CN113468817B - Ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimized ELM (ELM) - Google Patents
Ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimized ELM (ELM) Download PDFInfo
- Publication number
- CN113468817B CN113468817B CN202110788967.0A CN202110788967A CN113468817B CN 113468817 B CN113468817 B CN 113468817B CN 202110788967 A CN202110788967 A CN 202110788967A CN 113468817 B CN113468817 B CN 113468817B
- Authority
- CN
- China
- Prior art keywords
- locust
- elm
- data
- wind power
- igoa
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000013507 mapping Methods 0.000 claims abstract description 17
- 238000011217 control strategy Methods 0.000 claims abstract description 14
- 230000035772 mutation Effects 0.000 claims abstract description 14
- 238000005457 optimization Methods 0.000 claims abstract description 14
- 230000000739 chaotic effect Effects 0.000 claims abstract description 12
- 238000011156 evaluation Methods 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000012360 testing method Methods 0.000 claims abstract description 8
- 238000012544 monitoring process Methods 0.000 claims abstract description 4
- 230000003247 decreasing effect Effects 0.000 claims description 15
- 241000238814 Orthoptera Species 0.000 claims description 9
- 238000009434 installation Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000002474 experimental method Methods 0.000 claims description 5
- 230000001965 increasing effect Effects 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000005291 chaos (dynamical) Methods 0.000 claims description 3
- 230000007423 decrease Effects 0.000 claims description 3
- 238000005315 distribution function Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 claims 1
- 230000006870 function Effects 0.000 description 10
- 230000000694 effects Effects 0.000 description 5
- 238000011161 development Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 241000254032 Acrididae Species 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000011664 nicotinic acid Substances 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/06—Wind turbines or wind farms
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
- Y02E40/70—Smart grids as climate change mitigation technology in the energy generation sector
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Strategic Management (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Entrepreneurship & Innovation (AREA)
- Medical Informatics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Power Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimization ELM, which comprises the following steps: acquiring historical wind speed and wind power data from a wind power plant data acquisition and monitoring control system, preprocessing the data, selecting a training data sample and a test data sample, and establishing an ELM prediction model; improving GOA, and initializing the population of GOA by adopting a cube chaotic mapping method; updating the decremental coefficients by a sine function based adjustment parameter control strategy; introducing cauchy mutation operation at the position updating position of the locust individual to obtain IGOA; optimizing parameters of the ELM by using the IGOA to obtain optimal parameters; inputting the obtained optimal parameters and test data samples into an ELM prediction model to obtain a prediction result, and selecting three commonly used evaluation indexes in wind power prediction: the prediction performance of the model is evaluated by the root mean square error RMSE, the average absolute error MAE and the maximum absolute error Emax. The method effectively predicts the ultra-short-term wind power and improves the precision of the prediction model.
Description
Technical Field
The invention relates to the field of wind power prediction, in particular to an ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimization ELM (electronic dynamic range).
Background
With the increasing maturity of wind power generation technology, the capacity of a single wind power unit and the scale of a grid-connected wind power plant are continuously enlarged, and the proportion of wind power to the total power generation amount of a power system is increased year by year. The penetrating power of the wind power plant is continuously increased, a series of problems brought to the power system are increasingly prominent, and the serious power system and the power system are safe, stable, economical and reliable to operate. The wind power is predicted timely and accurately, so that the safety, stability, economy and controllability of the power system can be obviously enhanced.
The extreme learning machine (Extreme Learning Machine, ELM) is used to train a single hidden layer feedforward neural network (SLFN) that, unlike conventional SLFN training algorithms, randomly generates input layer weights and hidden layer biases, and in training, only the number of hidden layer neurons needs to be set to obtain a unique optimal solution without adjustment. However, since the gradient descent method requires multiple iterations to achieve the purpose of correcting the weight and the threshold, the training process takes a long time and is easy to sink into local minima. For this reason, a more efficient optimization algorithm is needed to improve the prediction accuracy of the extreme learning machine.
The locust algorithm (Grasshopper Optimization Algorithm, GOA) is a meta-heuristic bionic optimization algorithm proposed by the year 2017 by Saremi, but is easy to sink into local extreme points in the later stage of the algorithm, the convergence speed is slow, and the prediction accuracy is reduced. The method comprises the steps of introducing a cube chaotic mapping method, updating a decreasing coefficient through a sine function-based adjustment parameter control strategy, introducing a cauchy variation operation at the position update position of an individual locust to improve a traditional locust algorithm, optimizing an extreme learning machine by utilizing the improved locust algorithm, acquiring optimal parameters, establishing a new extreme learning machine prediction model, and realizing ultra-short-term wind power prediction.
