CN117218405A - Digital fault diagnosis method for battery production equipment - Google Patents
Digital fault diagnosis method for battery production equipment Download PDFInfo
- Publication number
- CN117218405A CN117218405A CN202310917399.9A CN202310917399A CN117218405A CN 117218405 A CN117218405 A CN 117218405A CN 202310917399 A CN202310917399 A CN 202310917399A CN 117218405 A CN117218405 A CN 117218405A
- Authority
- CN
- China
- Prior art keywords
- data
- generator
- training
- binarization
- fault
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 55
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000003745 diagnosis Methods 0.000 title claims abstract description 46
- 238000012549 training Methods 0.000 claims abstract description 56
- 238000012545 processing Methods 0.000 claims abstract description 16
- 238000013135 deep learning Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 35
- 230000008569 process Effects 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 9
- 238000005315 distribution function Methods 0.000 claims description 6
- 230000007246 mechanism Effects 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 4
- 238000002405 diagnostic procedure Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000013145 classification model Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000008485 antagonism Effects 0.000 claims description 2
- 230000001172 regenerating effect Effects 0.000 claims description 2
- 230000006872 improvement Effects 0.000 claims 1
- 230000007257 malfunction Effects 0.000 claims 1
- 230000008929 regeneration Effects 0.000 description 4
- 238000011069 regeneration method Methods 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000006227 byproduct Substances 0.000 description 1
- 239000003054 catalyst Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Abstract
The application discloses a digital fault diagnosis method for battery production equipment, and relates to the technical field of equipment fault diagnosis. The application comprises the following steps: acquiring a normal state and fault state sample set in the battery production process; performing binarization operation on the normal state data and the fault state data; inputting the binarized normal state data set and the binarized fault state data set into a generated countermeasure network; training a fault diagnosis model of battery production equipment by deep learning sample characteristics; acquiring production equipment operation data in real time, and importing the production equipment operation data into a generated type countermeasure network; after the generated countermeasure network outputs the characteristic image, a device fault diagnosis model is input to identify and diagnose the device. According to the application, the normal state data and the fault state data are subjected to binarization operation, and the binarization processing data are utilized, so that the workload of the fault diagnosis model of the battery production equipment is directly lightened, and the processing efficiency and the accuracy of the fault diagnosis model of the battery production equipment are improved.
Description
Technical Field
The application belongs to the technical field of fault diagnosis, and particularly relates to a digital fault diagnosis method for battery production equipment.
Background
The automation level is higher in the power battery production process, the integration level and the complexity are improved increasingly, and the association coupling between different process variables is realized, so that any tiny problems such as manual misoperation, equipment part abnormality and the like can cause chain reaction, the operation failure of the whole system is caused, and huge economic loss is caused. The fault detection and diagnosis technology is one of key technologies for guaranteeing industrial production safety and reducing maintenance cost, and is mainly divided into a qualitative method and a quantitative method:
the qualitative method analyzes and diagnoses the faults based on the experience knowledge accumulated by the field expert or the professional mechanic, and has the advantages of clear diagnosis mechanism, strong interpretability, but lower fault recognition rate and classification rate; the quantification method further includes a model and data based method. The method comprises the steps of constructing a mechanism model of an industrial process based on a model method, obtaining estimated values of intermediate variables and output variables in a normal state according to system input simulation, and comparing measured data to identify faults and anomalies; the data-based method does not depend on the system structure and parameters, and only judges whether faults exist by constructing a data model.
Compared with the prior art, due to the complexity of a modern industrial system, the model-based method is difficult to accurately describe the dynamic characteristics of the process variables, the data-based method bypasses the process mechanism and is based on machine learning models such as a support vector machine, a Bayesian network, a random forest, a neural network and the like, and fault diagnosis is realized according to nonlinear association relations between input and output fitting system variables, so that the method has wide application prospects.
However, the common diagnostic model based on data has a common problem, namely, the interpretation problem of complex models. Taking a neural network as an example, the feature transformation performed internally is a black box which is difficult for a user to understand, and the fault cause and effect variables cannot be obtained from the black box, so that the stability of the diagnosis performance is difficult to guarantee. The industrial system fault diagnosis method must have high reliability, which requires that the data-based method can effectively find the physical quantity change causing the fault, thereby revealing the cause of the fault, realizing fault diagnosis analysis and improving the reliability of the diagnosis method.
