CN115526112A - Building carbon emission dynamic optimization management and control system and method based on machine learning - Google Patents

Building carbon emission dynamic optimization management and control system and method based on machine learning Download PDF

Info

Publication number
CN115526112A
CN115526112A CN202211272055.9A CN202211272055A CN115526112A CN 115526112 A CN115526112 A CN 115526112A CN 202211272055 A CN202211272055 A CN 202211272055A CN 115526112 A CN115526112 A CN 115526112A
Authority
CN
China
Prior art keywords
carbon emission
building
energy
carbon
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211272055.9A
Other languages
Chinese (zh)
Inventor
丁勇
罗梓淇
罗庆
张东林
蒋祥婷
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University
Original Assignee
Chongqing University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University filed Critical Chongqing University
Priority to CN202211272055.9A priority Critical patent/CN115526112A/en
Publication of CN115526112A publication Critical patent/CN115526112A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/06—Power analysis or power optimisation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention discloses a building carbon emission dynamic optimization management and control system and method based on machine learning, which comprises a building carbon emission dynamic regulation and control system and a building carbon emission dynamic management system; the building carbon emission dynamic regulation and control system comprises a building monitoring module, an energy resource prediction module, a carbon emission quota calculation module, a current carbon emission calculation module and a carbon emission regulation and control module; the building carbon emission dynamic management system comprises a building operation period carbon emission calculation module, a building inherent carbon calculation module and a building low-carbon performance evaluation module; the invention can obtain a real-time dynamic building energy resource model according to the building operation information and energy resource consumption data on the same day or a period of time, further obtain a dynamic carbon emission quota and influence factors, and pertinently regulate and control main objects influencing carbon emission until the carbon emission of building operation reaches the optimum under the allowable condition, thereby reducing the carbon emission of the building operation period.