Disclosure of Invention
The invention aims to: the invention aims to solve the technical problems that parameters are difficult to determine, local optimum is easy to fall into and the like in the existing ELM prediction model, and improves the accuracy of the prediction model.
The technical scheme is as follows: an ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimization ELM (ELM) comprises the following steps:
step 1: acquiring historical wind speed and wind power data from a wind power plant data acquisition and monitoring control system, preprocessing the data, selecting a training data sample and a test data sample, and establishing an ELM prediction model;
step 2: improving GOA, and initializing the population of GOA by adopting a cube chaotic mapping method;
step 3: updating the decremental coefficients by a sine function based adjustment parameter control strategy;
step 4: introducing cauchy mutation operation at the position updating position of the locust individual to obtain IGOA;
step 5: optimizing parameters of the ELM by using the IGOA to obtain optimal parameters;
step 6: inputting the obtained optimal parameters and test data samples into an ELM prediction model to obtain a prediction result, and selecting three commonly used evaluation indexes in wind power prediction: root mean square error RMSE, mean absolute error MAE and maximum absolute error E max And evaluating the prediction performance of the model.
Further, the preprocessing of the data in the step 1 specifically includes:
step 1.1.1: if the collected power data is larger than the wind power installation capacity, replacing the collected power data with the wind power installation capacity, and if the collected power data is negative, replacing the collected power data with zero;
step 1.1.2: for the missing data between adjacent moments, replacing the missing data by an average value of the two adjacent data, and for the abnormal data, replacing the data at the previous moment;
step 1.1.3: normalization processing is performed on the data according to the formula (1):
wherein: y is normalized data, x is raw data, x max And x min Respectively the maximum value and the minimum value in the original data, and the normalization range is 0,1]。
Further, the building of the ELM prediction model in step 1 specifically includes:
step 1.2.1: training data samplesWherein x is i And t i For training data true values, N is the number of nodes of an input layer, N and m are vector space dimensions, and an activation function g (■) of an implicit layer and the number of nodes L of the implicit layer are input into an algorithm model;
step 1.2.2: determining input layer weights w i ∈[-1,1]And hidden layer bias b i ∈[0,1]；
Step 1.2.3: the output matrix H of the hidden layer is calculated according to equation (2),
wherein H= { H 1 ,h 2 ,…,h i }，h i ＝g(w i x i +b i ) Is the output of the hidden layer i-th neuron.
Step 1.2.4: solving an output weight beta, and obtaining the output weight beta through least square solution, namely the formula (3):
wherein: h is the generalized inverse of Moore-Penrose of matrix H;
step 1.2.5: ELM prediction model is obtained by the output weight, namely (4):
wherein: beta i For the connection weight between the i node of the hidden layer and the output layer, w i B, for the connection weight between the input layer and the i node of the hidden layer i Is the bias of the hidden layer i-th node.
Further, step 2 specifically includes:
step 2.1: individuals in a plurality of locust populations are arranged in the D-dimensional space, and the individuals in the locust populations are generated through a cube chaotic mapping formula (5), namely Y= (Y) 1 ,y 2 ,...,y n ) And the value range is y n ∈(0,1)；
Wherein a is a control parameter.
Step 2.2: iterating each dimension of the individual of the locust population N-1 times, thereby generating N-1 individuals of the artificial locust population;
step 2.3: when the individual iteration is completed in all artificial locust populations, mapping into solution space according to formula (6):
wherein: n is locust population number, ub d 、lb d For the upper and lower bounds of the search space,is the d-th dimensional coordinate of the individual of the ith artificial locust population,/for>Is the d-th dimension of the individual of the ith artificial locust population.
Further, in step 2.1, the chaos of the Cubic mapping is related to the value of the control parameter, and when a=2.6, the Cubic mapping has better chaos traversal.
Further, the step 3 specifically includes:
the decreasing coefficient c of the traditional locust algorithm is used for balancing the effects of global search and local development and is proportional to the iteration times, but cannot meet the actual requirements of the algorithm in different periods: the decreasing coefficient c is required to be larger and slowly decreased in the early stage of the algorithm, and the method has the capability and time of carrying out global search to reach the approximate range of the global optimal solution; the decreasing coefficient c is required to be smaller and rapidly decreased in the later stage of the algorithm, the algorithm is rapidly converged to a local optimal solution, and the control strategy of the adjustment parameters based on the sine function is shown as a formula (7):
in the formula, T is the current iteration number of the algorithm, T is the maximum iteration number, and k and u are adjustment parameters.
