CN109397703B - Fault detection method and device - Google Patents
Fault detection method and device Download PDFInfo
- Publication number
- CN109397703B CN109397703B CN201811269612.5A CN201811269612A CN109397703B CN 109397703 B CN109397703 B CN 109397703B CN 201811269612 A CN201811269612 A CN 201811269612A CN 109397703 B CN109397703 B CN 109397703B
- Authority
- CN
- China
- Prior art keywords
- printer
- parameters
- induction
- detected
- state
- 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
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C64/00—Additive manufacturing, i.e. manufacturing of three-dimensional [3D] objects by additive deposition, additive agglomeration or additive layering, e.g. by 3D printing, stereolithography or selective laser sintering
- B29C64/30—Auxiliary operations or equipment
- B29C64/386—Data acquisition or data processing for additive manufacturing
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C64/00—Additive manufacturing, i.e. manufacturing of three-dimensional [3D] objects by additive deposition, additive agglomeration or additive layering, e.g. by 3D printing, stereolithography or selective laser sintering
- B29C64/30—Auxiliary operations or equipment
- B29C64/386—Data acquisition or data processing for additive manufacturing
- B29C64/393—Data acquisition or data processing for additive manufacturing for controlling or regulating additive manufacturing processes
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B33—ADDITIVE MANUFACTURING TECHNOLOGY
- B33Y—ADDITIVE MANUFACTURING, i.e. MANUFACTURING OF THREE-DIMENSIONAL [3-D] OBJECTS BY ADDITIVE DEPOSITION, ADDITIVE AGGLOMERATION OR ADDITIVE LAYERING, e.g. BY 3-D PRINTING, STEREOLITHOGRAPHY OR SELECTIVE LASER SINTERING
- B33Y50/00—Data acquisition or data processing for additive manufacturing
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B33—ADDITIVE MANUFACTURING TECHNOLOGY
- B33Y—ADDITIVE MANUFACTURING, i.e. MANUFACTURING OF THREE-DIMENSIONAL [3-D] OBJECTS BY ADDITIVE DEPOSITION, ADDITIVE AGGLOMERATION OR ADDITIVE LAYERING, e.g. BY 3-D PRINTING, STEREOLITHOGRAPHY OR SELECTIVE LASER SINTERING
- B33Y50/00—Data acquisition or data processing for additive manufacturing
- B33Y50/02—Data acquisition or data processing for additive manufacturing for controlling or regulating additive manufacturing processes
Landscapes
- Chemical & Material Sciences (AREA)
- Engineering & Computer Science (AREA)
- Materials Engineering (AREA)
- Manufacturing & Machinery (AREA)
- Physics & Mathematics (AREA)
- Mechanical Engineering (AREA)
- Optics & Photonics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The application provides a fault detection method and a fault detection device, wherein the method comprises the following steps: acquiring sensing parameters detected by a sensor arranged on the printer, wherein the sensing parameters are used for reflecting the current working state of the printer; then, extracting the characteristics of the obtained induction parameters to obtain characteristic vectors; inputting the obtained feature vector into a classifier model obtained by pre-training, and determining a state evaluation value, wherein the state evaluation value is used for reflecting the possibility of the printer failure; and finally, when the state evaluation value is larger than a first preset threshold value, performing alarm processing. By the mode, the working state of the printer can be monitored in real time, and the alarm is given when the printer breaks down, so that the working efficiency is improved.
Description
Technical Field
The present application relates to the field of fault detection technologies, and in particular, to a fault detection method and apparatus.
Background
Three-dimensional (3Dimensions, 3D) printing technology is one of the rapid prototyping technologies, and by virtue of the advantage that almost any complex parts can be generated without using a mold and a tool, people's lives are rapidly entered, and production efficiency is greatly improved.
However, in the prior art, the time consumed for printing a complete and usable product is long, the problem of printing failure caused by material shortage or 3D printing machine failure and the like easily occurs in the printing process, and the supervision of the printing state of the 3D printer in the prior art mainly uses manpower, so that the efficiency is low.
Disclosure of Invention
In view of this, an object of the present application is to provide a method and an apparatus for detecting a fault, so as to monitor a state of a 3D printer in real time, and perform a state alarm when a fault occurs, thereby improving work efficiency.
In a first aspect, an embodiment of the present application provides a fault detection method, where the method includes:
acquiring sensing parameters detected by a sensor arranged on a printer, wherein the sensing parameters are used for reflecting the current working state of the printer;
extracting the characteristics of the obtained induction parameters to obtain characteristic vectors;
inputting the obtained feature vector into a classifier model obtained by pre-training, and determining a state evaluation value, wherein the state evaluation value is used for reflecting the possibility of the printer failure;
and when the state evaluation value is larger than a first preset threshold value, performing alarm processing.
With reference to the first aspect, an embodiment of the present application provides a first possible implementation manner of the first aspect, where the sensing parameter includes: a frequency detected by a vibration sensor; acceleration detected by an acceleration sensor.
With reference to the first aspect, an embodiment of the present application provides a second possible implementation manner of the first aspect, where the performing feature extraction on the obtained sensing parameters to obtain a feature vector specifically includes:
summing induction parameters respectively detected by the same type of sensors arranged at different positions of the printer at the same time to obtain first data corresponding to each type of sensor;
randomly combining induction parameters respectively detected by all sensors arranged on a printer;
respectively carrying out summation operation on each group of induction parameters obtained after random combination to obtain second data corresponding to each group of induction parameters;
performing a division operation on each group of induction parameters obtained after random combination to obtain third data corresponding to each group of induction parameters;
and generating the characteristic vector according to the first data, the second data, the third data and the sensing parameters detected by each sensor installed on a printer.