Description

Building carbon emission dynamic optimization management and control system and method based on machine learning
Technical Field
The invention relates to the field of quota and control of building carbon emission, in particular to a dynamic optimization control system and method for building carbon emission based on machine learning.
Background
The intelligent building energy consumption monitoring system aims at solving the problems of high building energy consumption, high carbon emission and the like, and by means of an informatization technology and utilizing the advantages of large data collection and analysis prediction, building energy consumption efficiency is improved, building energy consumption analysis is refined, building operation and maintenance are optimized, and building carbon emission is reduced, so that the intelligent building energy consumption monitoring system is a key development direction which is beneficial to improving the development level of the building industry in science and technology and promoting energy conservation, emission reduction and efficiency improvement of the building industry.
But at present, the following defects exist in the aspect of controlling the carbon emission of the building: firstly, due to the regulation limitation of energy consumption limitation, annual benchmarking is often performed according to annual energy consumption data, the time period is too long, the current situation of building energy consumption cannot be known in time, and the carbon emission control cannot be performed on related energy consumption behaviors in time; carbon emission comprises carbon emission of energy and resources, but few methods bring the resources into a control range; thirdly, the uncertainty of carbon emission influence factors, different building characteristics and different weight of the factors influencing carbon emission cause that regulation and control objects are indiscriminately determined, the carbon emission control efficiency is reduced, and how to find the carbon emission influence factors and carry out timely and targeted regulation and control is also an important problem.
In recent years, machine learning technology has been developed greatly in the field of building energy consumption prediction, and by utilizing the advantages of a machine learning algorithm, rapid modeling analysis can be performed according to different and continuously updated data sources, so that independent variable and dependent variable related models can be obtained, influence factors can be known, and excellent technical support can be provided for building carbon emission control.
Therefore, in order to solve the problems of inflexible limit of carbon emission, incomplete range, unrefined influence factors, inaccurate regulation and control object, untimely feedback of regulation and control effect and the like, the building carbon emission control system and the method for building carbon emission control based on machine learning dynamic quota of building carbon emission, post-calibration feedback regulation and control, carbon emission reduction and optimal carbon emission effect search have better practical significance.
Disclosure of Invention
The invention aims to provide a building carbon emission dynamic optimization management and control system based on machine learning, which comprises a building carbon emission dynamic regulation and control system;
the building carbon emission dynamic regulation and control system comprises a building monitoring module, an energy resource prediction module, a carbon emission quota calculation module, a current carbon emission calculation module and a carbon emission regulation and control module;
the building monitoring module is used for acquiring building operation information and energy resource consumption data and inputting the building operation information and the energy resource consumption data into the energy resource prediction module and the current carbon emission calculation module;
the energy resource prediction module stores an energy prediction model and a resource prediction model which are constructed based on a machine learning algorithm;
according to the building operation information and the energy resource consumption data, the energy prediction model outputs energy prediction data to the carbon emission quota calculation module, and the resource prediction model outputs resource prediction data to the carbon emission quota calculation module;
forming an energy prediction model according to the building operation information and the energy consumption data, obtaining the influence factors of the current energy consumption by the model, and transmitting the influence factors to a carbon emission regulation and control module;
forming the resource prediction model according to the building operation information and the resource consumption data, obtaining the influence factors of the current resource consumption by the model, and transmitting the influence factors to the carbon emission regulation and control module;
the carbon emission quota calculation module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulation and control module;
the current carbon emission calculation module converts the energy resource consumption data into hour operation carbon emission data according to the carbon emission factor and transmits the hour operation carbon emission data to the carbon emission regulation and control module;
the carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data is larger than the hourly carbon emission quota, the carbon emission regulation and control module performs system regulation and control of energy and resources according to the influence factors of current energy consumption and the influence factors of current resource consumption and controls the building monitoring module to continue working.
Further, the building operation information comprises electromechanical equipment operation performance and state, an environment state, renewable energy system performance and non-traditional water source system performance; the energy resource consumption data comprises total energy and total resources actually consumed by all systems of the building.
Further, the operation performance and the state of the electromechanical equipment in the building operation information comprise the number, the duration and the power of lamps started in a lighting socket system, and the number, the duration and the power of electric appliances using sockets; the number, duration and performance of elevator runs in the power system; the method comprises the following steps that cold and heat sources in an air-conditioning system run long, frequency, power, inlet and outlet water temperature and flow, air-conditioning water pumps run quantity, time, frequency and power, air-conditioning fans run quantity, time, frequency and power, cooling tower fans run quantity, time, frequency and power, and air-conditioning tail end equipment run quantity, time and performance; the running performance of a hot water boiler and a water pump in a domestic hot water system; the number, duration, frequency, power and performance of the devices operating in the energy consumption system; the using times and duration of the water heaters in the water supply and drainage system are increased; the times and duration of greening spraying operation are increased; the environmental conditions include set air conditioner temperature and air speed, actual ambient temperature, relative humidity, and CO in room and public area 2 And (4) concentration.
Further, when a building is built with a renewable energy system and a non-traditional water source system, the building operation information further comprises the performance of the renewable energy system and the performance of the non-traditional water source system;
the performance of the renewable energy system comprises the installed capacity, the starting rate and the renewable energy generation amount of renewable energy equipment;
the performance of the non-traditional water source system comprises the capacity, collection amount and usage amount of a rainwater and reclaimed water recovery device.
Further, the energy resource consumption data comprises total energy and total resources actually consumed by all systems of the building; the energy sources comprise electric power, natural gas, coal, gasoline, diesel oil and liquefied petroleum gas; the resources include tap water, self-contained supply water, and bottled water.
Further, the step of establishing an energy prediction model and a resource prediction model comprises:
1) Performing variable preprocessing according to the independent variable parameter type; when the independent variable is a continuous variable, the preprocessing method comprises Z-Score standardization; when the independent variable is a subtype variable, the preprocessing method comprises One-hot coding; carrying out normal distribution conversion on the dependent variable; the independent variable is building operation information, and the dependent variable is energy resource consumption data;
2) Dividing the building operation information and the energy resource consumption data set S into a training set D and a testing set T, wherein S = D memory,
Figure BDA0003895401120000031
3) Selecting a machine learning algorithm, taking the value of the hyper-parameter as an abscissa, and obtaining a decision coefficient R by a K-fold cross validation method 2 Drawing a learning curve as ordinate to determine the coefficient R 2 Carrying out super-parameter tuning by taking the growth rate not exceeding a preset threshold value as a standard; when the decision coefficient R is adjusted 2 After the growth rate does not exceed a preset threshold value, taking the parameter as an optimal hyper-parameter so as to finish the tuning of all hyper-parameters;
wherein the coefficient R is determined 2 As follows:
Figure BDA0003895401120000032
in the formula, y is a true value,
Figure BDA0003895401120000033
is the predicted value of the model and is,
Figure BDA0003895401120000034
is the arithmetic mean of the true value y, n is the number of samples;
4) And (5) carrying out training test again by using the optimal hyper-parameters to obtain an energy prediction model and a resource prediction model.
Further, the machine learning algorithm is an algorithm capable of identifying the importance of the factor features, and comprises a learning algorithm based on a linear kernel: linear regression, ridge regression, support vector regression-linear kernel, learning algorithm based on tree model: decision tree regression, random forest, gradient boosting tree.
Further, the influence factors of the current energy consumption obtained by the energy prediction model are one or more of the following energy consumption influence factors: the number and duration of the lamps; the number and duration of the socket electric appliances; the number and duration of elevator usage; the using quantity, duration and performance of each device of the air conditioning system; the using quantity, duration and performance of each device of the domestic hot water system; setting temperature and wind speed of air conditioners in rooms and public areas; installed capacity, turn-on rate and renewable energy generation amount of renewable energy equipment.
The influence factors of the current resource consumption obtained by the resource prediction model are one or more of the following energy consumption influence factors: the using times and duration of water appliances; the times and duration of greening spraying operation are increased; the capacity, the collection amount and the use amount of the rainwater and reclaimed water recovery device.
Further, the system also comprises a building carbon emission dynamic management system;
when the hourly operation carbon emission data is less than or equal to the hourly carbon emission quota, starting the building carbon emission dynamic management system;