Further, k= 2,u =2 is selected as the adjustment parameter.
Further, the step 4 specifically includes:
in the traditional algorithm, locust individuals update the position of the next step according to the distance between the current individual and other individuals in the population and the position of the optimal individual, and the problem that the locusts are prone to being in local optimal in the later period is solved. According to the cauchy standard distribution function formula (8), a cauchy variation formula (9) is constructed for the optimal individual, and the position updating formula of the locustus individual after variation is shown as (10):
wherein:for the algorithm iteration to the position of the ith locust in the d-th dimension at the t-th time, ++>For the algorithm to iterate to the position of the jth locust in the d dimension at the t time, d ij (t) is the distance between the ith and the jth locust of the locust population at the t iteration, ++>Is the current optimal individual position,/->The optimal locust position after Cauchy mutation is that r is a random number on (0, 1), p is mutation probability, and p=0.7 is selected through experiments.
Further, step 5 specifically includes:
step 5.1: initializing related algorithm parameters: setting locust population N, space dimension D, maximum iteration number T, upper bound ub and lower bound lb of search space and current iteration number t=1, and initializing to generate the space position of current locust by adopting chaos theory
Step 5.2: calculating the predicted value y trained by ELM network through a formula (11) i And training sample actual value t i As the root mean square error of the locust fitness value fitness, saving the locust position with the minimum current fitness value to a variable
Step 5.3: judging whether the current iteration times T reach the maximum iteration times T or not; if yes, executing the step 5.5, otherwise executing the step 5.4;
step 5.4: the decreasing coefficient c is updated by adjusting the parameter control strategy (7), the distance between locusts is standardized to be [1,4 ]]Updating the locusts by the cauchy variation formulas (9) and (10), returning the current iteration times t=t+1, returning to ELM network training to obtain a new predicted value, updating the fitness value of the locusts according to the formula (11), comparing with the current optimal solution, and updating if the fitness value is superior to the current optimal solutionOtherwise, executing the step 5.3;
step 5.5: outputting the optimal solutionFrom which the input layer weights w and hidden layer biases b required by the ELM network are extracted.
Further, in step 6, RMSE, MAE and E max The three evaluation index formulas are as follows:
E max ＝max(|P ti -P yi |) (14)
wherein: n is the number of samples, P ti For the actual power at time i, P yi For the predicted power at time i, P cap And starting up the total capacity for the wind farm.
The beneficial effects are that: compared with the prior art, the invention has the following remarkable advantages: firstly, initializing the position of a locust by adopting a cube chaotic mapping method for GOAThe method comprises the steps of constructing a decreasing coefficient c of an adjustment parameter control strategy based on a sine function and adding a cauchy mutation operation at a position update position of a locust, so that the mutated locust can increase population diversity, quicken convergence speed, help an algorithm jump out of local optimum, find a new optimum solution and enhance algorithm reliability; and then, the parameters of the ELM are optimized by utilizing the IGOA, so that the aim of improving the accuracy of the prediction model is fulfilled.
Drawings
FIG. 1 is a flow chart of an ultra-short-term wind power prediction method based on IGOA optimization ELM.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent.
Aiming at the defects of the prior art, the method comprises the steps of initializing the position of a locust by adopting a cube chaotic mapping method for GOA, constructing a decreasing coefficient c of an adjustment parameter control strategy based on a sine function, adding a Cauchy mutation operation at the position updating position of an individual locust, enabling the mutated locust to increase population diversity, accelerating convergence speed, helping an algorithm to jump out of local optimum, finding a new optimal solution, enhancing the reliability of the algorithm, optimizing the parameters of ELM by utilizing IGOA, and establishing a new ELM prediction model.
In the invention, the actual measurement data of a wind farm 2021, which is continuously measured for 250 hours in 5 months, is taken as an embodiment, and example simulation is carried out to verify the effect of the invention. FIG. 1 is a flowchart of an ultra-short-term wind power prediction model based on IGOA optimization ELM, and the implementation steps are as follows:
step 1: historical wind speed and wind power data with sampling time resolution of 10min, which are continuously 250h for 5 months, are obtained from a wind power plant data acquisition and monitoring control system, are subjected to a series of preprocessing, training data samples and test data samples are selected, and an ELM prediction model is established.
Wherein performing a series of pre-processing of the data comprises:
step 1.1.1: if the collected power data is larger than the wind power installation capacity, replacing the collected power data with the wind power installation capacity, and if the collected power data is negative, replacing the collected power data with zero;
step 1.1.2: for the missing data between adjacent moments, replacing the missing data by an average value of the two adjacent data, and for the abnormal data, replacing the data at the previous moment;
step 1.1.3: in order to facilitate data processing, the data is normalized according to formula (1).