Based on this, there is a need in the art for new battery production equipment failure digital diagnostic methods to address the above-described problems.
Disclosure of Invention
The application aims to provide a digital fault diagnosis method for battery production equipment, which improves the efficiency of model processing data by performing binarization processing on acquired production equipment operation data, realizes equipment identification and diagnosis by using a generated type countermeasure network processed input equipment fault diagnosis model, and solves the problems of untimely fault discovery and low reliability in the existing battery production process.
In order to solve the technical problems, the application is realized by the following technical scheme:
the application relates to a digital diagnosis method for faults of battery production equipment, which comprises the following steps:
step S1: acquiring a normal state and fault state sample set in the battery production process;
step S2: performing binarization operation on the normal state data and the fault state data;
step S3: inputting the binarized normal state data set and the binarized fault state data set into a generated countermeasure network;
step S4: training a fault diagnosis model of battery production equipment by deep learning sample characteristics;
step S5: acquiring production equipment operation data in real time, and importing the production equipment operation data into a generated type countermeasure network;
step S6: after the characteristic image is output by the generated countermeasure network, the fault diagnosis model of the equipment is input;
step S7: the equipment fault diagnosis model outputs characteristic distribution of abnormal data to identify and diagnose equipment.
As a preferable embodiment, in the step S1, the normal state and the fault state samples include physical quantities capable of reflecting the battery production process, the physical quantities including process variables and control variables; and acquiring physical quantity actual measurement values of a period of continuous time under normal state and fault state by a data acquisition system to form a normal state and fault state sample set.
As a preferred embodiment, in the step S2, the binarizing operation includes:
step S21: carrying out mixed sequencing on the physical quantities of the normal state and the fault state, and searching the average value of all adjacent physical quantities of the normal state and the fault state as a binarization reference point;
step S22: for each physical quantity, constructing a Gaussian distribution function according to the mean value and the variance of a normal state sample, calculating the quantity r of the fault state sample within the range of 3 times of standard deviation of the normal state sample Gaussian distribution function, and setting an upper limit T and a lower limit T of a quantity threshold value t 、T b To obtain a binarization point;
step S23: for each physical quantity, it is mapped into a binarization feature according to the relative magnitudes of its value and all binarization point values, all of which constitute a binarization vector.
As a preferred technical solution, in the step S22, the obtaining of the binarized point includes three cases: when r > T t When the method is used, the maximum value and the minimum value of the corresponding physical quantity binarization reference points are used as binarization points; when T is b ≤r≤T t When the method is used, the statistical value of the binarization reference point of the corresponding physical quantity is used as a binarization point; when T is b When the method is used, all binarization reference points of the corresponding physical quantity are taken as binarization points.
As an preferable technical scheme, in the step S3, the normal state data and the fault state data after the binarization processing are divided into a training set and a testing set; the data of the normal state after the binarization processing is divided into a training set and a testing set; the data of the fault state after the binarization processing are all divided into test sets.
As a preferable technical solution, in the step S3, the generating type countermeasure network includes a data converter, a generator, and a discriminator; the training set of the time sequence data set is sequentially input into a data converter, a generator and a discriminator; the data converter converts the binarized data into two-dimensional image data; the generator includes a first encoder, a second encoder, and a decoder; the first encoder is used for learning the representation of the original characteristics of the sample; the second encoder is configured to generate a regenerative signature; the decoder is used for reconstructing a reproduction characteristic; the discriminator is used to distinguish whether an input sample is a false sample generated by the generator or a true sample from the training data.
As a preferable technical scheme, the encoder and decoder of the generator adopt jump connection between layers to obtain a U-Net type picture generator; the plurality of picture generators and the plurality of discriminators are in one-to-one correspondence to obtain a group of generation type antagonism network groups; the image generators are all input by taking an input mode as a condition, a required PET image is generated as a learning target, and the discriminators are all input by taking the input mode of the corresponding image generator, two-dimensional image data corresponding to the input mode and an output result, so that a generated type countermeasure network group is obtained.