the building carbon emission dynamic management system comprises a building operation period carbon emission calculation module, a building inherent carbon calculation module and a building low-carbon performance evaluation module;
the building operation period carbon emission calculation module accumulates hour operation carbon emission data according to the building operation time, calculates to obtain the building operation period carbon emission, and transmits the building operation period carbon emission to the low-carbon performance evaluation module of the building;
the building inherent carbon calculation module calculates building inherent carbon according to building information and a carbon emission factor, estimates the carbon emission in the building demolition period according to the fitting relationship among the building area, the number of building layers and the carbon emission in the building demolition period, and transmits the carbon emission to the building low-carbon performance evaluation module;
the building low-carbon performance evaluation module adds the carbon emission in the building operation period, the inherent carbon of the building and the carbon emission in the building demolition period to obtain the carbon emission in the building life cycle; the building life cycle carbon emission is inversely related to the building low carbon performance score. The building inherent carbon comprises building material preparation carbon emission and construction carbon emission.
A use method of a building carbon emission dynamic optimization management and control system based on machine learning comprises the following steps:
1) Building operation information and energy resource consumption data are collected by a building monitoring module and input into an energy resource prediction module and a current carbon emission calculation module;
2) Establishing an energy prediction model and a resource prediction model based on machine learning by using a historical data set, and simultaneously obtaining current energy influence factors and current resource influence factors in T time by using the models;
3) Building operation information which does not participate in model building is respectively input into the energy prediction model and the resource prediction model, and energy prediction data and resource prediction data are obtained through calculation;
4) The energy prediction model transmits energy prediction data to the carbon emission quota calculation module and transmits influence factors of current energy consumption to the carbon emission regulation module;
the resource prediction model transmits resource prediction data to the carbon emission quota calculation module and transmits influence factors of current resource consumption to the carbon emission regulation and control module;
5) Updating a historical data set after regulation and control, and constructing an energy prediction model and a resource prediction model based on machine learning by using the new historical data set;
6) The carbon emission quota calculating module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulating module; the current carbon emission calculation module converts current energy resource consumption data into hourly operation carbon emission data according to the carbon emission factor and transmits the hourly operation carbon emission data to the carbon emission regulation and control module;
7) And the carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data is greater than the hourly carbon emission quota, the carbon emission regulation and control module performs energy and resource system regulation and control according to the influence factors of the current energy consumption and the influence factors of the current resource consumption, and returns to the step 1).
The technical effect of the invention is undoubted, the invention can obtain a real-time dynamic building energy resource model according to the building operation information and energy resource consumption data of the same day or a period of time, further obtain a dynamic carbon emission quota and influence factors, and pertinently regulate and control main objects influencing carbon emission until the carbon emission of the building operation reaches the optimum under the allowable condition, thereby reducing the carbon emission of the building operation period.
Drawings
FIG. 1 is a flow chart of the use of the system;
FIG. 2 is a structural diagram of a building carbon emission dynamic regulation system;
fig. 3 is a structural diagram of a building carbon emission dynamic management system.
Detailed Description
The present invention will be further described with reference to the following examples, but it should be understood that the scope of the subject matter described above is not limited to the following examples. Various substitutions and alterations can be made without departing from the technical idea of the invention and the scope of the invention is covered by the present invention according to the common technical knowledge and the conventional means in the field.
Example 1:
referring to fig. 1, 2 and 3, a building carbon emission dynamic optimization management and control system based on machine learning includes a building carbon emission dynamic regulation and control system;
the building carbon emission dynamic regulation and control system comprises a building monitoring module, an energy resource prediction module, a carbon emission quota calculation module, a current carbon emission calculation module and a carbon emission regulation and control module;
the building monitoring module is used for acquiring building operation information and energy resource consumption data and inputting the building operation information and the energy resource consumption data into the energy resource prediction module and the current carbon emission calculation module;
the energy resource prediction module stores an energy prediction model and a resource prediction model which are constructed based on a machine learning algorithm;
according to the building operation information and the energy resource consumption data, the energy prediction model outputs energy prediction data to the carbon emission quota calculation module, and the resource prediction model outputs resource prediction data to the carbon emission quota calculation module;
forming an energy prediction model according to the building operation information and the energy consumption data, obtaining the influence factors of the current energy consumption by the model, and transmitting the influence factors to a carbon emission regulation and control module;
forming the resource prediction model according to the building operation information and the resource consumption data, obtaining the influence factors of the current resource consumption by the model, and transmitting the influence factors to the carbon emission regulation and control module;
the carbon emission quota calculating module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulating module;
the current carbon emission calculation module converts the energy resource consumption data into hourly operation carbon emission data according to the carbon emission factor and transmits the hourly operation carbon emission data to the carbon emission regulation and control module;
the conversion mode is as follows: and acquiring a corresponding carbon emission factor according to national standards and specifications, and multiplying the data by the carbon emission factor to obtain carbon emission prediction data. For example, the total power consumption of a building is known, and the carbon emission factor corresponding to one electric power is directly multiplied to obtain the carbon emission amount of the electric power consumption; knowing the total consumption of natural gas (tap water, gasoline and diesel oil) of the building, directly multiplying the total consumption by a carbon emission factor corresponding to the natural gas (tap water, gasoline and diesel oil) to obtain the carbon emission of different types of energy resources, wherein the sum of the carbon emission and the carbon emission is the total carbon emission of the building.
The carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data is larger than the hourly carbon emission quota, the carbon emission regulation and control module performs system regulation and control of energy and resources according to the influence factors of current energy consumption and the influence factors of current resource consumption and controls the building monitoring module to continue working.
The building operation information comprises the operation performance and state of electromechanical equipment, the environment state, the performance of a renewable energy source system and the performance of a non-traditional water source system; the energy resource consumption data comprises total energy and total resources actually consumed by all systems of the building.
The operation performance and the state of the electromechanical equipment in the building operation information comprise the number, the duration and the power of lamps started in a lighting socket system, and the number, the duration and the power of electric appliances using sockets; the number, duration and performance of elevator operation in the power system; the method comprises the following steps that an air-conditioning system is provided with an inter-cooling heat source, wherein the inter-cooling heat source comprises the operation time length, frequency, power, inlet and outlet water temperature and flow, the operation number, time length, frequency and power of an air-conditioning water pump, the operation number, time length, frequency and power of air-conditioning fans, the operation number, time length, frequency and power of cooling tower fans, and the operation number, time length and performance of air-conditioning terminal equipment; the running performance of a hot water boiler and a water pump in a domestic hot water system; the number, duration, frequency, power and performance of devices operating in a special (other) energy-consuming system; the using times and duration of the water heaters in the water supply and drainage system are increased; the number and duration of greening spraying operation are increased; the environment state comprises set temperature and wind speed of air conditioner, actual environment temperature, relative humidity and CO of room and public area 2 And (4) concentration.
When a renewable energy system and a non-traditional water source system are built in the building, the building operation information also comprises the performance of the renewable energy system and the performance of the non-traditional water source system;
the performance of the renewable energy system comprises the installed capacity, the starting rate and the renewable energy generation amount of renewable energy equipment;
the performance of the non-traditional water source system comprises the capacity, collection amount and usage amount of a rainwater and reclaimed water recovery device.
The energy resource consumption data comprises total energy and total resources actually consumed by all systems of the building; the energy source comprises electric power, natural gas, coal, gasoline, diesel oil and liquefied petroleum gas; the resources include tap water, self-contained supply water, and barreled water.
The steps of establishing the energy prediction model and the resource prediction model comprise:
1) Performing variable preprocessing according to the independent variable parameter type; when the independent variable is a continuous variable, the pretreatment method comprises Z-Score standardization; when the independent variable is a subtype variable, the preprocessing method comprises One-hot coding; carrying out normal distribution conversion on the dependent variable; the independent variable is building operation information, and the dependent variable is energy resource consumption data;