Wherein: y is normalized data, x is raw data, x max And x min Respectively the maximum value and the minimum value in the original data, and the normalization range is 0,1]。
The building of the ELM prediction model includes:
step 1.2.1: training data samplesx i And t i For training data true values, N is the number of nodes of an input layer, N and m are vector space dimensions, and an activation function g (■) of an implicit layer and the number of nodes L of the implicit layer are input into an algorithm model;
step 1.2.2: determining input layer weights w i ∈[-1,1]And hidden layer bias b i ∈[0,1]；
Step 1.2.3: calculating an output matrix H of the hidden layer, i.e. equation (2):
wherein H= { H 1 ,h 2 ,…,h i }，h i ＝g(w i x i +b i ) Is the output of the hidden layer i-th neuron.
Step 1.2.4: solving the output weight beta, which can be obtained by least squares solution, namely formula (3):
wherein: h is the generalized inverse of Moore-Penrose of matrix H;
step 1.2.5: ELM prediction model is obtained by the output weight, namely (4):
wherein: beta i For the connection weight between the i node of the hidden layer and the output layer, w i B, for the connection weight between the input layer and the i node of the hidden layer i Is the bias of the hidden layer i-th node.
Step two: the GOA is improved, and the population of the GOA is initialized by adopting a cube chaotic mapping method.
The method for initializing the population of the GOA by the cube chaotic mapping method comprises the following steps:
step 2.1: individuals in a plurality of locust populations are arranged in the D-dimensional space, and the individuals in the locust populations are generated through a cube chaotic mapping formula (5), namely Y= (Y) 1 ,y 2 ,...,y n ) And the value range is y n ∈(0,1)；
Wherein a is a control parameter; the chaos of the cube map is related to the value of the parameter a, and through looking up related data and experiments, when a=2.6 is selected, the cube map has better chaos ergodic property.
Step 2.2: then iterating each dimension of the individuals of the locust population for N-1 times, thereby generating N-1 individuals of the artificial locust population;
step 2.3: finally, when the iteration of the individuals in all artificial locust populations is completed, mapping the artificial locust populations into a solution space according to a formula (6):
in the above formula: n is locust population number, ub d 、lb d For the upper and lower bounds of the search space,is the d-th dimensional coordinate of the individual of the ith artificial locust population,/for>Is the d-th dimension of the individual of the ith artificial locust population.
Step three: the decrementing coefficients are then updated by a sine function based tuning parameter control strategy.
The decreasing coefficient c of the traditional locust algorithm is used for balancing the effects of global search and local development and is proportional to the iteration times, but cannot meet the actual requirements of the algorithm in different periods: the decreasing coefficient c is required to be larger and slowly decreased in the early stage of the algorithm, so that the algorithm has enough capacity and time to perform global searching to reach the approximate range of the global optimal solution; the decrease coefficient c is required to be smaller and decrease rapidly in the later stage of the algorithm, so that the algorithm can be quickly converged to the local optimal solution. For this purpose, a sine function-based adjustment parameter control strategy is provided, and the formula is shown as (7):
in the formula, T is the current iteration number of the algorithm, T is the maximum iteration number, k and u are adjustment parameters, and k= 2,u =2 is selected by the invention through experiments.
Step four: and finally, introducing a cauchy mutation operation at the position update position of the locust individual.
In the traditional algorithm, locust individuals update the position of the next step according to the distance between the current individual and other individuals in the population and the position of the optimal individual, and the problem that the locusts are prone to being in local optimal in the later period is solved. According to the cauchy standard distribution function formula (8), a cauchy variation formula (9) is constructed for the optimal individual, and the position updating formula of the locustus individual after variation is shown as (10):
wherein:for the algorithm iteration to the position of the ith locust in the d-th dimension at the t-th time, ++>For the algorithm to iterate to the position of the jth locust in the d dimension at the t time, d ij (t) is the distance between the ith and the jth locust of the locust population at the t iteration, ++>Is the current optimal individual position,/->The optimal locust position after Cauchy mutation is that r is a random number on (0, 1), p is mutation probability, and p=0.7 is selected through experiments.
Step five: and optimizing the parameters of the ELM by using the IGOA to obtain the optimal parameters.