With reference to the first aspect, an embodiment of the present application provides a third possible implementation manner of the first aspect, where after obtaining the sensing parameter detected by the sensor installed on the printer, before performing feature extraction on the obtained sensing parameter, the method further includes:
calculating the average value of the same type of induction parameters respectively collected under each time window in continuous time, wherein the time window is the time length consumed by obtaining N same type of induction parameters, and N is a positive integer;
calculating the standard deviation between the average values respectively corresponding to different time windows;
normalizing each obtained induction parameter in the same induction parameter by using the calculated average value and standard deviation to obtain the induction parameter after normalization;
the characteristic extraction of the acquired induction parameters comprises the following steps:
and performing feature extraction on the induction parameters after the normalization processing.
With reference to the first aspect, an embodiment of the present application provides a fourth possible implementation manner of the first aspect, where the obtained classifier model is trained according to the following manner:
acquiring a plurality of sample data with status labels, wherein the status labels comprise a first label indicating that the printer is in failure and a second label indicating that the printer is normal;
sequentially inputting the plurality of sample data with the state labels into a classifier model to obtain a training result of each sample data, wherein the training result is the state of the printer detected by the classifier model, and the state of the printer can be a normal state or a fault state;
and when the accuracy of the training result is smaller than a second preset threshold, adjusting the model parameters of the classifier model until the accuracy of the training result is larger than or equal to the second preset threshold.
In a second aspect, an embodiment of the present application further provides a fault detection apparatus, including:
the system comprises an acquisition module, a processing module and a control module, wherein the acquisition module is used for acquiring sensing parameters detected by a sensor arranged on a printer, and the sensing parameters are used for reflecting the current working state of the printer;
the characteristic extraction module is used for extracting the characteristics of the obtained induction parameters to obtain characteristic vectors;
the determining module is used for inputting the obtained feature vector into a classifier model obtained by pre-training and determining a state evaluation value, wherein the state evaluation value is used for reflecting the possibility of the printer failure;
and the alarm module is used for carrying out alarm processing when the state evaluation value is greater than a first preset threshold value.
With reference to the second aspect, the present application provides a first possible implementation manner of the second aspect, where the sensing parameter includes: a frequency detected by a vibration sensor; acceleration detected by an acceleration sensor.
With reference to the second aspect, an embodiment of the present application provides a second possible implementation manner of the second aspect, where the feature extraction module, when performing feature extraction on the acquired sensing parameters, is specifically configured to:
summing induction parameters respectively detected by the same type of sensors arranged at different positions of the printer at the same time to obtain first data corresponding to each type of sensor;
randomly combining induction parameters respectively detected by all sensors arranged on a printer;
respectively carrying out summation operation on each group of induction parameters obtained after random combination to obtain second data corresponding to each group of induction parameters;
performing a division operation on each group of induction parameters obtained after random combination to obtain third data corresponding to each group of induction parameters;
and generating the characteristic vector according to the first data, the second data, the third data and the sensing parameters detected by each sensor installed on a printer.
With reference to the second aspect, this application provides a third possible implementation manner of the second aspect, where after obtaining the sensing parameters detected by the sensor installed on the printer, before performing feature extraction on the obtained sensing parameters, the apparatus is further configured to:
calculating the average value of the same type of induction parameters respectively collected under each time window in continuous time, wherein the time window is the time length consumed by obtaining N same type of induction parameters, and N is a positive integer;
calculating the standard deviation between the average values respectively corresponding to different time windows;
normalizing each obtained induction parameter in the same induction parameter by using the calculated average value and standard deviation to obtain the induction parameter after normalization;
the characteristic extraction of the acquired induction parameters comprises the following steps:
and performing feature extraction on the induction parameters after the normalization processing.
With reference to the second aspect, an embodiment of the present application provides a fourth possible implementation manner of the second aspect, where the determining module, when training the classifier model, is specifically configured to:
acquiring a plurality of sample data with status labels, wherein the status labels comprise a first label indicating that the printer is in failure and a second label indicating that the printer is normal;
sequentially inputting the plurality of sample data with the state labels into a classifier model to obtain a training result of each sample data, wherein the training result is the state of the printer detected by the classifier model, and the state of the printer can be a normal state or a fault state;
and when the accuracy of the training result is smaller than a second preset threshold, adjusting the model parameters of the classifier model until the accuracy of the training result is larger than or equal to the second preset threshold.
In a third aspect, an embodiment of the present application further provides an electronic device, including: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, the processor and the memory communicate via the bus when the electronic device is running, and the machine-readable instructions are executed by the processor to perform the steps of the fault detection method described above in the first aspect and any possible implementation manner of the first aspect.
In a fourth aspect, this embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and the computer program is executed by a processor to perform the steps of the foregoing first aspect, or the foregoing fault detection method in any possible implementation manner of the first aspect.