2) Dividing the building operation information and the energy resource consumption data set S into a training set D and a testing set T, wherein S = D memory,
Figure BDA0003895401120000071
3) Selecting a machine learning algorithm, taking the value of the hyper-parameter as an abscissa, and obtaining a decision coefficient R by a K-fold cross validation method 2 Drawing a learning curve as ordinate to determine the coefficient R 2 Carrying out super-parameter tuning by taking the growth rate not exceeding a preset threshold value as a standard; when adjusted, determining the coefficient R 2 After the growth rate does not exceed a preset threshold value, taking the parameter as an optimal hyper-parameter so as to finish the tuning of all hyper-parameters;
wherein the coefficient R is determined 2 As follows:
Figure BDA0003895401120000072
in the formula, y is a true value,
Figure BDA0003895401120000073
is the predicted value of the model and is,
Figure BDA0003895401120000074
is the arithmetic mean of the true value y, n is the number of samples;
4) And carrying out training test again by using the optimal hyper-parameter to obtain an energy prediction model and a resource prediction model.
The machine learning algorithm is an algorithm capable of identifying the importance of the feature of the factor, and comprises a learning algorithm based on a linear kernel: linear regression, ridge regression, support vector regression-linear kernel, learning algorithm based on tree model: decision tree regression, random forest, gradient boosting tree.
The influence factors of the current energy consumption obtained by the energy prediction model are one or more of the following energy consumption influence factors: the number and duration of the lamps; the number and duration of the socket electric appliances; the number and duration of elevator usage; the using quantity, duration and performance of each device of the air conditioning system; the using quantity, duration and performance of each device of the domestic hot water system are determined; setting temperature and wind speed of air conditioners in rooms and public areas; installed capacity, turn-on rate and renewable energy generation amount of renewable energy equipment.
The influence factors of the current resource consumption obtained by the resource prediction model are one or more of the following energy consumption influence factors: the using times and duration of water appliances; the number and duration of greening spraying operation are increased; capacity, collection amount and usage amount of the rainwater and reclaimed water recovery device.
The system also comprises a building carbon emission dynamic management system;
when the hourly operation carbon emission data is less than or equal to the hourly carbon emission quota, starting the building carbon emission dynamic management system;
the building carbon emission dynamic management system comprises a building operation period carbon emission calculation module, a building inherent carbon calculation module and a building low carbon performance evaluation module;
the building operation period carbon emission calculation module accumulates hour operation carbon emission data according to the building operation time, calculates to obtain the building operation period carbon emission, and transmits the building operation period carbon emission to the low-carbon performance evaluation module of the building;
the building inherent carbon calculation module calculates building inherent carbon according to the building information and the carbon emission factor, estimates the carbon emission in the building demolition period according to the fitting relationship among the building area, the number of building layers and the carbon emission in the building demolition period, and transmits the carbon emission to the building low-carbon performance evaluation module;
the building low-carbon performance evaluation module adds the carbon emission of the building in the operation period, the inherent carbon of the building and the carbon emission of the building in the demolition period to obtain the carbon emission of the life cycle of the building; the building life cycle carbon emission is inversely related to the building low carbon performance score. The building inherent carbon comprises building material preparation carbon emission and construction carbon emission.
A use method of a building carbon emission dynamic optimization management and control system based on machine learning comprises the following steps:
1) Building operation information and energy resource consumption data are collected by a building monitoring module and input into an energy resource prediction module and a current carbon emission calculation module;
2) An energy prediction model and a resource prediction model based on machine learning are built by utilizing a historical data set, and the model simultaneously obtains current energy influence factors and current resource influence factors in T time;
3) Building operation information which does not participate in model building is respectively input into the energy prediction model and the resource prediction model, and energy prediction data and resource prediction data are obtained through calculation;
4) The energy prediction model transmits energy prediction data to the carbon emission quota calculation module and transmits influence factors of current energy consumption to the carbon emission regulation module;
the resource prediction model transmits resource prediction data to the carbon emission quota calculation module and transmits influence factors of current resource consumption to the carbon emission regulation module;
5) Updating a historical data set after regulation and control, and constructing an energy prediction model and a resource prediction model based on machine learning by using the new historical data set;
6) The carbon emission quota calculation module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulation and control module; the current carbon emission calculation module converts current energy resource consumption data into hourly operation carbon emission data according to the carbon emission factor and transmits the hourly operation carbon emission data to the carbon emission regulation and control module;
7) And the carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data is greater than the hourly carbon emission quota, the carbon emission regulation and control module performs energy and resource system regulation and control according to the influence factors of the current energy consumption and the influence factors of the current resource consumption, and returns to the step 1).
Example 2:
a method for using a building carbon emission dynamic optimization management and control system based on machine learning comprises the following steps:
firstly, a regulation part comprises the following steps:
s1: collecting building operation information and energy resource consumption data through a building monitoring system to construct a data set;
s2: based on the data sets, a machine learning algorithm capable of identifying feature importance is adopted to construct a building energy prediction model and a resource prediction model, and the models respectively obtain energy influence factors and resource influence factors;
s3: converting the energy resource data obtained by the model into carbon emission data through a carbon emission factor, and obtaining a building carbon emission quota by a quota level method or a sequencing method;
s4: calculating the energy resource consumption data through the carbon emission factor to obtain the building operation carbon emission, and carrying out benchmarking with the carbon emission quota obtained in the step S3;
s5: when the carbon emission of the building operation exceeds the carbon emission quota, determining a regulation object according to the influence factors obtained by the model, regulating and controlling the system from two aspects of energy resources, and reducing the carbon emission of the building operation;
s6: after the regulation and control are completed, the monitoring system collects new data again, the steps S1 to S4 are repeated, if the data do not reach the standard, the step S5 and the step are continued until the carbon emission of the building operation is lower than or equal to the carbon emission quota, the regulation and control at the stage are finished, and a management part is entered;
in the management part, further calculating the operating carbon emission meeting the quota requirement to obtain the carbon emission of the building in the operating period, wherein the data can be used for carbon transaction in the building field; calculating the inherent carbon of the building according to the building information and the carbon emission factor, wherein the carbon emission of building material preparation and the carbon emission of construction are included; and estimating the carbon emission in the building demolition period according to the fitting relationship among the building area, the number of building layers and the carbon emission in the building demolition period. And adding the carbon emission in the building operation period, the inherent carbon of the building and the carbon emission in the building demolition period to obtain the carbon emission of the life cycle of the building, wherein the data can be used for evaluating the low-carbon performance of the building.
The monitored building operation information comprises the operation performance and state of electromechanical equipment, the environment state, the performance of a renewable energy system and the performance of a non-traditional water source system; the energy resource consumption data comprises the total energy and total resources actually consumed by all the systems of the building.
When a machine learning algorithm is used for building a model, building operation information is used as an independent variable, and energy resource consumption data is used as a dependent variable;
in a further refinement, the step S2 includes the steps of:
s21: performing variable pretreatment according to the independent variable parameter types, standardizing continuous variables by adopting a Z-Score method, coding classified variables by adopting a One-hot method, and performing normal distribution transformation on dependent variables;
s22: the data set S is divided into a training set D and a test set T, satisfying S = dout,
Figure BDA0003895401120000101
Figure BDA0003895401120000102
s23: selecting a machine learning algorithm, taking the value of the hyper-parameter as an abscissa, and obtaining a decision coefficient R by a K-fold cross validation method 2 As ordinate, a learning curve is drawn with R 2 Increasing the rate to be not more than 0.1 percent as a threshold standard to carry out super-parameter tuning;
s24: and carrying out training test again by using the optimal hyper-parameter to obtain an energy prediction model and a resource prediction model.
The machine learning algorithm needs to select an algorithm that can identify the importance of the feature of the factor.
The carbon emission quota level is selected by taking comprehensive consideration of the self condition and policy requirements of the building.
Example 3:
a use method of a building carbon emission dynamic optimization management and control system based on machine learning comprises the following steps:
s1: building operation information and energy resource consumption data are collected through a building monitoring system, and a data set is constructed;
s2: based on the data sets, a machine learning algorithm capable of identifying feature importance is adopted to construct a building energy prediction model and a resource prediction model, and the models respectively obtain energy influence factors and resource influence factors;
s3: converting the energy resource data obtained by the model into carbon emission data through a carbon emission factor, and obtaining a building carbon emission quota by a quota level method or a sequencing method;