As a preferred technical solution, the arbiter includes a plurality of sets of convolution and learkyrlu activation functions; the discriminator takes the input mode of the corresponding picture generator, the two-dimensional image data corresponding to the input mode and the output result as input to acquire the corresponding label picture; carrying out parameter estimation on the generator by adopting a loss function, and carrying out parameter estimation on the discriminator by adopting a cross entropy loss function to obtain a loss function between an output result of the picture generator and a label picture and a loss function of the discriminator; combining the loss functions of each generated type countermeasure network to obtain a combined loss function;
the generator formula is:
wherein G is the number of generators, N is the number of generators, X is the generated data, Z is the noise figureAmount of G i Is the ith generator;
the generators share input data and a discrimination network, and a mixed structure of the generators provides learning signals for the discriminators; the generated countermeasure network includes: the number of generators, the classification model mechanism of the generated countermeasure network, the training times, the starting training times, the adjusting parameters and the batch size.
As a preferable technical solution, the generator needs to calculate a lifting value of the generating capability in the previous training, and the specific process is as follows:
wherein D is t-1 Is a discriminator after the t-1 th training, D t-2 Is a discriminator after the t-2 th training, G i,t-1 Is the ith generator after the t-1 th training, G i,t-2 Is the ith generator after the t-2 th training, z t-1 Is the noise sample sampled in the t-1 training, z t-2 Is the noise sample sampled in the t-2 th training,is when z t-1 ～p z The desired value of the time period is set,is when z t-2 ～p z Expected value at that time.
As a preferred technical solution, the loss function of the generator is:
wherein z is t Is the noise sample sampled in the t-th training, G i,t-1 Is the ith generator after the t-1 th training, D t Is a discriminant after the t-th training,is the loss of the generatorA function.
The application has the following beneficial effects:
(1) According to the application, the normal state data and the fault state data are subjected to binarization operation, and the binarization processing data are utilized, so that the workload of a fault diagnosis model of the battery production equipment is directly lightened, and the processing efficiency and the accuracy of the fault diagnosis model of the battery production equipment are improved;
(2) According to the application, the equipment fault diagnosis model is manufactured through the production equipment operation data, the equipment data to be predicted is processed by using the generated countermeasure network, and the equipment data is converted into the two-dimensional image data input equipment fault diagnosis model to realize equipment identification and diagnosis, so that the equipment fault diagnosis accuracy and prediction efficiency are improved.
Of course, it is not necessary for any one product to practice the application to achieve all of the advantages set forth above at the same time.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for digitally diagnosing faults in a battery production facility according to the present application;
fig. 2 is a schematic structural view of a fault digital diagnostic device for battery production equipment according to the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
In addition, the technical features of the embodiments of the present application described below may be combined with each other as long as they do not collide with each other.
Example 1
Referring to fig. 1, the present application is a method for preprocessing fault data of a battery production device, comprising:
step S1: acquiring a normal state and fault state sample set in the battery production process; the normal state and fault state samples contain physical quantities capable of reflecting the battery production process, wherein the physical quantities comprise process variables and control variables; the data acquisition system is used for acquiring physical quantity actual measurement values of a period of continuous time under normal state and fault state to form a normal state and fault state sample set, and Tennessee Eastman (TE) process is taken as an example to acquire the normal state and fault state sample set. The TE process consists of 5 units and 8 species, specifically 4 inputs A, C, D and E,2 outputs G and H, catalyst B and byproduct F. The embodiment selects 22 process variables PV (1) -PV (22) and 11 operating variables OV (1) -OV (11) as main physical quantities reflecting the state of the industrial process, and selects 100 normal state samples and 80 fault state samples of 3 types of different faults F (2), F (7) and F (10) to form a sample set.
Step S2: performing binarization operation on the normal state data and the fault state data;
the binarization operation includes:
step S21: carrying out mixed sequencing on the physical quantities of the normal state and the fault state, and searching the average value of all adjacent physical quantities of the normal state and the fault state as a binarization reference point;
step S22: for each physical quantity, constructing a Gaussian distribution function according to the mean value and the variance of a normal state sample, calculating the quantity r of the fault state sample within the range of 3 times of standard deviation of the normal state sample Gaussian distribution function, and setting an upper limit T and a lower limit T of a quantity threshold value t 、T b To obtain a binarization point;
step S23: for each physical quantity, it is mapped into a binarization feature according to the relative magnitudes of its value and all binarization point values, all of which constitute a binarization vector.