Wherein optimizing parameters of the ELM using the IGOA, obtaining optimal parameters comprises:
step 5.1: initializing related algorithm parameters: setting locust population N, space dimension D, maximum iteration number T, upper bound ub and lower bound lb of search space and current iteration number t=1, and initializing to generate the space position of current locust by adopting chaos theory
Step 5.2: calculating the predicted value y trained by ELM network through a formula (11) i And training sample actual value t i As the root mean square error of the locust fitness value fitness, and saving the locust position with the smallest current fitness value to the variable
Step 5.3: and judging whether the current iteration times T reach the maximum iteration times T or not. If yes, executing step S5, otherwise executing step S4;
step 5.4: updating the decremental coefficient c by adjusting the parameter control strategy formula (7), normalizing the distance between locusts to [1,4 ]]Updating the locusts by the cauchy variation formulas (9) and (10), returning the current iteration times t=t+1, returning to ELM network training to obtain a new predicted value, updating the fitness value of the locusts according to the formula (11), comparing with the current optimal solution, and updating if the fitness value is superior to the current optimal solutionOtherwise, executing the step S3;
step 5.5: outputting the optimal solutionAnd extracts therefrom the input layer weights w and hidden layer biases b required by the ELM network.
Step (a)Sixth,: inputting the obtained optimal parameters and test data samples into an ELM prediction model to obtain a prediction result, and selecting three commonly used evaluation indexes in wind power prediction: root mean square error RMSE, mean absolute error MAE and maximum absolute error E max And evaluating the prediction performance of the model.
Wherein RMSE, MAE and E max The three evaluation index formulas are (12) to (14):
E max ＝max(|P ti -P yi |) (14)
wherein: n is the number of samples, P ti For the actual power at time i, P yi For the predicted power at time i, P cap And starting up the total capacity for the wind farm.
After preprocessing the acquired data, 1050 samples of the first 175h were used for training, 450 samples of the last 75h were used for prediction, and error evaluation analysis was performed on the unoptimized ELM, GOA-ELM, and IGOA-ELM, respectively, and the results are shown in table 1:
TABLE 1
As can be seen from Table 1, the RMSE, MAE and E of the IGOA-ELM model and the GOA-ELM model max The prediction effect is better than that of the non-optimized ELM model, the prediction precision of the ELM model is improved, and the ultra-short-term wind power is effectively predicted.
Aiming at the technical problems that parameters are difficult to determine, local optimum is easy to fall into and the like of the existing ELM prediction model, firstly, a method of cube chaotic mapping is adopted for GOA to initialize the position of a locust, a decreasing coefficient c of a parameter control strategy is adjusted based on a sine function is constructed, and a Cauchy variation operation is added at the position update position of an individual locust, so that the population diversity of the mutated locust can be increased, the convergence speed is accelerated, the algorithm is helped to jump out of local optimum, a new optimal solution is found, the reliability of the algorithm is enhanced, then the parameters of the ELM are optimized by using IGOA, a new ELM prediction model is built, and the ultra-short-term wind power is effectively predicted by taking wind power prediction as an example, so that the effect of the invention is verified.
What is not elaborated on the invention belongs to the prior art which is known to the person skilled in the art.
Claims (7)
1. An ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimization ELM (ELM) is characterized by comprising the following steps:
step 1: acquiring historical wind speed and wind power data from a wind power plant data acquisition and monitoring control system, preprocessing the data, selecting a training data sample and a test data sample, and establishing an ELM prediction model;
the preprocessing of the data in the step 1 specifically includes:
step 1.1.1: if the collected power data is larger than the wind power installation capacity, replacing the collected power data with the wind power installation capacity, and if the collected power data is negative, replacing the collected power data with zero;
step 1.1.2: for the missing data between adjacent moments, replacing the missing data by an average value of the two adjacent data, and for the abnormal data, replacing the data at the previous moment;
step 1.1.3: normalization processing is performed on the data according to the formula (1):
wherein: y is normalized data, x is raw data, x max And x min Respectively the maximum value and the minimum value in the original data, and the normalization range is 0,1]；
Step 2: improving GOA, and initializing the population of GOA by adopting a cube chaotic mapping method;
step 3: updating the decremental coefficients by means of an adjustment parameter control strategy based on a sinusoidal function;
the step 3 specifically includes:
the descending of the decreasing coefficient c is slow at the early stage of the algorithm, and the approximate range of the global optimal solution can be searched globally; the decreasing coefficient c rapidly decreases in the later stage of the algorithm, the algorithm rapidly converges to a local optimal solution, and the control strategy of the adjustment parameters based on the sine function is shown in a formula (2):
in the formula, T is the current iteration times of the algorithm, T is the maximum iteration times, and k and u are adjustment parameters;
step 4: introducing cauchy mutation operation at the position updating position of the locust individual to obtain IGOA;
step 5: optimizing parameters of the ELM by using the IGOA to obtain optimal parameters;
the step 5 specifically includes:
step 5.1: initializing related algorithm parameters: setting locust population N, space dimension D, maximum iteration number T, upper bound ub and lower bound lb of search space and current iteration number t=1, and initializing to generate the space position of current locust by adopting chaos theory
Step 5.2: calculating a predicted value y trained by ELM network through a formula (3) i And training sample actual value t i As the root mean square error of the locust fitness value fitness, saving the locust position with the minimum current fitness value to a variable
Step 5.3: judging whether the current iteration times T reach the maximum iteration times T or not; if yes, executing the step 5.5, otherwise executing the step 5.4;
step 5.4: updating the decremental coefficient c by adjusting the parameter control strategy formula (2), normalizing the distance between locusts to [1,4 ]]Updating the position of the locust by the cauchy variation formula, returning the current iteration times t=t+1, obtaining a new predicted value by ELM network training, updating the fitness value of the locust according to the formula (3), comparing with the current optimal solution, and updating if the fitness value is superior to the current optimal solutionOtherwise, executing the step 5.3;
step 5.5: outputting the optimal solutionExtracting an input layer weight w and an hidden layer bias b required by the ELM network from the ELM network;
step 6: inputting the obtained optimal parameters and test data samples into an ELM prediction model to obtain a prediction result, and selecting three commonly used evaluation indexes in wind power prediction: root mean square error RMSE, mean absolute error MAE and maximum absolute error E max And evaluating the prediction performance of the model.
2. The ultra-short-term wind power prediction method based on the IGOA optimization ELM according to claim 1, wherein the building of the ELM prediction model in the step 1 specifically comprises:
step 1.2.1: training data samplesWherein x is i And t i For training the true value of the data, N is the number of nodes of an input layer, N and m are vector space dimensions, an activation function g (■) of an implicit layer and the number L of nodes of the implicit layer are input into an algorithm moduleIn the form;
step 1.2.2: determining input layer weights w i ∈[-1,1]And hidden layer bias b i ∈[0,1]；
Step 1.2.3: the output matrix H of the hidden layer is calculated according to equation (4),
wherein H= { H 1 ,h 2 ,…,h i }，h i ＝g(w i x i +b i ) The output of the ith neuron which is the hidden layer;
step 1.2.4: solving an output weight beta, and obtaining the output weight beta through least square solution, namely the formula (5):
wherein: h is the generalized inverse of Moore-Penrose of matrix H;
step 1.2.5: ELM prediction model is obtained by the output weight, namely (6):
wherein: beta i For the connection weight between the i node of the hidden layer and the output layer, w i B, for the connection weight between the input layer and the i node of the hidden layer i Is the bias of the hidden layer i-th node.
3. The ultra-short-term wind power prediction method based on the IGOA optimization ELM according to claim 1, wherein the step 2 specifically comprises:
step 2.1: individuals in a plurality of locust populations are arranged in the D-dimensional space, and the individuals in the locust populations are generated through a cube chaotic mapping formula (7), namely Y= (Y) 1 ,y 2 ,...,y n ) And (2) andthe value range is y n ∈(0,1)；
Wherein a is a control parameter;
step 2.2: iterating each dimension of the individual of the locust population N-1 times, thereby generating N-1 individuals of the artificial locust population;
step 2.3: when the individual iteration is completed in all artificial locust populations, mapping into solution space according to equation (8):
wherein: n is locust population number, ub d 、lb d For the upper and lower bounds of the search space,is the d-th dimensional coordinate of the individual of the ith artificial locust population,/for>Is the d-th dimension of the individual of the ith artificial locust population.
4. The ultra-short-term wind power prediction method based on the IGOA-optimized ELM as set forth in claim 3, wherein in the step 2.1, the chaos of the cube map is related to the value of the control parameter, and when a=2.6, the cube map has chaos traversal.
5. The ultra-short term wind power prediction method based on the IGOA optimization ELM of claim 1, wherein the adjustment parameter is k= 2,u =2.
6. The ultra-short-term wind power prediction method based on the IGOA optimization ELM according to claim 1, wherein the step 4 specifically comprises:
adding cauchy mutation operation at the position updating position of the locust individual, increasing population diversity of the mutated locust, and constructing a cauchy mutation formula (10) for the optimal individual according to a cauchy standard distribution function formula (9), wherein the position updating formula of the mutated locust individual is shown as (11):
wherein:for the algorithm iteration to the position of the ith locust in the d-th dimension at the t-th time, ++>For the algorithm iteration to the position of the jth locust in the d dimension at the t-th time, dij (t) is the distance between the ith locust and the jth locust of the locust population at the t-th iteration,/>Is the current optimal individual position,/->The optimal locust position after Cauchy mutation is that r is a random number on (0, 1), p is mutation probability, and p=0.7 is selected through experiments.
7. The ultra-short term wind power prediction method based on IGOA optimization ELM according to claim 1, wherein in the step 6, RMSE, MAE and E are as follows max The three evaluation index formulas are as follows:
E max ＝max(|P ti -P yi |) (14)
wherein: n is the number of samples, P ti For the actual power at time i, P yi For the predicted power at time i, P cap And starting up the total capacity for the wind farm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110788967.0A CN113468817B (en) | 2021-07-13 | 2021-07-13 | Ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimized ELM (ELM) |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110788967.0A CN113468817B (en) | 2021-07-13 | 2021-07-13 | Ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimized ELM (ELM) |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113468817A CN113468817A (en) | 2021-10-01 |
| CN113468817B true CN113468817B (en) | 2023-08-22 |
Family
ID=77880042
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110788967.0A Active CN113468817B (en) | 2021-07-13 | 2021-07-13 | Ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimized ELM (ELM) |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113468817B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114723217B (en) * | 2022-02-15 | 2023-04-28 | 石河子大学 | Comprehensive energy supply side configuration method and system under carbon neutralization target |
| CN116031879A (en) * | 2023-02-28 | 2023-04-28 | 四川大学 | Hybrid intelligent feature selection method suitable for transient voltage stability evaluation of power system |
| CN116526478B (en) * | 2023-07-03 | 2023-09-19 | 南昌工程学院 | Short-term wind power prediction method and system based on improved snake group optimization algorithm |
| CN117557361B (en) * | 2023-11-10 | 2024-04-26 | 纬创软件(武汉)有限公司 | User credit risk assessment method and system based on data analysis |
| CN117348419A (en) * | 2023-11-17 | 2024-01-05 | 南通开放大学(南通建筑职业技术学校、南通市社区教育服务指导中心) | PLC self-adaptive control method and system based on artificial intelligence |
| CN117635245A (en) * | 2023-11-30 | 2024-03-01 | 广东电力交易中心有限责任公司 | Power price prediction method and system based on multilevel frequency domain decomposition and IBM H optimization DELM |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107516150A (en) * | 2017-08-25 | 2017-12-26 | 广东工业大学 | A kind of Forecasting Methodology of short-term wind-electricity power, apparatus and system |
| CN110348631A (en) * | 2019-07-09 | 2019-10-18 | 武汉四创自动控制技术有限责任公司 | A kind of regionality Methods of electric load forecasting and system |
| CN110555548A (en) * | 2019-08-05 | 2019-12-10 | 三峡大学 | ICS-ELM ultra-short-term wind power prediction method based on data mining original error correction |
| CN111105045A (en) * | 2019-11-13 | 2020-05-05 | 温州大学 | Method for constructing prediction model based on improved locust optimization algorithm |
| CN112686464A (en) * | 2021-01-07 | 2021-04-20 | 云南电力技术有限责任公司 | Short-term wind power prediction method and device |
| CN112733462A (en) * | 2021-01-21 | 2021-04-30 | 国网辽宁省电力有限公司阜新供电公司 | Ultra-short-term wind power plant power prediction method combining meteorological factors |
| CN112819238A (en) * | 2021-02-19 | 2021-05-18 | 河海大学 | Short-term wind power prediction method based on chaotic chicken flock optimization algorithm |
-
2021
- 2021-07-13 CN CN202110788967.0A patent/CN113468817B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107516150A (en) * | 2017-08-25 | 2017-12-26 | 广东工业大学 | A kind of Forecasting Methodology of short-term wind-electricity power, apparatus and system |
| CN110348631A (en) * | 2019-07-09 | 2019-10-18 | 武汉四创自动控制技术有限责任公司 | A kind of regionality Methods of electric load forecasting and system |
| CN110555548A (en) * | 2019-08-05 | 2019-12-10 | 三峡大学 | ICS-ELM ultra-short-term wind power prediction method based on data mining original error correction |
| CN111105045A (en) * | 2019-11-13 | 2020-05-05 | 温州大学 | Method for constructing prediction model based on improved locust optimization algorithm |
| CN112686464A (en) * | 2021-01-07 | 2021-04-20 | 云南电力技术有限责任公司 | Short-term wind power prediction method and device |
| CN112733462A (en) * | 2021-01-21 | 2021-04-30 | 国网辽宁省电力有限公司阜新供电公司 | Ultra-short-term wind power plant power prediction method combining meteorological factors |
| CN112819238A (en) * | 2021-02-19 | 2021-05-18 | 河海大学 | Short-term wind power prediction method based on chaotic chicken flock optimization algorithm |
Non-Patent Citations (1)
| Title |
|---|
| SGOA: annealing-behaved grasshopper optimizer for global tasks;Caiyang Yu, Mengxiang Chen等;Engineering with Computers;第38卷;3761–3788 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113468817A (en) | 2021-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113468817B (en) | Ultra-short-term wind power prediction method based on IGOA (insulated gate bipolar transistor) optimized ELM (ELM) | |
| CN114092832B (en) | High-resolution remote sensing image classification method based on parallel hybrid convolutional network | |
| WO2018045642A1 (en) | A bus bar load forecasting method | |
| CN109242223B (en) | Quantum support vector machine evaluation and prediction method for urban public building fire risk | |
| CN108205717A (en) | A kind of photovoltaic generation power Multiple Time Scales Forecasting Methodology | |
| CN104077632B (en) | A kind of wind electric field power prediction method based on deep neural network | |
| CN106251001A (en) | A kind of based on the photovoltaic power Forecasting Methodology improving fuzzy clustering algorithm | |
| CN107346459B (en) | Multi-mode pollutant integrated forecasting method based on genetic algorithm improvement | |
| CN106650784A (en) | Feature clustering comparison-based power prediction method and device for photovoltaic power station | |
| CN110942205A (en) | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM | |
| CN112149883A (en) | Photovoltaic power prediction method based on FWA-BP neural network | |
| CN114462718A (en) | CNN-GRU wind power prediction method based on time sliding window | |
| CN114282646B (en) | Optical power prediction method and system based on two-stage feature extraction and BiLSTM improvement | |
| CN115374995A (en) | Distributed photovoltaic and small wind power station power prediction method | |
| CN107909221A (en) | Power-system short-term load forecasting method based on combination neural net | |
| CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
| CN115995810A (en) | Wind power prediction method and system considering weather fluctuation self-adaptive matching | |
| CN113344243B (en) | Wind speed prediction method and system for optimizing ELM (ELM) based on improved Harris eagle algorithm | |
| CN117154690A (en) | Photovoltaic power generation power prediction method and system based on neural network | |
| CN115759389A (en) | Day-ahead photovoltaic power prediction method based on weather type similar day combination strategy | |
| CN116345555A (en) | CNN-ISCA-LSTM model-based short-term photovoltaic power generation power prediction method | |
| CN115032682A (en) | Multi-station seismic source parameter estimation method based on graph theory | |
| CN111488974B (en) | Ocean wind energy downscaling method based on deep learning neural network | |
| CN113435595A (en) | Two-stage optimization method for extreme learning machine network parameters based on natural evolution strategy | |
| CN116826737A (en) | Photovoltaic power prediction method, device, storage medium and equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20211001 Assignee: Shanghai Yanqiao Information Technology Co.,Ltd. Assignor: HUAIYIN INSTITUTE OF TECHNOLOGY Contract record no.: X2023980047724 Denomination of invention: A Ultra Short Term Wind Power Prediction Method Based on IGOA Optimized ELM Granted publication date: 20230822 License type: Common License Record date: 20231121 |
|
| EE01 | Entry into force of recordation of patent licensing contract | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240717 Address after: 230000 Room 203, building 2, phase I, e-commerce Park, Jinggang Road, Shushan Economic Development Zone, Hefei City, Anhui Province Patentee after: Hefei Jiuzhou Longteng scientific and technological achievement transformation Co.,Ltd. Country or region after: China Address before: 223000 No.1, Meicheng East Road, Huaian Economic and Technological Development Zone, Jiangsu Province Patentee before: HUAIYIN INSTITUTE OF TECHNOLOGY Country or region before: China |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20241104 Address after: 137100 Er Long Xiang Guang Ming Cun, Taonan City, Baicheng City, Jilin Province Patentee after: DATANG XIANGYANG WIND POWER GENERATION Co.,Ltd. Country or region after: China Address before: 230000 Room 203, building 2, phase I, e-commerce Park, Jinggang Road, Shushan Economic Development Zone, Hefei City, Anhui Province Patentee before: Hefei Jiuzhou Longteng scientific and technological achievement transformation Co.,Ltd. Country or region before: China |