s4: calculating the energy resource consumption data through the carbon emission factor to obtain the building operation carbon emission, and carrying out benchmarking with the carbon emission quota obtained in the step S3;
s5: when the carbon emission of the building operation exceeds the carbon emission quota, determining a regulation object according to the influence factors obtained by the model, regulating and controlling the system from two aspects of energy resources, and reducing the carbon emission of the building operation;
s6: after the regulation and control are completed, the monitoring system collects new data again, the steps S1 to S4 are repeated, if the data do not reach the standard, the step S5 and the step are continued until the carbon emission of the building operation is lower than or equal to the carbon emission quota, the regulation and control at the stage are finished, and a management part is entered;
in the management part, further calculating the operating carbon emission meeting the quota requirement to obtain the carbon emission of the building in the operating period, wherein the data can be used for carbon transaction in the building field; calculating the inherent carbon of the building according to the building information and the carbon emission factor, wherein the carbon emission of building material preparation and the carbon emission of construction are included; and estimating the carbon emission in the building demolition period according to the fitting relationship among the building area, the number of building layers and the carbon emission in the building demolition period. And adding the carbon emission in the building operation period, the inherent carbon of the building and the carbon emission in the building demolition period to obtain the carbon emission in the building life cycle, wherein the data can be used for evaluating the low-carbon performance of the building.
Specifically, the building operation information collected by the building monitoring system includes the operation performance and state of the electromechanical devices of the building, such as the number, duration and power of lamps turned on in the lighting socket system, and the number, duration and power of electrical appliances using the sockets; the number, duration and performance of elevator runs in the power system; the method comprises the following steps that cold and heat sources in an air-conditioning system run long, frequency, power, inlet and outlet water temperature and flow, air-conditioning water pumps run quantity, time, frequency and power, air-conditioning fans run quantity, time, frequency and power, cooling tower fans run quantity, time, frequency and power, and air-conditioning tail end equipment run quantity, time and performance; the running performance of a hot water boiler and a water pump in a domestic hot water system; the number, duration, frequency, power and performance of devices operating in a particular (other) energy consuming system; the using times and duration of the water heaters in the water supply and drainage system are increased; the times and duration of greening spraying operation are increased; also included are environmental conditions such as air conditioning set temperature and wind speed, actual ambient temperature, relative humidity, CO in rooms and public areas 2 Concentration; if the building is built with a renewable energy system and a non-traditional water source system, the operation information also comprises the performance of the renewable energy system, such as the installed capacity, the starting rate and the renewable energy generation amount of the renewable energy equipment; also include non-traditional water source system performance, such as rainwater and mid-water recovery unit capacity, collection, usage;
another aspect of the energy resource consumption data collected by the building monitoring system includes the total energy and total resources actually consumed by all systems of the building. For example, energy sources including electricity, natural gas, coal, gasoline, diesel, liquefied petroleum gas, and the like; resources include tap water, self-contained supply water, bottled water, etc.
After the data set collection is completed, a machine learning algorithm capable of identifying the importance of the factor features is selected, for example, a learning algorithm based on linear kernels: linear regression, ridge regression, support vector regression-linear kernel; another example is a learning algorithm based on a tree model: regression of decision trees, random forests and gradient boosting trees; and establishing an energy resource model of the building within a period of time by taking all factors in the building operation information as independent variables and energy resource consumption data as dependent variables.
In the embodiment, one hour is taken as a data mapping, hour data in a month before a week is taken as model training and testing data, hour operation data in a recent week is input into the model to obtain an energy resource model value, and the model value is used for carrying out carbon emission quota.
The machine learning algorithm model is specifically described as follows:
s21: carrying out variable preprocessing according to independent variable parameter types, standardizing continuous variables such as ' water pump running time ', ' lamp starting number ', ' elevator running time ' and ' number of computers used by adopting a Z-Score method, coding classified variables such as ' whether a fan adopts frequency conversion ' by adopting a One-hot method, and carrying out normal distribution conversion, such as logarithmic conversion, when the dependent variables do not meet the requirement of energy resources in normal distribution;
s22: dividing a data set S containing operation information and energy resource consumption data of each hour in a month before a week into a training set D and a testing set T, wherein S = D ^ T,
Figure BDA0003895401120000121
s23: according to the selected machine learning algorithm, taking the value of the hyper-parameter as the abscissa and obtaining a decision coefficient R by a K-turn cross-validation method 2 For ordinate, a learning curve is plotted, with R 2 Increasing by no more than 0.1% as standard to perform hyper-parameter tuning, and obtaining model R when the adjusted hyper-parameter 2 After the growth rate does not exceed 0.1%, taking the parameter as the optimal hyper-parameter, and adjusting all the hyper-parameters without changing other hyper-parameters;
s24: and carrying out training test again by using the optimal hyper-parameter to obtain an energy prediction model and a resource prediction model.
After the model is obtained, the main factors influencing the energy consumption and the main factors influencing the resource consumption in the period of time can be obtained, and the factors can be used as alternative objects for regulation and control.
Inputting hour operation information in the data set within a week to a model to obtain an energy resource consumption model predicted value, converting the predicted value into carbon emission data through a carbon emission factor, and obtaining an hour carbon emission quota of the building by a quota level method or a sequencing method; selecting a quota level method when the carbon emission data meet a certain probability density distribution model, and otherwise selecting a sequencing method; the quota level in quota is dynamically determined after comprehensive consideration is carried out according to the self condition and policy requirements of the building, for example, the building is built with a renewable energy system, so that the quota level can be reduced; the quota level can be improved if the building has inevitable large-scale equipment;
calculating the latest acquired current hour energy resource consumption data according to the carbon emission factor to obtain the current hour operation carbon emission of the building, and comparing the current hour operation carbon emission with the hour carbon emission quota obtained in the last step;
and if the hourly operation carbon emission exceeds the hourly carbon emission quota, performing system regulation and control on energy resources according to influence elements obtained by the model, such as reducing the number of lamps used, reducing the service life of an elevator, reducing the opening time of an air conditioner, improving the set temperature of the air conditioner in a room, reducing the opening time of green spraying, increasing the use amount of rainwater and the like, and reducing the operation carbon emission of the building.
After a series of regulation and control are completed, the monitoring system can acquire new data again, and continue to perform the steps until the hourly carbon emission is lower than or equal to the hourly carbon emission quota when the benchmarks are judged, the monitoring system jumps out of the regulation and control part and enters the management part.
In the management part, the carbon emission in the operation period of the building can be obtained by multiplying the carbon emission in the operation period by the expected life of the building, and the data can be used for carbon transaction in the field of the building; calculating the inherent carbon of the building according to the building information and the carbon emission factor, wherein the carbon emission of building material preparation and the carbon emission of construction are included; and estimating the carbon emission in the building demolition period according to the fitting relationship among the building area, the number of building layers and the carbon emission in the building demolition period.
And adding the carbon emission in the building operation period, the inherent carbon of the building and the carbon emission in the building demolition period to obtain the carbon emission in the building life cycle, wherein the data can be used for evaluating the low-carbon performance of the building.
Example 4:
a building carbon emission dynamic optimization management and control system based on machine learning comprises a building carbon emission dynamic regulation and control system;
the building carbon emission dynamic regulation and control system comprises a building monitoring module, an energy resource prediction module, a carbon emission quota calculation module, a current carbon emission calculation module and a carbon emission regulation and control module;
the building monitoring module is used for acquiring building operation information and energy resource consumption data and inputting the building operation information and the energy resource consumption data into the energy resource prediction module and the current carbon emission calculation module;
the energy resource prediction module stores an energy prediction model and a resource prediction model which are constructed based on a machine learning algorithm;
according to the building operation information and the energy resource consumption data, the energy prediction model outputs energy prediction data to the carbon emission quota calculation module, and the resource prediction model outputs resource prediction data to the carbon emission quota calculation module;
forming an energy prediction model according to the building operation information and the energy consumption data, obtaining the influence factors of the current energy consumption by the model, and transmitting the influence factors to a carbon emission regulation and control module;
forming the resource prediction model according to the building operation information and the resource consumption data, obtaining the influence factors of the current resource consumption by the model, and transmitting the influence factors to the carbon emission regulation and control module;
the carbon emission quota calculation module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulation and control module;