In step S22, the acquisition of the binarized point includes three cases: when r > T t In the time-course of which the first and second contact surfaces,taking the maximum value and the minimum value of the corresponding physical quantity binarization reference points as binarization points; when T is b ≤r≤T t When the method is used, the statistical value of the binarization reference point of the corresponding physical quantity is used as a binarization point; when T is b When the method is used, all binarization reference points of the corresponding physical quantity are taken as binarization points.
Step S3: inputting the binarized normal state data set and the binarized fault state data set into a generated countermeasure network; dividing the binarized normal state data and fault state data into a training set and a testing set; the data of the normal state after the binarization processing is divided into a training set and a testing set; the data of the fault state after the binarization processing are all divided into test sets.
The innovation point of one aspect of the embodiment of the application is that the normal state data and the fault state data are subjected to binarization operation, and the binarization processing data are utilized, so that the workload of the fault diagnosis model of the battery production equipment is directly lightened, and the processing efficiency and the accuracy of the fault diagnosis model of the battery production equipment are improved.
Example two
The application relates to a digital diagnosis method for faults of battery production equipment, which comprises the following steps:
step S4: training a fault diagnosis model of battery production equipment by deep learning sample characteristics;
step S5: acquiring production equipment operation data in real time, and importing the production equipment operation data into a generated type countermeasure network;
step S6: after the characteristic image is output by the generated countermeasure network, the fault diagnosis model of the equipment is input;
step S7: the equipment fault diagnosis model outputs characteristic distribution of abnormal data to identify and diagnose equipment.
The generating type countermeasure network comprises a data converter, a generator and a discriminator; the training set of the time sequence data set is sequentially input into a data converter, a generator and a discriminator; the data converter converts the binarized data into two-dimensional image data; the generator comprises a first encoder, a second encoder and a decoder; the first encoder is used for learning the representation of the original characteristics of the sample; the second encoder is used for generating a regeneration characteristic; the decoder is used for reconstructing the regeneration characteristics; the discriminator is used to distinguish whether the input samples are false samples generated by the generator or real samples from the training data.
The encoder and decoder of the generator adopt jump connection between each layer to obtain a U-Net type picture generator; the plurality of picture generators and the plurality of discriminators are in one-to-one correspondence to obtain a group of generation type countermeasure network groups; the image generation method comprises the steps that a plurality of image generators are input by taking an input mode as a condition, a required PET image is generated as a learning target, and a plurality of discriminators are input by taking the input mode corresponding to the image generators, two-dimensional image data corresponding to the input mode and an output result, so that a generation type countermeasure network group is obtained.
The discriminator comprises a plurality of sets of convolution sum and LearkyReLU activation functions; the discriminator takes the input mode of the corresponding picture generator, the two-dimensional image data corresponding to the input mode and the output result as input to acquire the corresponding label picture; carrying out parameter estimation on the generator by adopting a loss function, and carrying out parameter estimation on the discriminator by adopting a cross entropy loss function to obtain a loss function between an output result of the picture generator and a label picture and a loss function of the discriminator; combining the loss functions of each generated type countermeasure network to obtain a combined loss function;
the generator formula is:
where G is the generator, N is the number of generators, X is the generated data, Z is the noise scalar, G i Is the ith generator;
input data and a discrimination network are shared among the generators, and a mixed structure of the generators provides learning signals for the discriminators; the generation type countermeasure network comprises: the number of generators, the classification model mechanism of the generated countermeasure network, the training times, the starting training times, the adjusting parameters and the batch size.
The generator needs to calculate the lifting value of the generating capacity in the previous training, and the specific process is as follows:
wherein D is t-1 Is a discriminator after the t-1 th training, D t-2 Is a discriminator after the t-2 th training, G i,t-1 Is the ith generator after the t-1 th training, G i,t-2 Is the ith generator after the t-2 th training, z t-1 Is the noise sample sampled in the t-1 training, z t-2 Is the noise sample sampled in the t-2 th training,is when z t-1 ～p z The desired value of the time period is set,is when z t-2 ～p z Expected value at that time.
The loss function of the generator is:
wherein z is t Is the noise sample sampled in the t-th training, G i,t-1 Is the ith generator after the t-1 th training, D t Is a discriminant after the t-th training,is the loss function of the generator.