According to the fault detection method and device provided by the embodiment of the application, the sensing parameters detected by the sensor arranged on the printer are obtained, wherein the sensing parameters are used for reflecting the current working state of the printer; then, extracting the characteristics of the obtained induction parameters to obtain characteristic vectors; inputting the obtained feature vector into a classifier model obtained by pre-training, and determining a state evaluation value, wherein the state evaluation value is used for reflecting the possibility of the printer failure; and finally, when the state evaluation value is larger than a first preset threshold value, performing alarm processing. By the mode, the working state of the printer can be monitored in real time, and the alarm is given when the printer breaks down, so that the working efficiency is improved.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
Fig. 1 is a schematic view illustrating an application scenario of a fault detection method and apparatus provided in an embodiment of the present application;
fig. 2 is a schematic flowchart illustrating a fault detection method and apparatus provided in an embodiment of the present application;
FIG. 3 illustrates a waveform of raw data of sensing parameters provided by an embodiment of the present application;
FIG. 4 is a waveform diagram of the sensing parameters after noise reduction provided by the embodiment of the present application;
FIG. 5 is a graph illustrating normalized sensing parameter waveforms provided by an embodiment of the present application;
FIG. 6 is a flow chart illustrating classifier model training provided by an embodiment of the present application;
fig. 7 shows an architecture diagram of a fault monitoring apparatus 700 provided by an embodiment of the present application;
fig. 8 shows a schematic structural diagram of an electronic device 800 provided in an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.
At present, the state of the printer is mainly monitored through manpower, and the efficiency is low. In view of the above-mentioned problems, embodiments of the present application provide a method and an apparatus for fault detection, which are described below by way of embodiments.
First, referring to fig. 1, an application scenario diagram applicable to the embodiment of the present application is exemplarily illustrated. The scene comprises the following steps: 3D printer, computer and cloud platform. Wherein, a sensor is arranged on the 3D printer, and the sensor can be a vibration sensor, an acceleration sensor and the like; before the induction parameters detected by the sensor are sent to the computer, the induction parameters detected by the sensor can be converted into a data format which can be recognized by the computer through the single chip microcomputer. For example, the singlechip can be an Arduino type singlechip, and is used for receiving the sensing parameters detected by the sensor and sending the sensing parameters to the computer; the computer is used for receiving the induction parameters sent by the Arduino and sending the induction parameters to the cloud platform; the cloud platform detects the state of the printer based on the trained classifier model and the induction parameters sent by the computer, and judges whether the working state of the 3D printer is in a fault state or a normal state.
The method comprises the following steps that a pre-trained classifier model can be trained by a computer, specifically, a large number of induction parameters stored by the computer are selected, different labels are respectively configured for the induction parameters stored by the computer, the labels can be labels representing printer faults or labels representing normal operation of the printer, then a large number of induction parameters with labels are sequentially input into a pre-established classifier model to obtain training results of the induction parameters with labels, and the accuracy of the training results is determined by comparing the training results of each induction parameter with the label states of the induction parameters; adjusting the model parameters until the accuracy of the training result of the classifier model is greater than or equal to a preset threshold;
in the embodiment of the application, when the method is actually applied, the sensing parameters can be stored in the cloud platform, and the computer acquires the sensing parameters from the cloud platform, completes operations such as model training and printer state detection and the like so as to prompt a user of the state of the printer; or the cloud platform stores the induction parameters and completes operations such as model training and printer state detection.
To facilitate understanding of the present embodiment, a method for detecting a fault disclosed in the embodiments of the present application will be described in detail first.
Example one
Referring to fig. 2, a schematic flow chart of a fault detection method and apparatus provided in the embodiment of the present application includes the following steps:
s201, acquiring sensing parameters detected by a sensor installed on the printer.
In one possible embodiment, the printer may be a 3D printer; the sensor can be a vibration sensor and an acceleration sensor; the sensing parameter may be a frequency detected by a vibration sensor and an acceleration detected by an acceleration sensor.
The sensor may be debugged before the sensing parameters are obtained by the sensor, for example, the sensitivity of the sensor may be debugged.
In the embodiment of the present application, a vibration sensor and an acceleration sensor are used, but in practical applications, these sensors are not limited to these two sensors. For example, other alternative sensors, such as temperature and humidity sensors, may also be used. After a large number of sensing parameters are acquired from different sensors, the sensing parameters most relevant to the printer fault can be obtained by analyzing the sensing parameters detected by the different sensors.
After the sensor detects the sensing parameters, the sensing parameters cannot be directly identified by the computer, and the sensing parameters need to be converted into a data format which can be identified by the computer. In one possible implementation mode, the sensor is connected to the microcomputer Arduino, the sensing parameters detected by the sensor are converted into a data format which can be recognized by the computer through the microcomputer Arduino, then the sensing parameters are sent to the computer through a communication interface of the microcomputer Arduino, then the sensing parameters are forwarded to the cloud platform through the computer, the cloud platform performs data analysis on the sensing parameters, and the current working state of the 3D printer is judged.
In one possible implementation, a vibration sensor and an acceleration sensor are mounted on each of the axes of the 3D printer X, Y, Z, wherein a sensor mounted on the X-axis is used to detect a sensing parameter in the lateral direction, a sensor mounted on the Y-axis is used to detect a sensing parameter in the longitudinal direction, and a sensor mounted on the Z-axis is used to detect a sensing parameter in the depth direction; and simultaneously detects the vibration frequency and acceleration detected by the sensor installed on the X, Y, Z shaft to detect the printing state of the 3D printer.
S202, extracting the characteristics of the obtained induction parameters to obtain characteristic vectors.
After the cloud platform acquires the induction parameters detected by the sensor installed on the printer, data preprocessing can be performed on the acquired induction parameters before feature extraction, wherein the data preprocessing comprises noise reduction processing and normalization processing on the induction parameters, and specifically:
calculating the average value of the same type of induction parameters respectively collected under each time window in continuous time, wherein the time window is the time length consumed by obtaining N same type of induction parameters, and N is a positive integer; because the numerical fluctuation of the same sensing parameter is large when noise interference exists, the numerical fluctuation degree can be reduced after the average value of the same sensing parameter collected under each time window in different time windows is calculated, and the purpose of noise reduction is achieved; then calculating the standard deviation between the average values respectively corresponding to different time windows; and finally, performing normalization processing on each induction parameter in the acquired same induction parameter by using the calculated average value and standard deviation to obtain the induction parameter after the normalization processing.
Specifically, when the induction parameters of any time window are normalized, the following formula is adopted:
wherein X is an induction parameter of any time window; mu is the average value of the same induction parameter respectively collected in different time windows in continuous time; sigma is the standard deviation between the average values respectively corresponding to different time windows; x*The induction parameters after normalization.
For example, to perform data preprocessing on the acceleration detected by the acceleration sensor in the 10 th to 50 th seconds, if the acceleration sensor collects data 30 times per second, 1200 pieces of acceleration data can be obtained in the 10 th to 50 th seconds. Setting a time window as the time length consumed by acquiring 50 data, wherein the process of performing noise reduction processing on the data of each time window is to calculate the average value of the 50 data under each time window, then using the calculated average value as the acceleration value of the time of the last data under each time window, and calculating the average value of all the acceleration data according to the time windows to obtain 1200/50-24 acceleration data after noise reduction.
Because the value ranges of the induction parameters detected by the acceleration sensor and the vibration sensor are different, the induction parameters after noise reduction are normalized to be suitable for the same classifier model. If the acceleration is to be normalized, the average value and the standard deviation of the acceleration data after noise reduction can be calculated, then each acceleration data after noise reduction is calculated according to the formula (1-1), and finally the acceleration data after normalization processing is obtained.
Specifically, the result of the data preprocessing is shown in fig. 3, 4 and 5, in which the curves represent the sensed parameters detected by different sensors, such as frequency, acceleration, temperature, etc. FIG. 3 shows the original sensing parameters detected by the sensor, the original data is interfered by noise and has larger fluctuation; fig. 4 shows data obtained by denoising the original sensing parameters, where the data fluctuation after denoising is small, but the value ranges of the sensing parameters detected by different sensors are different; fig. 5 is data obtained by performing normalization processing on the data after noise reduction, where the normalized data has small fluctuation and the value ranges of the sensing parameters are close to each other.
In one possible implementation, the sensing parameters detected by the vibration sensor and the acceleration sensor mounted on the shaft of the 3D printer X, Y, Z may be subjected to feature extraction, specifically:
summing induction parameters respectively detected by the same type of sensors arranged at different positions of the printer at the same time to obtain first data corresponding to each type of sensor;
randomly combining induction parameters respectively detected by all sensors arranged on a printer;
respectively carrying out summation operation on each group of induction parameters obtained after random combination to obtain second data corresponding to each group of induction parameters;
performing a division operation on each group of induction parameters obtained after random combination to obtain third data corresponding to each group of induction parameters;
and generating a characteristic vector according to the first data, the second data and the third data and the sensing parameters detected by each sensor installed on the printer.
For example, the sum of the frequencies detected by the vibration sensors installed on the shaft of the 3D printer X, Y, Z at the same time may be obtained by first performing a summation operation; then, summing the accelerations detected by the acceleration sensors arranged on the X, Y, Z shafts at the same time to obtain the sum of the accelerations; the sum of the acceleration and the sum of the frequency jointly form first data;
then, induction parameters detected by all sensors arranged on the 3D printer at the same time are combined in pairs, and the total induction parameters are total
The combination is used for carrying out summation operation aiming at each group of induction parameters to obtain second data corresponding to each group of induction parameters; and performing division operation on each group of induction parameters to obtain third data corresponding to each group of induction parameters.
For example, the sensing parameters detected by the sensors are respectively a1, a2, A3, a4, a5 and a6, and the combination obtained by combining the sensing parameters two by two is as follows: a1A2, A1A3, A1A4, A1A5, A1A6, A2A3, A2A4, A2A5, A2A6, A3A4, A3A5, A3A6, A4A5, A4A6, A5 A6; and aiming at each group of the obtained induction parameters, carrying out summation operation to obtain second data of each group of the induction parameters, and carrying out division operation to obtain third data of each group of the induction parameters.
And then inputting the first data, the second data and the third data at the same time and the sensing parameters detected by each sensor installed on the printer into a classifier model trained in advance as a group of feature vectors.
And S203, inputting the obtained feature vector into a classifier model obtained by pre-training, and determining a state evaluation value.
Before this step is executed, the classifier model needs to be trained, in this embodiment, sample data with a first label indicating that the printer is in a fault and sample data with a second label indicating that the printer is normal may be referred to train the classifier model, specifically, referring to the flowchart of classifier model training shown in fig. 6:
601, acquiring a plurality of sample data with status labels, wherein the status labels comprise a first label indicating that a printer has a fault and a second label indicating that the printer is normal;
step 602, sequentially inputting a plurality of sample data with state labels into a classifier model to obtain a training result of each sample data, wherein the training result is the state of the printer detected by the classifier model, and the state of the printer can be a normal state or a fault state;
step 603, comparing the training result of each sample data with the state label of the sample data, and determining the accuracy of the training result;
step 604, judging whether the accuracy of the training result is smaller than a second preset threshold, if so, ending the process; if the determination result is no, go to step 605;
step 605, adjusting model parameters of the classifier model.
The calculation process for the model adjustment is as follows:
first, a linear regression function is constructed:
z＝θ0+θ1x1+θ2x2+…+θnxn＝θTx
wherein, thetaiAs a model parameter, xiFor the input feature vector, i is a positive integer from 1 to n, n is the number of input data, i.e. the firstData, second data, third data, and a total number of sensing parameters detected by each sensor mounted on the printer, x ═ x1,x2,x3,x4,...xn}。
Then, a linear regression model is constructed as follows:
hθ(x) The model output represents the probability of the printer failure, and the value of the probability is between 0 and 1. h isθ(x)<0.5 indicates that the probability of the printer failure is less than 50%, and the working state of the printer is considered to be normal, hθ(x)>0.5 indicates that the probability of the printer failing is more than 50%, and the operating state of the printer is considered to be a failure.
Let y be 0 for hθ(x)<0.5, let y be 1 for hθ(x)>0.5, then:
P(y＝1|x；θ)＝hθ(x)
P(y＝0|x；θ)＝1-hθ(x)
the corresponding probability function is:
P(y|x；θ)＝(hθ(x))y*(1-hθ(x))1-y
the total number of the input m data, namely the first data, the second data, the third data and the sensing parameters detected by each sensor installed on the printer is set as m, and the corresponding likelihood function is as follows:
the log-likelihood function is:
finally, let
Wherein J (theta) is a loss function used for measuring the inconsistency degree of the predicted value and the true value of the classifier model. Therefore, the model training process is to continuously adjust the model parameter thetaiThereby minimizing the loss function J (θ).
For example, the model parameter θ may beiIs set to {2, 3, 5, 6, 9}, {8, 5, 4, 6, 1}, {5, 8, 6, 9, 5}, etc., and the classifier model is trained based on each model parameter. If the accuracy obtained when the model parameter value is {2, 3, 5, 6, 9} is the highest and exceeds the preset threshold, the accuracy obtained when the model parameter value is {8, 5, 4, 6, 1} also exceeds the preset threshold, but the accuracy obtained when no model parameter value is {2, 3, 5, 6, 9} is high, then determining {2, 3, 5, 6, 9} as the model parameter of the classifier model obtained by training.
After model training, the optimized model parameter theta can be obtainediSending the trained model to the cloud platform, and receiving the sensing parameters of the sensor by the cloud platform into the trained classifier model received by the cloud platform, wherein h can be obtained by outputtingθ(x)，hθ(x) That is, the state evaluation value.
And S204, when the state evaluation value is larger than a first preset threshold value, performing alarm processing.
Inputting the feature vector into a classifier model trained in advance, outputting a state evaluation value which can be obtained, and when the state evaluation value is greater than a first preset threshold value, determining that the state of the printer is a fault state and performing alarm processing; when the state evaluation value is less than or equal to the first preset threshold value, the printer state can be confirmed to be a normal state, and input calculation of a newly acquired feature vector can be performed.
By the method, the acquired sensing parameters detected by the sensor arranged on the printer are analyzed, the working state of the printer can be monitored in real time, manual inspection is avoided, and the working efficiency is improved.
Example two
An embodiment of the present application provides a fault detection apparatus, and as shown in fig. 7, is an architecture schematic diagram of a fault monitoring apparatus 700 provided in an embodiment of the present application, the apparatus 700 includes: the system comprises an acquisition module 701, a feature extraction module 702, a determination module 703 and an alarm module 704.
Specifically, the acquiring module 701 is configured to acquire sensing parameters detected by a sensor installed on a printer, where the sensing parameters are used to reflect a current working state of the printer;
a feature extraction module 702, configured to perform feature extraction on the obtained sensing parameters to obtain feature vectors;
a determining module 703, configured to input the obtained feature vector into a classifier model obtained through pre-training, and determine a state evaluation value, where the state evaluation value is used to reflect a possibility that the printer fails;
an alarm module 704, configured to perform alarm processing when the state evaluation value is greater than a first preset threshold.
In one possible embodiment, the sensing parameters include: a frequency detected by a vibration sensor; acceleration detected by an acceleration sensor.
In addition, in a possible implementation manner, the feature extraction module 702, when performing feature extraction on the acquired sensing parameters, is specifically configured to:
summing induction parameters respectively detected by the same type of sensors arranged at different positions of the printer at the same time to obtain first data corresponding to each type of sensor;
randomly combining induction parameters respectively detected by all sensors arranged on a printer;
respectively carrying out summation operation on each group of induction parameters obtained after random combination to obtain second data corresponding to each group of induction parameters;
performing a division operation on each group of induction parameters obtained after random combination to obtain third data corresponding to each group of induction parameters;
and generating the characteristic vector according to the first data, the second data, the third data and the sensing parameters detected by each sensor installed on a printer.
In a possible embodiment, after acquiring the sensing parameters detected by the sensor installed on the printer, and before performing feature extraction on the acquired sensing parameters, the apparatus is further configured to:
calculating the average value of the same type of induction parameters respectively collected under each time window in continuous time, wherein the time window is the time length consumed by obtaining N same type of induction parameters, and N is a positive integer;
calculating the standard deviation between the average values respectively corresponding to different time windows;
normalizing each obtained induction parameter in the same induction parameter by using the calculated average value and standard deviation to obtain the induction parameter after normalization;
the characteristic extraction of the acquired induction parameters comprises the following steps:
and performing feature extraction on the induction parameters after the normalization processing.
In addition, in a possible implementation, the determining module 703 is specifically configured to, when training the classifier model:
acquiring a plurality of sample data with status labels, wherein the status labels comprise a first label indicating that the printer is in failure and a second label indicating that the printer is normal;
sequentially inputting the plurality of sample data with the state labels into a classifier model to obtain a training result of each sample data, wherein the training result is the state of the printer detected by the classifier model, and the state of the printer can be a normal state or a fault state;
and when the accuracy of the training result is smaller than a second preset threshold, adjusting the model parameters of the classifier model until the accuracy of the training result is larger than or equal to the second preset threshold.
EXAMPLE III
As shown in fig. 8, a schematic structural diagram of an electronic device 800 according to a fourth embodiment of the present application includes: a processor 801, a memory 802, and a bus 803;
the memory 802 stores machine-readable instructions executable by the processor 801 (for example, including execution instructions corresponding to the obtaining module 701, the feature extracting module 702, the determining module 703, and the alarm module 704 in fig. 7, and the like), when the electronic device 800 runs, the processor 801 communicates with the memory 802 through the bus 803, and when the processor 801 executes the following processes:
acquiring sensing parameters detected by a sensor arranged on a printer, wherein the sensing parameters are used for reflecting the current working state of the printer;
extracting the characteristics of the obtained induction parameters to obtain characteristic vectors;
inputting the obtained feature vector into a classifier model obtained by pre-training, and determining a state evaluation value, wherein the state evaluation value is used for reflecting the possibility of the printer failure;
and when the state evaluation value is larger than a first preset threshold value, performing alarm processing.
In addition, in the processing executed by the processor 801, the sensing parameters include: a frequency detected by a vibration sensor; acceleration detected by an acceleration sensor.
In addition, in the processing executed by the processor 801, the extracting features of the acquired sensing parameters to obtain a feature vector specifically includes:
summing induction parameters respectively detected by the same type of sensors arranged at different positions of the printer at the same time to obtain first data corresponding to each type of sensor;
randomly combining induction parameters respectively detected by all sensors arranged on a printer;
respectively carrying out summation operation on each group of induction parameters obtained after random combination to obtain second data corresponding to each group of induction parameters;
performing a division operation on each group of induction parameters obtained after random combination to obtain third data corresponding to each group of induction parameters;
and generating the characteristic vector according to the first data, the second data, the third data and the sensing parameters detected by each sensor installed on a printer.
In addition, the processor 801 further includes, after acquiring the sensing parameter detected by the sensor mounted on the printer and before performing feature extraction on the acquired sensing parameter, in the processing executed by the processor:
calculating the average value of the same type of induction parameters respectively collected under each time window in continuous time, wherein the time window is the time length consumed by obtaining N same type of induction parameters, and N is a positive integer;
calculating the standard deviation between the average values respectively corresponding to different time windows;
and normalizing each acquired induction parameter in the same induction parameter by using the calculated average value and standard deviation to obtain the induction parameter after normalization.
The characteristic extraction of the acquired induction parameters comprises the following steps:
and performing feature extraction on the induction parameters after the normalization processing.
In addition, the processor 801 performs a process of training the obtained classifier model according to the following manner:
acquiring a plurality of sample data with status labels, wherein the status labels comprise a first label indicating that the printer is in failure and a second label indicating that the printer is normal;
sequentially inputting the plurality of sample data with the state labels into a classifier model to obtain a training result of each sample data, wherein the training result is the state of the printer detected by the classifier model, and the state of the printer can be a normal state or a fault state;
and when the accuracy of the training result is smaller than a second preset threshold, adjusting the model parameters of the classifier model until the accuracy of the training result is larger than or equal to the second preset threshold.
Example four
An embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer program performs the steps of the fault detection method described in any of the above embodiments.
Specifically, the storage medium can be a general-purpose storage medium, such as a removable disk, a hard disk, or the like, and when a computer program on the storage medium is executed, the steps of the fault detection method can be executed, so that the working state of the printer can be monitored in real time, and the working efficiency can be improved.
The computer program product for performing the fault detection method provided in the embodiment of the present application includes a computer-readable storage medium storing a program code, where instructions included in the program code may be used to execute the method described in the foregoing method embodiment, and specific implementation may refer to the method embodiment, which is not described herein again.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the system and the apparatus described above may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used for illustrating the technical solutions of the present application, but not limiting the same, and the scope of the present application is not limited thereto, and although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the exemplary embodiments of the present application, and are intended to be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (8)
1. A method of fault detection, comprising:
acquiring sensing parameters detected by a sensor arranged on a printer, wherein the sensing parameters are used for reflecting the current working state of the printer;
extracting the characteristics of the obtained induction parameters to obtain characteristic vectors;
inputting the obtained feature vector into a classifier model obtained by pre-training, and determining a state evaluation value, wherein the state evaluation value is used for reflecting the possibility of the printer failure;
when the state evaluation value is larger than a first preset threshold value, performing alarm processing;
the extracting the characteristics of the obtained induction parameters to obtain the characteristic vectors specifically comprises:
summing induction parameters respectively detected by the same type of sensors arranged at different positions of the printer at the same time to obtain first data corresponding to each type of sensor;
combining induction parameters respectively detected by all sensors arranged on a printer at the same time in pairs;
summing operation is respectively carried out on each group of induction parameters obtained after pairwise combination to obtain second data corresponding to each group of induction parameters;
performing subtraction operation on each group of induction parameters obtained after pairwise combination to obtain third data corresponding to each group of induction parameters;
generating the feature vector according to the first data, the second data, the third data and the sensing parameters detected by each sensor installed on a printer;
training the obtained classifier model according to the following modes:
acquiring a plurality of sample data with status labels, wherein the status labels comprise a first label indicating that the printer is in failure and a second label indicating that the printer is normal;
sequentially inputting the plurality of sample data with the state labels into a classifier model to obtain a training result of each sample data, wherein the training result is the state of the printer detected by the classifier model, and the state of the printer can be a normal state or a fault state;
and when the accuracy of the training result is smaller than a second preset threshold, adjusting the model parameters of the classifier model until the accuracy of the training result is larger than or equal to the second preset threshold.
2. The method of claim 1, wherein the sensed parameter comprises: a frequency detected by a vibration sensor; acceleration detected by an acceleration sensor.
3. The method according to claim 1, wherein after acquiring the sensing parameters detected by the sensor installed on the printer and before performing feature extraction on the acquired sensing parameters, the method further comprises:
calculating the average value of the same type of induction parameters respectively collected under each time window in continuous time, wherein the time window is the time length consumed by obtaining N same type of induction parameters, and N is a positive integer;
calculating the standard deviation between the average values respectively corresponding to different time windows;
normalizing each obtained induction parameter in the same induction parameter by using the calculated average value and standard deviation to obtain the induction parameter after normalization;
the characteristic extraction of the acquired induction parameters comprises the following steps:
and performing feature extraction on the induction parameters after the normalization processing.
4. A fault detection device, comprising:
the system comprises an acquisition module, a processing module and a control module, wherein the acquisition module is used for acquiring sensing parameters detected by a sensor arranged on a printer, and the sensing parameters are used for reflecting the current working state of the printer;
the characteristic extraction module is used for extracting the characteristics of the obtained induction parameters to obtain characteristic vectors;
the determining module is used for inputting the obtained feature vector into a classifier model obtained by pre-training and determining a state evaluation value, wherein the state evaluation value is used for reflecting the possibility of the printer failure;
the alarm module is used for carrying out alarm processing when the state evaluation value is greater than a first preset threshold value;
the feature extraction module is specifically configured to, when performing feature extraction on the acquired sensing parameters:
summing induction parameters respectively detected by the same type of sensors arranged at different positions of the printer at the same time to obtain first data corresponding to each type of sensor;
combining induction parameters respectively detected by all sensors arranged on a printer at the same time in pairs;
summing operation is respectively carried out on each group of induction parameters obtained after pairwise combination to obtain second data corresponding to each group of induction parameters;
performing subtraction operation on each group of induction parameters obtained after pairwise combination to obtain third data corresponding to each group of induction parameters;
generating the feature vector according to the first data, the second data, the third data and the sensing parameters detected by each sensor installed on a printer;
the determining module, when training the classifier model, is specifically configured to:
acquiring a plurality of sample data with status labels, wherein the status labels comprise a first label indicating that the printer is in failure and a second label indicating that the printer is normal;
sequentially inputting the plurality of sample data with the state labels into a classifier model to obtain a training result of each sample data, wherein the training result is the state of the printer detected by the classifier model, and the state of the printer can be a normal state or a fault state;
and when the accuracy of the training result is smaller than a second preset threshold, adjusting the model parameters of the classifier model until the accuracy of the training result is larger than or equal to the second preset threshold.
5. The apparatus of claim 4, wherein the sensed parameter comprises: a frequency detected by a vibration sensor; acceleration detected by an acceleration sensor.
6. The apparatus according to claim 4, wherein after acquiring the sensing parameters detected by the sensor mounted on the printer, before performing feature extraction on the acquired sensing parameters, the apparatus is further configured to:
calculating the average value of the same induction parameters collected in each time window in continuous time, wherein
The time window is the time length consumed for obtaining N same induction parameters, and N is a positive integer;
calculating the standard deviation between the average values respectively corresponding to different time windows;
using the calculated average value and standard deviation to obtain each induction parameter in the same induction parameter
Carrying out normalization processing to obtain induction parameters after normalization processing;
the characteristic extraction of the acquired induction parameters comprises the following steps:
and performing feature extraction on the induction parameters after the normalization processing.
7. An electronic device, comprising: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory communicating over the bus when the electronic device is operating, the machine-readable instructions when executed by the processor performing the steps of the fault detection method of any of claims 1 to 3.
8. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the fault detection method according to one of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811269612.5A CN109397703B (en) | 2018-10-29 | 2018-10-29 | Fault detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811269612.5A CN109397703B (en) | 2018-10-29 | 2018-10-29 | Fault detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109397703A CN109397703A (en) | 2019-03-01 |
| CN109397703B true CN109397703B (en) | 2020-08-07 |
Family
ID=65469708
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811269612.5A Active CN109397703B (en) | 2018-10-29 | 2018-10-29 | Fault detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109397703B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110004573B (en) * | 2019-04-03 | 2020-09-08 | 东北大学 | Yarn fault detection method and device based on vibration data |
| CN110142973A (en) * | 2019-05-10 | 2019-08-20 | 西安理工大学 | Fusion sediment formula 3D printer and method with condition monitoring and fault diagnosis |
| CN110370649B (en) * | 2019-07-11 | 2021-05-11 | 中国科学院自动化研究所 | Online monitoring device and system of 3D printing equipment |
| CN111120094B (en) * | 2019-11-29 | 2021-02-23 | 潍柴动力股份有限公司 | Engine fire detection method and device, storage medium and terminal |
| CN110906508B (en) * | 2019-12-09 | 2020-12-15 | 珠海格力电器股份有限公司 | Fault detection method and system for air conditioner sensor |
| CN111651505B (en) * | 2020-06-05 | 2023-05-16 | 中国民用航空厦门空中交通管理站 | Equipment operation situation analysis and early warning method and system based on data driving |
| CN113064559A (en) * | 2021-03-12 | 2021-07-02 | 珠海奔图电子有限公司 | Image forming apparatus monitoring method, apparatus, system and storage medium |
| CN113282433B (en) * | 2021-06-10 | 2023-04-28 | 天翼云科技有限公司 | Cluster anomaly detection method, device and related equipment |
| CN114311683B (en) * | 2021-12-31 | 2023-11-17 | 深圳拓竹科技有限公司 | Method for 3D printer and 3D printer |
| SE2250597A1 (en) * | 2022-05-19 | 2023-11-20 | Cellink Bioprinting Ab | Multi-sensor evaluation of a printing process |
| CN116484306B (en) * | 2023-06-20 | 2023-09-26 | 蘑菇物联技术（深圳）有限公司 | Positioning method and device of abnormal sensor, computer equipment and storage medium |
| CN117931103B (en) * | 2024-01-24 | 2024-08-09 | 浙江沧田智能信息科技有限公司 | Laser printer fault identification method and system based on remote interaction |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08286719A (en) * | 1995-04-19 | 1996-11-01 | Amada Washino Co Ltd | Input history storage device for working machine |
| US7631168B1 (en) * | 2006-05-10 | 2009-12-08 | The Math Works, Inc. | Graphical interface for grouping concurrent computing units executing a concurrent computing process |
| CN103325384A (en) * | 2012-03-23 | 2013-09-25 | 杜比实验室特许公司 | Harmonicity estimation, audio classification, pitch definition and noise estimation |
| CN105786711A (en) * | 2016-03-25 | 2016-07-20 | 广州华多网络科技有限公司 | Data analysis method and device |
| CN108491305B (en) * | 2018-03-09 | 2021-05-25 | 网宿科技股份有限公司 | Method and system for detecting server fault |
-
2018
- 2018-10-29 CN CN201811269612.5A patent/CN109397703B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109397703A (en) | 2019-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109397703B (en) | Fault detection method and device | |
| EP3441947B1 (en) | System and method for remaining useful life determination | |
| US11544554B2 (en) | Additional learning method for deterioration diagnosis system | |
| Wang et al. | Online class imbalance learning and its applications in fault detection | |
| US20150346066A1 (en) | Asset Condition Monitoring | |
| CN108181105B (en) | Rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence | |
| CN113485302B (en) | Vehicle operation process fault diagnosis method and system based on multivariate time sequence data | |
| CN109740617A (en) | A kind of image detecting method and device | |
| KR101539896B1 (en) | Method for diagnosis of induction motor fault | |
| CN110837718B (en) | Switch fault detection method and device, electronic equipment and storage medium | |
| CN111190804A (en) | Multi-level deep learning log fault detection method for cloud native system | |
| CN110134571A (en) | Rotary-type mechanical equipment health status monitoring method and device | |
| CN111678699B (en) | Early fault monitoring and diagnosing method and system for rolling bearing | |
| CN111964909A (en) | Rolling bearing operation state detection method, fault diagnosis method and system | |
| EP2135144B1 (en) | Machine condition monitoring using pattern rules | |
| CN112070180B (en) | Power grid equipment state judging method and device based on information physical bilateral data | |
| CN113424119A (en) | Work efficiency evaluation method, work efficiency evaluation device, and program | |
| CN114548280A (en) | Fault diagnosis model training method, fault diagnosis method and electronic equipment | |
| JP7188463B2 (en) | ANALYSIS DEVICE, ANALYSIS METHOD, AND PROGRAM | |
| JP6323121B2 (en) | Unknown data analyzer | |
| JP2020107248A (en) | Abnormality determination device and abnormality determination method | |
| Zhang et al. | Applied sensor fault detection and validation using transposed input data PCA and ANNs | |
| CN114391093B (en) | Abnormality determination device and abnormality determination method | |
| CN115238904A (en) | Method, apparatus and computer program for creating training data in a vehicle | |
| CN114355234A (en) | Intelligent quality detection method and system for power module |
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 |