the current carbon emission calculation module converts the energy resource consumption data into hour operation carbon emission data according to the carbon emission factor and transmits the hour operation carbon emission data to the carbon emission regulation and control module;
the carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data are larger than the hourly carbon emission quota, the carbon emission regulation and control module performs energy and resource system regulation and control according to the influence factors of the current energy consumption and the influence factors of the current resource consumption and controls the building monitoring module to continue working.
Example 5:
a building carbon emission dynamic optimization management and control system based on machine learning mainly comprises an embodiment 4, wherein the building operation information comprises electromechanical equipment operation performance and state, an environment state, renewable energy system performance and non-traditional water source system performance; the energy resource consumption data comprises total energy and total resources actually consumed by all systems of the building.
Example 6:
a building carbon emission dynamic optimization management and control system based on machine learning mainly comprises an embodiment 4, wherein the electromechanical equipment operation performance and state in building operation information comprise the number, duration and power of lamps turned on in a lighting socket system, and the number, duration and power of electric appliances using sockets; the number, duration and performance of elevator operation in the power system; the method comprises the following steps that an air-conditioning system is provided with an inter-cooling heat source, wherein the inter-cooling heat source comprises the operation time length, frequency, power, inlet and outlet water temperature and flow, the operation number, time length, frequency and power of an air-conditioning water pump, the operation number, time length, frequency and power of air-conditioning fans, the operation number, time length, frequency and power of cooling tower fans, and the operation number, time length and performance of air-conditioning terminal equipment; the running performance of a hot water boiler and a water pump in a domestic hot water system; the number, duration, frequency, power and performance of the devices operating in the energy consumption system; the using times and duration of the water heaters in the water supply and drainage system are increased; the number and duration of greening spraying operation are increased; the environmental conditions include set air conditioner temperature and air speed, actual ambient temperature, relative humidity, and CO in room and public area 2 And (4) concentration.
Example 7:
a building carbon emission dynamic optimization management and control system based on machine learning mainly comprises embodiment 4, wherein when a renewable energy system and a non-traditional water source system are built in a building, the building operation information further comprises the performance of the renewable energy system and the performance of the non-traditional water source system;
the performance of the renewable energy system comprises the installed capacity, the starting rate and the renewable energy generation amount of renewable energy equipment;
the performance of the non-traditional water source system comprises the capacity, collection amount and usage amount of a rainwater and reclaimed water recovery device.
Example 8:
a building carbon emission dynamic optimization management and control system based on machine learning is mainly disclosed in embodiment 4, wherein the energy resource consumption data comprises total energy and total resources actually consumed by all systems of a building; the energy sources comprise electric power, natural gas, coal, gasoline, diesel oil and liquefied petroleum gas; the resources include tap water, self-contained supply water, and bottled water.
Example 9:
a building carbon emission dynamic optimization management and control system based on machine learning is mainly disclosed in embodiment 4, wherein the step of establishing an energy prediction model and a resource prediction model comprises the following steps:
1) Performing variable preprocessing according to the independent variable parameter type; when the independent variable is a continuous variable, the pretreatment method comprises Z-Score standardization; when the independent variable is a subtype variable, the preprocessing method comprises One-hot coding; carrying out normal distribution conversion on the dependent variable; the independent variable is building operation information, and the dependent variable is energy resource consumption data;
2) Dividing a building operation information and energy resource consumption data set S into a training set D and a testing set T, wherein S = D ≦ T,
Figure BDA0003895401120000151
3) Selecting a machine learning algorithm, taking the value of the hyper-parameter as an abscissa, and obtaining a decision coefficient R by a K-fold cross validation method 2 Drawing a learning curve as the ordinate to determine the coefficient R 2 The growth rate does not exceed a predetermined thresholdCarrying out hyper-parameter tuning with the value as a standard; when adjusted, determining the coefficient R 2 After the growth rate does not exceed a preset threshold value, taking the parameter as an optimal hyper-parameter so as to finish the tuning of all hyper-parameters;
wherein the coefficient R is determined 2 As follows:
Figure BDA0003895401120000152
in the formula, y is a true value,
Figure BDA0003895401120000153
is the predicted value of the model and is,
Figure BDA0003895401120000154
is the calculated number average of the true value y, and n is the number of samples;
4) And carrying out training test again by using the optimal hyper-parameter to obtain an energy prediction model and a resource prediction model.
Example 10:
a building carbon emission dynamic optimization management and control system based on machine learning is mainly disclosed in embodiment 4, wherein the machine learning algorithm is an algorithm capable of identifying the importance of factor features, and comprises a learning algorithm based on linear kernel: linear regression, ridge regression, support vector regression-linear kernel, learning algorithm based on tree model: decision tree regression, random forest, gradient boosting tree.
Example 11:
a building carbon emission dynamic optimization management and control system based on machine learning is mainly disclosed in embodiment 4, wherein the influence factors of the current energy consumption obtained by the energy prediction model are one or more of the following energy consumption influence factors: the number and duration of the lamps; the number and duration of the socket electric appliances; the number and duration of elevator usage; the using quantity, duration and performance of each device of the air conditioning system; the using quantity, duration and performance of each device of the domestic hot water system; setting temperature and wind speed of air conditioners in rooms and public areas; installed capacity, starting rate and renewable energy generation amount of renewable energy equipment.
The influence factors of the current resource consumption obtained by the resource prediction model are one or more of the following energy consumption influence factors: the using times and duration of the water-using appliance; the number and duration of greening spraying operation are increased; the capacity, the collection amount and the use amount of the rainwater and reclaimed water recovery device.
Example 12:
a building carbon emission dynamic optimization management and control system based on machine learning is disclosed in an embodiment 4, wherein the system further comprises a building carbon emission dynamic management system;
when the hourly operation carbon emission data is less than or equal to the hourly carbon emission quota, starting a building carbon emission dynamic management system;
the building carbon emission dynamic management system comprises a building operation period carbon emission calculation module, a building inherent carbon calculation module and a building low carbon performance evaluation module;
the building operation period carbon emission calculation module accumulates hourly operation carbon emission data according to building operation time, calculates to obtain building operation period carbon emission, and transmits the building operation period carbon emission to the low-carbon performance evaluation module of the building;
the building inherent carbon calculation module calculates building inherent carbon according to the building information and the carbon emission factor, estimates the carbon emission in the building demolition period according to the fitting relationship among the building area, the number of building layers and the carbon emission in the building demolition period, and transmits the carbon emission to the building low-carbon performance evaluation module;
the building low-carbon performance evaluation module adds the carbon emission of the building in the operation period, the inherent carbon of the building and the carbon emission of the building in the demolition period to obtain the carbon emission of the life cycle of the building; the building life cycle carbon emission is inversely related to the building low carbon performance score. The building inherent carbon comprises building material preparation carbon emission and construction carbon emission.
Example 13:
a use method of a building carbon emission dynamic optimization management and control system based on machine learning comprises the following steps:
1) Building operation information and energy resource consumption data are collected by a building monitoring module and input into an energy resource prediction module and a current carbon emission calculation module;
2) An energy prediction model and a resource prediction model based on machine learning are built by utilizing a historical data set, and the model simultaneously obtains current energy influence factors and current resource influence factors in T time;
3) Building operation information which does not participate in model building is respectively input into the energy prediction model and the resource prediction model, and energy prediction data and resource prediction data are obtained through calculation;
4) The energy prediction model transmits energy prediction data to the carbon emission quota calculation module and transmits influence factors of current energy consumption to the carbon emission regulation module;
the resource prediction model transmits resource prediction data to the carbon emission quota calculation module and transmits influence factors of current resource consumption to the carbon emission regulation module;
5) Updating a historical data set after regulation and control, and constructing an energy prediction model and a resource prediction model based on machine learning by using the new historical data set;
6) The carbon emission quota calculation module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulation and control module; the current carbon emission calculation module converts current energy resource consumption data into hourly operation carbon emission data according to the carbon emission factor and transmits the hourly operation carbon emission data to the carbon emission regulation and control module;
7) And the carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data are greater than the hourly carbon emission quota, the carbon emission regulation and control module performs energy and resource system regulation and control according to the influence factors of the current energy consumption and the influence factors of the current resource consumption, and returns to the step 1).