The application relates to a file generation type countermeasure network, which comprises a data converter, a generator and a discriminator; the training set of the time sequence data set is sequentially input into a data converter, a generator and a discriminator; the data converter converts the one-dimensional time series data into two-dimensional image data; the generator comprises a first encoder, a second encoder and a decoder; the first encoder is used for learning the representation of the original characteristics of the sample; the second encoder is used for generating a regeneration characteristic; the decoder is used for reconstructing the regeneration characteristics; the discriminator is used for distinguishing whether the input sample is a false sample generated by the generator or a real sample from the training data; the generator and the discriminator are trained iteratively, the more the training times are, the more the data generated by the generator are 'vivid', and the more the discriminator has strong false judging capability. It is obvious that the generator is constantly learning new "false making" capability and the arbiter is constantly improving "false making" capability.
The encoder and decoder of the generator adopt jump connection between each layer to obtain a U-Net type picture generator; the plurality of picture generators and the plurality of discriminators are in one-to-one correspondence to obtain a group of generation type countermeasure network groups; the image generation method comprises the steps that a plurality of image generators are input by taking an input mode as a condition, a required PET image is generated as a learning target, and a plurality of discriminators are input by taking the input mode corresponding to the image generators, two-dimensional image data corresponding to the input mode and an output result, so that a generation type countermeasure network group is obtained.
The discriminator comprises a plurality of sets of convolution sum and LearkyReLU activation functions; the discriminator takes the input mode of the corresponding picture generator, the two-dimensional image data corresponding to the input mode and the output result as input to acquire the corresponding label picture; carrying out parameter estimation on the generator by adopting a loss function, and carrying out parameter estimation on the discriminator by adopting a cross entropy loss function to obtain a loss function between an output result of the picture generator and a label picture and a loss function of the discriminator; and combining the loss functions of the various generated countermeasure networks to obtain a joint loss function.
The discriminator judges the probability that the sample is a true sample. The goal of the generator is to have the generated sample not be judged by the arbiter as being generated, while the goal of the arbiter is to be able to judge whether the sample is generated or real. When the countermeasure training of the generator and the discriminator reaches a steady state, the generator can generate a very real sample from random noise, thereby achieving the purpose of sample generation.
Example III
As shown in fig. 2, a battery production facility failure digital diagnosis apparatus of the present embodiment includes one or more processors and a memory. One processor is taken as an example in fig. 2.
The processor and the memory may be connected by a bus or otherwise, for example in fig. 2.
The memory is used as a nonvolatile computer-readable storage medium for storing a nonvolatile software program and a nonvolatile computer-executable program, as in example 1, a battery production apparatus failure digital diagnosis method. The processor executes a battery production facility fault digital diagnostic method by running non-volatile software programs and instructions stored in the memory.
The memory may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory may optionally include memory located remotely from the processor, the remote memory being connectable to the processor through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The program instructions/modules are stored in the memory and, when executed by the one or more processors, perform a battery production equipment failure digital diagnostic method of embodiment 1 described above, for example, performing the steps of fig. 1 described above.
It should be noted that, in the above system embodiment, each unit included is only divided according to the functional logic, but not limited to the above division, so long as the corresponding function can be implemented; in addition, the specific names of the functional units are also only for distinguishing from each other, and are not used to limit the protection scope of the present application.
In addition, those skilled in the art will appreciate that all or part of the steps in implementing the methods of the embodiments described above may be implemented by a program to instruct related hardware, and the corresponding program may be stored in a computer readable storage medium.
The preferred embodiments of the application disclosed above are intended only to assist in the explanation of the application. The preferred embodiments are not exhaustive or to limit the application to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the application and the practical application, to thereby enable others skilled in the art to best understand and utilize the application. The application is limited only by the claims and the full scope and equivalents thereof.