Claims (10)

1. The utility model provides a building carbon discharges developments management and control system that optimizes based on machine learning which characterized in that: the building carbon emission dynamic regulation system is included.
The building carbon emission dynamic regulation and control system comprises a building monitoring module, an energy resource prediction module, a carbon emission quota calculation module, a current carbon emission calculation module and a carbon emission regulation and control module;
the building monitoring module is used for acquiring building operation information and energy resource consumption data and inputting the building operation information and the energy resource consumption data into the energy resource prediction module and the current carbon emission calculation module;
the energy resource prediction module stores an energy prediction model and a resource prediction model which are constructed based on a machine learning algorithm;
according to the building operation information and the energy resource consumption data, the energy prediction model outputs energy prediction data to the carbon emission quota calculation module, and the resource prediction model outputs resource prediction data to the carbon emission quota calculation module;
forming an energy prediction model according to the building operation information and the energy consumption data, obtaining the influence factors of the current energy consumption by the model, and transmitting the influence factors to a carbon emission regulation and control module;
forming the resource prediction model according to the building operation information and the resource consumption data, obtaining the influence factors of the current resource consumption by the model, and transmitting the influence factors to the carbon emission regulation and control module;
the carbon emission quota calculating module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulating module;
the current carbon emission calculation module converts the energy resource consumption data into hourly operation carbon emission data according to the carbon emission factor and transmits the hourly operation carbon emission data to the carbon emission regulation and control module;
the carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data is larger than the hourly carbon emission quota, the carbon emission regulation and control module performs system regulation and control of energy and resources according to the influence factors of current energy consumption and the influence factors of current resource consumption and controls the building monitoring module to continue working.
2. The machine learning-based building carbon emission dynamic optimization management and control system according to claim 1, wherein: the building operation information comprises the operation performance and state of electromechanical equipment, the environment state, the performance of a renewable energy source system and the performance of a non-traditional water source system; the energy resource consumption data comprises total energy and total resources actually consumed by all systems of the building.
3. The machine learning-based building carbon emission dynamic optimization management and control system according to claim 2, wherein: the operation performance and the state of the electromechanical equipment in the building operation information comprise the number, the duration and the power of lamps started in a lighting socket system, and the number, the duration and the power of electric appliances using sockets; the number, duration and performance of elevator runs in the power system; the method comprises the following steps that an air-conditioning system is provided with an inter-cooling heat source, wherein the inter-cooling heat source comprises the operation time length, frequency, power, inlet and outlet water temperature and flow, the operation number, time length, frequency and power of an air-conditioning water pump, the operation number, time length, frequency and power of air-conditioning fans, the operation number, time length, frequency and power of cooling tower fans, and the operation number, time length and performance of air-conditioning terminal equipment; the running performance of a hot water boiler and the running performance of a water pump in a domestic hot water system; the number, duration, frequency, power and performance of the devices operating in the energy consumption system; the using times and duration of the water heaters in the water supply and drainage system are increased; the number and duration of greening spraying operation are increased; the environment state comprises set temperature and wind speed of air conditioner in room and public area, actual environment temperature, relative humidity, and CO 2 And (4) concentration.
4. The building carbon emission dynamic optimization management and control system based on machine learning according to claim 3, characterized in that: when a building is built with a renewable energy system and a non-traditional water source system, the building operation information also comprises the performance of the renewable energy system and the performance of the non-traditional water source system;
the performance of the renewable energy system comprises the installed capacity, the starting rate and the renewable energy generation amount of renewable energy equipment;
the performance of the non-traditional water source system comprises the capacity, collection amount and usage amount of a rainwater and reclaimed water recovery device.
5. The building carbon emission dynamic optimization management and control system based on machine learning according to claim 1, characterized in that: the energy resource consumption data comprises total energy and total resources actually consumed by all systems of the building; the energy sources comprise electric power, natural gas, coal, gasoline, diesel oil and liquefied petroleum gas; the resources include tap water, self-contained supply water, and bottled water.
6. The machine learning-based building carbon emission dynamic optimization management and control system according to claim 1, wherein the step of establishing an energy prediction model and a resource prediction model comprises:
1) Performing variable preprocessing according to the independent variable parameter type; when the independent variable is a continuous variable, the pretreatment method comprises Z-Score standardization; when the independent variable is a subtype variable, the preprocessing method comprises One-hot coding; carrying out normal distribution conversion on the dependent variable; the independent variable is building operation information, and the dependent variable is energy resource consumption data;
2) Dividing a building operation information and energy resource consumption data set S into a training set D and a testing set T, wherein S = D ≦ T,
Figure FDA0003895401110000021
3) Selecting a machine learning algorithm, taking the value of the hyper-parameter as the abscissa, and obtaining a decision coefficient R by a K-fold cross-validation method 2 Drawing a learning curve as the ordinate to determine the coefficient R 2 Carrying out super-parameter tuning on the standard that the growth rate does not exceed a preset threshold; when adjusted, determining the coefficient R 2 After the growth rate does not exceed a preset threshold value, taking the parameter as an optimal hyper-parameter so as to finish the tuning of all hyper-parameters;
wherein the coefficient R is determined 2 As follows:
Figure FDA0003895401110000031
in the formula, y is a true value,
Figure FDA0003895401110000032
is the predicted value of the model and is,
Figure FDA0003895401110000033
is the arithmetic mean of the true value y, n is the number of samples;
4) And (5) carrying out training test again by using the optimal hyper-parameters to obtain an energy prediction model and a resource prediction model.
7. The machine learning-based building carbon emission dynamic optimization management and control system according to claim 1, wherein: the machine learning algorithm is an algorithm capable of identifying the importance of the feature of the factor, and comprises a learning algorithm based on a linear kernel: linear regression, ridge regression, support vector regression-linear kernel, learning algorithm based on tree model: decision tree regression, random forest, gradient boosting tree.
8. The building carbon emission dynamic optimization management and control system based on machine learning according to claim 1, characterized in that: the influence factors of the current energy consumption obtained by the energy prediction model are one or more of the following influence factors of the energy consumption: the number and duration of the lamps; the number and duration of the socket electric appliances; the number and duration of elevator usage; the using quantity, duration and performance of each device of the air conditioning system; the using quantity, duration and performance of each device of the domestic hot water system; setting temperature and wind speed of air conditioners in rooms and public areas; installed capacity, starting rate and renewable energy generation amount of renewable energy equipment.
The influence factors of the current resource consumption obtained by the resource prediction model are one or more of the following energy consumption influence factors: the using times and duration of the water-using appliance; the times and duration of greening spraying operation are increased; capacity, collection amount and usage amount of the rainwater and reclaimed water recovery device.
9. The machine learning-based building carbon emission dynamic optimization management and control system according to claim 1, wherein: the system also comprises a building carbon emission dynamic management system;
when the hourly operation carbon emission data is less than or equal to the hourly carbon emission quota, starting the building carbon emission dynamic management system;
the building carbon emission dynamic management system comprises a building operation period carbon emission calculation module, a building inherent carbon calculation module and a building low carbon performance evaluation module;
the building operation period carbon emission calculation module accumulates hour operation carbon emission data according to the building operation time, calculates to obtain the building operation period carbon emission, and transmits the building operation period carbon emission to the low-carbon performance evaluation module of the building;
the building inherent carbon calculation module calculates building inherent carbon according to the building information and the carbon emission factor, estimates the carbon emission in the building demolition period according to the fitting relationship among the building area, the number of building layers and the carbon emission in the building demolition period, and transmits the carbon emission to the building low-carbon performance evaluation module;
the building low-carbon performance evaluation module adds the carbon emission of the building in the operation period, the inherent carbon of the building and the carbon emission of the building in the demolition period to obtain the carbon emission of the life cycle of the building; the building life cycle carbon emission is inversely related to the building low carbon performance score. The building inherent carbon comprises building material preparation carbon emission and construction carbon emission.
10. A use method of a building carbon emission dynamic optimization management and control system based on machine learning is characterized by comprising the following steps:
1) Building operation information and energy resource consumption data are collected by a building monitoring module and input into an energy resource prediction module and a current carbon emission calculation module;
2) An energy prediction model and a resource prediction model based on machine learning are built by utilizing a historical data set, and the model simultaneously obtains current energy influence factors and current resource influence factors in T time;
3) Respectively inputting building operation information which does not participate in model building into an energy prediction model and a resource prediction model, and calculating to obtain energy prediction data and resource prediction data;
4) The energy prediction model transmits energy prediction data to the carbon emission quota calculation module and transmits influence factors of current energy consumption to the carbon emission regulation module;
the resource prediction model transmits resource prediction data to the carbon emission quota calculation module and transmits influence factors of current resource consumption to the carbon emission regulation module;
5) Updating a historical data set after regulation and control, and constructing an energy prediction model and a resource prediction model based on machine learning by using the new historical data set;
6) The carbon emission quota calculating module converts the energy source prediction data and the resource prediction data into carbon emission prediction data according to the carbon emission factor, calculates the hourly carbon emission quota of the building according to the carbon emission prediction data, and transmits the hourly carbon emission quota to the carbon emission regulating module; the current carbon emission calculation module converts current energy resource consumption data into hourly operation carbon emission data according to the carbon emission factor and transmits the hourly operation carbon emission data to the carbon emission regulation and control module;
7) And the carbon emission regulation and control module compares the hourly operation carbon emission data with the hourly carbon emission quota, and if the hourly operation carbon emission data is greater than the hourly carbon emission quota, the carbon emission regulation and control module performs energy and resource system regulation and control according to the influence factors of the current energy consumption and the influence factors of the current resource consumption, and returns to the step 1).
CN202211272055.9A 2022-10-18 2022-10-18 Building carbon emission dynamic optimization management and control system and method based on machine learning Pending CN115526112A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211272055.9A CN115526112A (en) 2022-10-18 2022-10-18 Building carbon emission dynamic optimization management and control system and method based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211272055.9A CN115526112A (en) 2022-10-18 2022-10-18 Building carbon emission dynamic optimization management and control system and method based on machine learning