Claims (10)
1. The digital fault diagnosis method for the battery production equipment is characterized by comprising the following steps of:
step S1: acquiring a normal state and fault state sample set in the battery production process;
step S2: performing binarization operation on the normal state data and the fault state data;
step S3: inputting the binarized normal state data set and the binarized fault state data set into a generated countermeasure network;
step S4: training a fault diagnosis model of battery production equipment by deep learning sample characteristics;
step S5: acquiring production equipment operation data in real time, and importing the production equipment operation data into a generated type countermeasure network;
step S6: after the characteristic image is output by the generated countermeasure network, the fault diagnosis model of the equipment is input;
step S7: the equipment fault diagnosis model outputs characteristic distribution of abnormal data to identify and diagnose equipment.
2. The method according to claim 1, wherein in the step S1, the normal state and the failure state samples contain physical quantities capable of reflecting the battery production process, the physical quantities including process variables and control variables; and acquiring physical quantity actual measurement values of a period of continuous time under normal state and fault state by a data acquisition system to form a normal state and fault state sample set.
3. The method for digitally diagnosing a fault in a battery production facility according to claim 1, wherein in said step S2, the binarizing operation comprises:
step S21: carrying out mixed sequencing on the physical quantities of the normal state and the fault state, and searching the average value of all adjacent physical quantities of the normal state and the fault state as a binarization reference point;
step S22: for each physical quantity, constructing a Gaussian distribution function according to the mean value and the variance of a normal state sample, calculating the quantity r of the fault state sample within the range of 3 times of standard deviation of the normal state sample Gaussian distribution function, and setting an upper limit T and a lower limit T of a quantity threshold value t 、T b To obtain a binarization point;
step S23: for each physical quantity, it is mapped into a binarization feature according to the relative magnitudes of its value and all binarization point values, all of which constitute a binarization vector.
4. A method for digitally diagnosing a fault in a battery production facility according to claim 3, wherein in said step S22, the acquisition of the binarization point includes three cases: when r > T t When the method is used, the maximum value and the minimum value of the corresponding physical quantity binarization reference points are used as binarization points; when T is b ≤r≤T t When the method is used, the statistical value of the binarization reference point of the corresponding physical quantity is used as a binarization point; when T is b When the method is used, all binarization reference points of the corresponding physical quantity are taken as binarization points.
5. The method for digital diagnosis of a battery production facility according to claim 1, wherein in the step S3, the normal state data and the fault state data after the binarization processing are divided into a training set and a test set; the data of the normal state after the binarization processing is divided into a training set and a testing set; the data of the fault state after the binarization processing are all divided into test sets.
6. The method according to claim 1, wherein in the step S3, the generated countermeasure network includes a data converter, a generator, and a discriminator; the training set of the time sequence data set is sequentially input into a data converter, a generator and a discriminator; the data converter converts the binarized data into two-dimensional image data; the generator includes a first encoder, a second encoder, and a decoder; the first encoder is used for learning the representation of the original characteristics of the sample; the second encoder is configured to generate a regenerative signature; the decoder is used for reconstructing a reproduction characteristic; the discriminator is used to distinguish whether an input sample is a false sample generated by the generator or a true sample from the training data.
7. The method for digital diagnosis of a battery production facility according to claim 6, wherein the encoder and decoder of the generator are connected by skip connection to obtain a U-Net type picture generator; the plurality of picture generators and the plurality of discriminators are in one-to-one correspondence to obtain a group of generation type antagonism network groups; the image generators are all input by taking an input mode as a condition, a required PET image is generated as a learning target, and the discriminators are all input by taking the input mode of the corresponding image generator, two-dimensional image data corresponding to the input mode and an output result, so that a generated type countermeasure network group is obtained.
8. The method for digital diagnosis of a battery production facility according to claim 7, wherein the discriminator comprises a plurality of sets of convolution and learkyrlu activation functions; the discriminator takes the input mode of the corresponding picture generator, the two-dimensional image data corresponding to the input mode and the output result as input to acquire the corresponding label picture; carrying out parameter estimation on the generator by adopting a loss function, and carrying out parameter estimation on the discriminator by adopting a cross entropy loss function to obtain a loss function between an output result of the picture generator and a label picture and a loss function of the discriminator; combining the loss functions of each generated type countermeasure network to obtain a combined loss function;
the generator formula is:
where G is the generator, N is the number of generators, X is the generated data, Z is the noise scalar, G i Is the ith generator;
the generators share input data and a discrimination network, and a mixed structure of the generators provides learning signals for the discriminators; the generated countermeasure network includes: the number of generators, the classification model mechanism of the generated countermeasure network, the training times, the starting training times, the adjusting parameters and the batch size.
9. The method for digital diagnosis of a malfunction of a battery production apparatus according to claim 6, wherein the generator is required to calculate the improvement value of the production capacity in the previous training, comprising:
wherein D is t-1 Is a discriminator after the t-1 th training, D t-2 Is a discriminator after the t-2 th training, G i,t-1 Is the ith generator after the t-1 th training, G i,t-2 Is the ith generator after the t-2 th training, z t-1 Is the noise sample sampled in the t-1 training, z t-2 Is the noise sample sampled in the t-2 th training,is when z t-1 ～p z Expected value of time->Is when z t-2 ～p z Expected value at that time.
10. The battery production facility failure digital diagnostic method of claim 9, wherein the generator loss function is:
wherein z is t Is the noise sample sampled in the t-th training, G i,t-1 Is the ith generator after the t-1 th training, D t Is a discriminant after the t-th training,is the loss function of the generator.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310917399.9A CN117218405A (en) | 2023-07-25 | 2023-07-25 | Digital fault diagnosis method for battery production equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310917399.9A CN117218405A (en) | 2023-07-25 | 2023-07-25 | Digital fault diagnosis method for battery production equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117218405A true CN117218405A (en) | 2023-12-12 |
Family
ID=89039621
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310917399.9A Pending CN117218405A (en) | 2023-07-25 | 2023-07-25 | Digital fault diagnosis method for battery production equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117218405A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117669388A (en) * | 2024-01-30 | 2024-03-08 | 武汉理工大学 | Fault sample generation method, device and computer medium |
-
2023
- 2023-07-25 CN CN202310917399.9A patent/CN117218405A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117669388A (en) * | 2024-01-30 | 2024-03-08 | 武汉理工大学 | Fault sample generation method, device and computer medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109086889B (en) | Terminal fault diagnosis method, device and system based on neural network | |
| US8868985B2 (en) | Supervised fault learning using rule-generated samples for machine condition monitoring | |
| CN112766342A (en) | Abnormity detection method for electrical equipment | |
| CN112416643A (en) | Unsupervised anomaly detection method and unsupervised anomaly detection device | |
| CN109472097B (en) | Fault diagnosis method for online monitoring equipment of power transmission line | |
| CN117218405A (en) | Digital fault diagnosis method for battery production equipment | |
| CN113111591B (en) | Automatic diagnosis method, device and equipment based on internal fault of modular power distribution terminal | |
| CN111695620A (en) | Method and system for detecting and correcting abnormal data of time sequence of power system | |
| CN115455746B (en) | Nuclear power device operation monitoring data anomaly detection and correction integrated method | |
| CN112559316A (en) | Software testing method and device, computer storage medium and server | |
| CN115238785A (en) | Rotary machine fault diagnosis method and system based on image fusion and integrated network | |
| CN117056865B (en) | Method and device for diagnosing operation faults of machine pump equipment based on feature fusion | |
| CN112380763A (en) | System and method for analyzing reliability of in-pile component based on data mining | |
| CN116108371A (en) | Cloud service abnormity diagnosis method and system based on cascade abnormity generation network | |
| CN114841196A (en) | Mechanical equipment intelligent fault detection method and system based on supervised learning | |
| CN115409052A (en) | Fault diagnosis method and system for wind generating set bearing under data imbalance | |
| KR102389317B1 (en) | Method of Determining Whether A Smart Farm Sensor has failed using a Recurrent Neural Network(RNN) | |
| CN113723592A (en) | Fault diagnosis method based on wind power gear box monitoring system | |
| CN117494588B (en) | Method, equipment and medium for optimizing residual effective life of fan bearing | |
| CN112598186A (en) | Improved LSTM-MLP-based small generator fault prediction method | |
| CN111310907A (en) | Microwave assembly fault diagnosis method, device and equipment | |
| CN115542880A (en) | Equipment fault prediction method based on generative countermeasure network | |
| Shi et al. | Data augmentation to improve the performance of ensemble learning for system failure prediction with limited observations | |
| CN117744874A (en) | Equipment fault prediction method and device and electronic equipment | |
| Li et al. | Fault diagnosis of automobile ECUs with data mining technologies |
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 |