Publications (1)

Publication Number Publication Date
CN115526112A true CN115526112A (en) 2022-12-27

Family

ID=84704487

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211272055.9A Pending CN115526112A (en) 2022-10-18 2022-10-18 Building carbon emission dynamic optimization management and control system and method based on machine learning

Country Status (1)

Country Link
CN (1) CN115526112A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118469243A (en) * 2024-07-09 2024-08-09 山东建筑大学 Method and system for planning and scheduling carbon bank in whole life of building

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118469243A (en) * 2024-07-09 2024-08-09 山东建筑大学 Method and system for planning and scheduling carbon bank in whole life of building

Similar Documents

Publication Publication Date Title
CN111735178B (en) Air conditioner energy saving system based on elite and SVR regression algorithm and optimization method
CN115759604B (en) Comprehensive energy system optimal scheduling method
CN111780384A (en) Central air-conditioning control system
CN113778215A (en) Method for realizing data center PUE prediction and consumption reduction strategy based on big data
CN115796393A (en) Energy network management optimization method, system and storage medium based on multi-energy interaction
CN117490193A (en) Group control method for central air-conditioning water system
CN110895029A (en) Building load prediction method based on temperature of chilled water
CN113294899A (en) Central air conditioner demand response capability evaluation method based on non-invasive load identification
CN115526112A (en) Building carbon emission dynamic optimization management and control system and method based on machine learning
CN113790516B (en) Global optimization energy-saving control method and system for central air-conditioning refrigeration station and electronic equipment
CN111968008B (en) Comprehensive energy monitoring system and method based on multi-energy cooperation
CN113609778B (en) Multi-objective optimization method and system for comprehensive energy system
CN105184408A (en) Energy-saving method based on electrical appliance electricity utilization effectiveness grading and user electricity utilization behavior recognition
CN116468302A (en) Building park energy structure evaluation method, system, electronic equipment and storage medium
CN114781274A (en) Comprehensive energy system control optimization method and system for simulation and decision-making alternative learning
CN114239922A (en) City multi-energy coordination optimization scheduling method
CN113222350A (en) Comprehensive evaluation method and system for multi-energy system based on sensitivity analysis
CN118630756B (en) Power optimization scheduling method and system for comprehensive energy system
CN118195279B (en) Heating management scheduling method and system based on renewable energy intelligent coupling
CN117646935B (en) Multifunctional air source heat pump control method and system
CN116841197B (en) Operation control method and device for building heat source system
CN111429040A (en) Optimized deployment method for metering equipment of building energy consumption detection system
CN118336835B (en) Renewable energy intelligent control method and system based on multi-energy coupling
CN220169626U (en) Central air conditioning energy management system
Zeng Multi-load prediction modelling for combined energy system of intelligent city buildings

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination