CN113485863A - Method for generating heterogeneous unbalanced fault samples based on improved generation countermeasure network - Google Patents
Method for generating heterogeneous unbalanced fault samples based on improved generation countermeasure network Download PDFInfo
- Publication number
- CN113485863A CN113485863A CN202110796968.XA CN202110796968A CN113485863A CN 113485863 A CN113485863 A CN 113485863A CN 202110796968 A CN202110796968 A CN 202110796968A CN 113485863 A CN113485863 A CN 113485863A
- Authority
- CN
- China
- Prior art keywords
- fault
- data
- heterogeneous
- sample
- countermeasure network
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000012544 monitoring process Methods 0.000 claims abstract description 45
- 238000003745 diagnosis Methods 0.000 claims abstract description 39
- 238000012549 training Methods 0.000 claims abstract description 32
- 230000009977 dual effect Effects 0.000 claims abstract description 18
- 230000008569 process Effects 0.000 claims abstract description 16
- 230000006870 function Effects 0.000 claims description 32
- 239000013598 vector Substances 0.000 claims description 16
- 238000007781 pre-processing Methods 0.000 claims description 12
- 238000013528 artificial neural network Methods 0.000 claims description 11
- 230000006872 improvement Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 5
- 238000010586 diagram Methods 0.000 description 6
- 238000009826 distribution Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 5
- 238000005070 sampling Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000013256 Gubra-Amylin NASH model Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000002194 synthesizing effect Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000013079 data visualisation Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
- G06F11/079—Root cause analysis, i.e. error or fault diagnosis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
- G06F11/0706—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation the processing taking place on a specific hardware platform or in a specific software environment
- G06F11/0736—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation the processing taking place on a specific hardware platform or in a specific software environment in functional embedded systems, i.e. in a data processing system designed as a combination of hardware and software dedicated to performing a certain function
- G06F11/0739—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation the processing taking place on a specific hardware platform or in a specific software environment in functional embedded systems, i.e. in a data processing system designed as a combination of hardware and software dedicated to performing a certain function in a data processing system embedded in automotive or aircraft systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/22—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing
- G06F11/26—Functional testing
- G06F11/263—Generation of test inputs, e.g. test vectors, patterns or sequences ; with adaptation of the tested hardware for testability with external testers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Computer Hardware Design (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention provides a method for generating heterogeneous unbalanced fault samples based on an improved generation countermeasure network. The method provides a hybrid automatic encoder and a dual-discriminator generation countermeasure network MAE-D2GAN, aiming at the problem of serious unbalance of monitoring data, a discriminator obtained by normal state sample training is added into a countermeasure process, and an improved generation countermeasure network with the dual discriminators is established; designing an automatic encoder comprising an encoder and a decoder, encoding heterogeneous monitoring data into continuous potential features through the encoder, and inputting the continuous potential features into a dual-discriminator generation countermeasure network model; and the decoder output layer uses a differentiable Gumbel-softmax to process the dual discriminators corresponding to the discrete variables to generate continuous characteristics for resisting network generation, so as to obtain heterogeneous fault samples. The fault sample generated by the countermeasure network MAE-D2GAN generated by the hybrid automatic encoder and the double discriminators provided by the invention is closer to a real fault sample, so that the influence of the similar overlapping problem can be effectively reduced, and the prediction accuracy of the fault diagnosis model can be better improved.
Description
Technical Field
The application relates to the field of fault diagnosis, in particular to processing of category imbalance problems in fault diagnosis of complex systems or equipment, and data expansion of heterogeneous fault samples is achieved by adopting an improved generation countermeasure network.
Background
With the rapid development of information technology, the degree of system intellectualization and integration is gradually improved, and the system structure and function are more and more complex. The cross-effects of the components in a complex system, sometimes a small fault, can also cause a chain reaction, resulting in damage to the entire system. This not only causes huge economic losses, but also endangers the life safety of the personnel involved. Therefore, condition monitoring and fault diagnosis technology is becoming more and more important in system health and safety management as a predictive maintenance means. Currently, there are many methods for fault diagnosis, such as an expert system model, a physical model, a data-driven model, and the like. Among them, the data-driven method is widely used in fault diagnosis of complex systems. By adopting the data driving model, the abnormal inspection model can be obtained quickly and cheaply only by enough related monitoring data and maintenance data, and the dependence on the prior knowledge of the equipment can be avoided. Although the method needs mass data to obtain the high-precision model, the deficiency of the data volume can be overcome by optimizing an algorithm, simulating data and strengthening learning.
Most data-driven fault diagnosis methods assume that the data sets are evenly distributed, i.e. the number of samples of different classes is close. However, data in practical applications is often unbalanced, and especially for a complex system or device with high reliability, the failure samples are inevitably far less than the normal ones. When these data-driven classification algorithms are used directly for fault diagnosis, it is difficult to obtain satisfactory results. The prediction results tend to be biased toward most categories, so that the accuracy of fault diagnosis is very low. However, in practical applications, fault class data is significantly more important. Therefore, in the face of unbalanced data, the bias caused by it must be overcome.
Currently, there are three main types of methods for processing unbalanced data:
(1) the resampling method comprises the following steps: such as majority undersampling, minority oversampling, and composite sampling;
(2) algorithm level method: such as modifying a loss function, modifying a classification threshold, cost sensitive learning, etc.;
(3) integrated learning: such as iterative training using enhanced combinations.
Most of the methods can improve the classification precision to a certain extent and are verified and popularized in various fields. Among them, the Synthetic minor sampling Technique (SMOTE) is a common method, and the classification performance is improved by adjusting data distribution by adding and synthesizing a small number of samples. Recently proposed generation countermeasure networks (GAN) are also used for generation of synthetic samples due to their efficiency and flexibility. Unlike SMOTE and its variants, which rely mainly on expert knowledge to design the generation rules for synthesizing a small population, the GAN method can automatically learn its intrinsic distribution and generate a small number of samples similar to real samples. A GAN includes two variable networks: a generator and a discriminator, denoted G and D respectively, which are trained to game each other in GAN. The sample generated by the generator G is judged and evaluated by the discriminator D, and then the generator G is optimized according to the evaluation result, so that the efficiency and the quality of the sample generation process can be greatly improved. At present, GAN and its variants have been successfully applied to a plurality of fields such as image restoration, scene synthesis, face recognition, and the like, and as for unbalanced data in fault diagnosis, GAN-based research has been gradually developed in recent years.
Previous research has focused on continuous and high-sampling frequency data, and actual data often contains both numerical variables and classification variables at low sampling frequencies, so-called heterogeneous data. For example, the factors affecting the brake system of the high-speed train include numerical variables such as voltage and current, and classified variables such as an operation mode and a brake state.
In the existing GAN, a class-type variable, which is also called a discrete variable, is often encoded into a numerical value using a one-hot code, but this causes the generated sample to have no engineering explanation, and the value is no longer discrete and may exceed the value range of the original variable. In addition to the heterogeneity of data, previous work mostly assumed that there were enough fault samples available to train the GAN model, and for a complex system, fault data was limited. Training the resulting GAN model on a limited number of failure samples often leads to severe overfitting problems.
Disclosure of Invention
To overcome the disadvantages of the prior art, the present invention proposes a method for generating heterogeneous fault samples by combining an improved generation countermeasure network with a dual discriminator and an improved Auto Encoder (AE). In this method, discrete variables and continuous variables can be automatically identified and processed separately. The improved AE includes an encoder for extracting the valid latent variables and feeding them to the GAN with dual discriminators and a decoder for converting the generated continuous data into heterogeneous data.
To achieve the above object, an aspect of the present invention provides a method for generating heterogeneous imbalance fault samples based on an improved generation countermeasure network, which includes the following steps:
step 1: collecting fault states and monitoring data of electromechanical product systems or equipment to obtain historical monitoring data;
step 2: performing data preprocessing on the historical monitoring data obtained in the step 1 to obtain a preprocessed data set [ X, Y ], wherein X is a relevant variable of a fault state of an electromechanical product system or equipment, Y is a fault state of the electromechanical product system or the equipment, X is heterogeneous data, and X comprises a continuous variable and a discrete variable;
and step 3: setting an automatic encoder to realize the interconversion of heterogeneous data and continuous characteristics, adding a normal sample discriminator in the existing classical generation countermeasure network, establishing an improved generation countermeasure network with double discriminators, inputting the preprocessed data set [ X, Y ] obtained in the step 2 into the automatic encoder and the improved generation countermeasure network with double discriminators, and obtaining a heterogeneous fault sample U, wherein the method specifically comprises the following steps:
step 31: the automatic encoder comprises an encoder and a decoder; the encoder functions to encode the heterogeneous monitoring data into a continuum of latent features; the decoder output layer uses differentiable Gumbel-softmax to realize the discretization of continuous characteristics, the function of the decoder is to decode continuous fault characteristic data into heterogeneous fault samples, and input the preprocessed data set [ X, Y ] obtained in the step 2 into the encoder to obtain fault sample characteristic data X and normal sample characteristic data;
step 32: said step 3 improved generation countermeasure network with dual discriminators comprises one generator G and two discriminators D and F; training the discriminator F according to the normal sample characteristic data obtained in the step 31; inputting white noise z into the generator G with the double-discriminator improved generation countermeasure network to obtain a characteristic data set G (z) corresponding to the potential spatial fault sample;
step 33: training the generator G and the discriminator D according to the fault sample feature data x and a feature data set G (z) corresponding to the potential spatial fault sample, wherein the improved generation of the countermeasure network with the double discriminators in the training process has the following loss functions V (D, G):
in the formula: d (x) is the probability that the generated sample comes from the fault sample feature data x;
calculating expected values for all x in the process; d (G (z)) is the probability that discriminator D discriminates whether the set of continuous variables G (z) is true; f (G (z)) is the probability that the discriminator F discriminates whether the set of continuous variables G (z) is normal or not;
calculating expected values for all z set by the pass;
the objective function of the improved generation of the countermeasure network with dual discriminators is
Based on the objective function
Iteratively training the generator G and discriminator D, the iterative training including training the generator G to minimize the loss function V (D, G) and training the discriminator D to maximize the loss function V (D, G); when the difference between the maximum value and the minimum value of the loss function V (D, G) in the model is continuously and iteratively trained for 100 times is less than 0.01, stopping iterative training; obtaining a trained generator G after the iterative training is stopped; inputting a group of white noises z into the trained generator G to obtain a characteristic data set G (z) corresponding to a potential spatial fault sample;
step 34: inputting the feature data set G (z) corresponding to the potential spatial fault sample obtained in the step 33 into a decoder in the automatic encoder to obtain a heterogeneous fault sample U;
and 4, step 4: adding a column of fault state I to the heterogeneous fault sample U obtained in the step 34 to obtain a heterogeneous fault sample set [ U, I ], wherein the fault state I is a column vector with elements of 1, and the dimension of the fault state I is consistent with that of the heterogeneous fault sample U.
Further, the specific step of obtaining the heterogeneous fault sample U in step 34 is: inputting the feature data set g (z) corresponding to the latent space fault sample obtained in the step 33 into the decoder, obtaining a decoded data set, where the decoded data set includes a column corresponding to a discrete variable and a column corresponding to a continuous variable, discretizing the column corresponding to the discrete variable with a differentiable Gumbel-softmax, obtaining a discretized column, and obtaining a heterogeneous fault sample U according to the column corresponding to the continuous variable and the discretized column.
Further, the encoding process using the encoder in step 31 specifically includes: and (3) carrying out single-hot coding on discrete variables in the preprocessed data set [ X, Y ] obtained in the step (2) to obtain single-hot coded variables, and uniformly coding the single-hot coded variables and continuous variables in the preprocessed data set [ X, Y ] obtained in the step (2) to obtain fault sample characteristic data X and normal sample characteristic data.
Preferably, the fault state in step 1 includes a fault and no fault.
Preferably, the monitoring data in step 1 is heterogeneous data, and the heterogeneous data includes continuous features, discrete features and signal features.
Further, the data preprocessing in the step 2 comprises missing value filling, abnormal value replacement, dimension gap processing and digitization; the missing value filling adopts a local mean filling method; the process dimension gap is a data normalization that takes the z-score method.
The invention also provides a method for performing real-time fault diagnosis according to the method for generating the heterogeneous unbalanced fault sample based on the improved generation countermeasure network, which specifically comprises the following steps:
step 1: collecting fault states and monitoring data of electromechanical product systems or equipment to obtain historical monitoring data;
step 2: performing data preprocessing on the historical monitoring data obtained in the step 1 to obtain a preprocessed data set [ X, Y ], wherein X is a relevant variable of a fault state of an electromechanical product system or equipment, Y is a fault state of the electromechanical product system or the equipment, X is heterogeneous data, and X comprises a continuous variable and a discrete variable;
and step 3: setting an automatic encoder to realize the interconversion of heterogeneous data and continuous characteristics, adding a normal sample discriminator in the existing classical generation countermeasure network, establishing an improved generation countermeasure network with double discriminators, inputting the preprocessed data set [ X, Y ] obtained in the step 2 into the automatic encoder and the improved generation countermeasure network with double discriminators, and obtaining a heterogeneous fault sample U;
and 4, step 4: adding a column of fault state I to the heterogeneous fault sample U obtained in the step 3 to obtain a heterogeneous fault sample set [ U, I ], wherein the fault state I is a column vector with elements of 1, and the dimension of the fault state I is consistent with that of the heterogeneous fault sample U;
and 5: will carry out the stepsHeterogeneous fault sample set [ U, I ] obtained in step 4]Incorporating the preprocessed data set [ X, Y ] obtained in step 2]In (1), obtaining a new data set
Based on the new data set
Training an Artificial Neural Network (ANN) algorithm to obtain a fault diagnosis model;
step 6: acquiring real-time monitoring data, performing data preprocessing on the real-time monitoring data, which is the same as the data preprocessing in the step 2, obtaining the preprocessed real-time monitoring data, inputting the preprocessed real-time monitoring data to the fault diagnosis model obtained in the step 5, obtaining a real-time fault state, and completing fault diagnosis.
Compared with the prior art, the invention provides a method for generating heterogeneous imbalance fault samples based on an improved generation countermeasure network, which has the advantages that:
1. the AE provided by the invention can automatically identify and process continuous variables and discrete variables simultaneously, and the differentiable Gumbel-softmax is used for reducing the discrete variables, so that an optimal model is more easily obtained;
2. the GAN with double discriminators is used for data expansion, so that the influence of the class overlapping problem can be effectively weakened;
3. the hybrid automatic encoder and the dual discriminator established by the method generate a confrontation network (Mixed AE-2Discriminators GAN, MAE-D2GAN) for fault diagnosis, and have the advantages that: the generated heterogeneous fault sample is more real and effective, accords with engineering practice, and can better improve the precision of fault diagnosis;
4. meanwhile, the method can be used for fault diagnosis and other operation and maintenance scenes, and has a good application prospect.
Drawings
FIG. 1 is a block diagram of an improved generation countermeasure network with dual discriminators that generates heterogeneous fault samples according to an embodiment of the present invention;
FIG. 2 is a structural diagram of an AE in an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a fault diagnosis model based on an artificial neural network ANN algorithm in an embodiment of the present invention;
FIG. 4 is a flowchart of a 10-fold cross-validation experiment in accordance with an embodiment of the present invention;
FIG. 5 is a t-SNE two-dimensional projection diagram of fault condition monitoring data of a brake system of a high-speed rail according to an embodiment of the invention.
Detailed Description
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
The embodiment of the invention provides a width learning-based real-time fault diagnosis method for an electromechanical product system or equipment, which comprises the following specific steps of:
step 1: collecting fault states and related monitoring data of the electrical product system or equipment in a period of time as historical monitoring data;
generally, the collected system or equipment fault state monitoring data is heterogeneous data, including continuous features, discrete features and signal features.
Step 2: and (4) carrying out data preprocessing on the historical monitoring data, including missing values, abnormal values, dimension differences, digital processing and the like, so as to obtain a preprocessed data set [ X, Y ]. X is related variable of fault state of the electromechanical product system or equipment, and X is heterogeneous data and comprises continuous variable and discrete variable; y is the fault state of the electromechanical product system or equipment, extracts the key indexes of the signal characteristics and digitalizes the discrete characteristics, and then the monitoring data can be converted into numerical data. The method and the device conduct research based on the numerical data obtained after conversion, and take the numerical data containing continuous features and discrete features as a preprocessed data set.
If the fault characteristic variables of the discrete data of the original fault characteristic variable data set contain the fault characteristic variables with only one state, the diagnosis of whether the system is in fault or not from the data is worthless, so that the fault characteristic variables need to be deleted, and subsequent research is carried out according to the remaining discrete data fault characteristic variables and the continuous data fault characteristic variables.
The historical monitoring data needs to be cleaned up and converted, such as filling missing values, replacing abnormal values and processing dimension gaps. The filling of the missing value adopts a local mean filling method, namely, the current missing value is filled by the mean value of the first 5 rows and the last 5 rows of data adjacent to the current monitoring data; the process dimension difference, i.e., data normalization, is performed by the z-score method, i.e., subtracting the mean and dividing by the standard deviation.
Various faults may occur in the system or the equipment, and the invention only distinguishes fault and no fault, and does not distinguish fault types.
The classical generation countermeasure network GAN can well process continuous data, an optimal model is obtained through first-order differentiation, parameters are difficult to update for discrete variables, and building a GAN model for heterogeneous class unbalanced data can face more challenges, such as difficulty in obtaining the optimal model, unrealistic generated samples and the like. To this end, the present invention proposes a combination of improved GAN and AE with dual discriminators and a differentiable Gumbel-softmax function to generate high quality heterogeneous fault samples. The improved GAN with the double discriminators is formed by adding a discriminator F on the basis of a classical GAN, wherein the discriminator F is used for distinguishing whether a generated sample is normal or not, the discriminator F is obtained only based on the training of the feature data of the normal sample, only a loss value is provided when the improved generation countermeasure network with the double discriminators is trained, and the discriminator F does not perform iterative optimization.
And step 3: training an improvement with dual discriminators to generate dual discriminators D and F in a countermeasure network after encoding with AE based on a preprocessed data set [ X, Y ], wherein F is optimized based on only normal sample feature data, where the normal sample feature data is feature data obtained after encoding with AE of normal samples in the preprocessed data set [ X, Y ], and D and G are combined to perform iterative optimization based on the generated continuous data set and fault sample feature data. The fault sample characteristic data X is the characteristic data of the fault sample in the preprocessed data set [ X, Y ] after AE encoding. Inputting white noise z into a generator G with a modified generation countermeasure network with dual discriminators generates a continuous data set G (z) of potential feature space. Converting the generated continuous data set g (z) into a heterogeneous data set U using a decoder in the AE;
in general, d-dimensional vector h can be normalized by softmax function to obtain each classification in the vector, and the sampling result is element h1,h2,...,hi,...,hdThereby converting h to a representation at h1,h2,...,hi,...,hdDiscrete variables of the sample.
h＝[h1,h2,...,hi,...,hd] (1)
y＝softmax(h)＝[p1,p2,...,pi,...,pd] (2)
Wherein: y is a vector of the d-dimensional vector h normalized by the softmax function; p is a radical ofiIs the ith element in the normalized vector.
softmax tends to make the probability of the largest element in h significantly larger than the other elements, but the vector p representing the probability has practically no probabilistic significance, and what is output based on softmax during model training is not trivial.
The Gumbel-softmax function is proposed based on softmax transformation, and can enable parameter updating to tend to an optimal solution, so that a generated sample is more real and effective, and a vector h is normalized by the Gumbel-softmax function and is as follows:
wherein:y' is a vector of a d-dimensional vector h normalized by a Gumbel-softmax function; g ═ g1,g2,...,gi,...,gd]，gi(i ═ 1,2, …, d) independently of one another and obey Gumbel distribution(ii) a τ is a control parameter for softness in the Gumbel-softmax function, and when τ → 0, y' → y. When τ → ∞, y' approximately satisfies the uniform distribution.
The invention adopts an encoder and a decoder to form AE to carry out feature reconstruction on the collected historical monitoring data, namely, heterogeneous monitoring data is encoded into continuous variables and feature extraction is carried out, and the dimension of the monitoring data is reduced by eliminating redundant information. In order to simultaneously encode and decode continuous and discrete variables in an auto-encoder, the present invention employs a hybrid model as shown in fig. 2. And, in order to prevent information loss, discrete variables are converted into multidimensional vectors based on one-hot encoding, then to generate discrete data, the decoder output needs to be one-hot vector, which is typically implemented by softmax function in the output layer. The invention uses a differentiable Gumbel-softmax function instead of the classical softmax function.
In order to make the generated fault sample similar to the actual fault data and different from the actual normal data, the countermeasure for generating the countermeasure network in the invention adopts a two-dimensional design, i.e. one countermeasure judges whether the generated sample is real or not, and the other judges whether the generated sample is normal or not, as shown in fig. 1, an improved generation countermeasure network structure diagram for generating heterogeneous fault samples is provided, and an additional item is added in a classical loss function to prevent the generated data from aggravating the problem of class overlapping.
The improvement with dual discriminators generates a loss function against the network as:
wherein: x is fault sample characteristic data, z is white gaussian noise input into the generator G, G (z) is a generated continuous data set, namely a characteristic data set corresponding to a potential space fault sample, D (x) is the probability of generating the sample from the fault sample characteristic data x, D (G (z)) is the probability of identifying whether G (z) is true or not, and F (G (z)) is the probability of identifying whether G (z) is normal or not.
To take all of the x's over to the desired value,
all z's set through are taken to the desired value.
Establishing an objective function for an improved generation of a competing network with dual discriminators as
And training the generator G to minimize the trend of the loss function, training the discriminator D to maximize the trend of the loss function, and stopping training until the difference between the maximum value and the minimum value of the loss function V (D, G) in continuous 100 iterations is less than 0.01, wherein the obtained generator G can be used for generating a feature data set G (z) corresponding to the fault sample in the potential space.
And 4, step 4: and E, using AE decoding G (z) to obtain a heterogeneous fault feature set U, adding a column of fault labels to U to obtain a generated heterogeneous fault sample set [ U, I ], wherein I is a fault state, I is a column vector with elements of 1, and the dimension of the fault state I is consistent with the number of the fault samples U.
The method comprises the following steps of establishing a hybrid automatic encoder and a dual discriminator generation countermeasure network (MAE-D2 GAN) for fault diagnosis by the method of the step 1-4, and specifically comprising the following steps:
the heterogeneous fault sample set [ U, I ] generated in the step 4 is processed]Merging into the preprocessed data set [ X, Y]In (3), get the new data set
The artificial neural network ANN model is trained based on the new data set to obtain a fault diagnosis model, the problem of unbalanced category can be effectively solved through the method, and the fault diagnosis precision is improved.
The structure of the artificial neural network ANN model is shown in fig. 3. The number of nodes of the input layer is the same as the number of columns of X, only one node is arranged on the output layer in a two-classification fault diagnosis scene, the structural complexity is approximately set according to the number of layers and the number of nodes of the hidden layer according to the sample amount and the characteristic number, and then manual adjustment is carried out to obtain an approximate local optimal solution.
And (3) acquiring real-time monitoring data of the system or equipment related to the fault state, preprocessing the real-time monitoring data in the same way as the step (2) to obtain preprocessed real-time monitoring data, inputting the preprocessed real-time monitoring data into a fault diagnosis model to obtain the real-time fault state of the product system or equipment, and completing fault diagnosis.
In order to check the effectiveness of the proposed method, a comparative experiment is designed based on a monitoring data set of a brake system of a high-speed train in the year-old running process, and the method is verified to be more effective than a common oversampling method. The comparison method comprises the following steps: random oversampling, combined oversampling SMOTE, marginal oversampling SMOTE-borderline1, marginal oversampling SMOTE-borderline2, ADASYN, and classical GAN with unique heat coding added. Experimental flow as shown in fig. 4, in 10-fold cross validation, the actual training data of each fault diagnosis model is a data set composed of the original training data and the generated data.
The brake system is a device for ensuring the effective deceleration of the high-speed train and is one of the most important components of the high-speed train. Data relating to brake system health may be collected by positioning sensors, capturing status information. The monitoring data set contains 43 variables that may lead to failure of the brake system, including 21 continuous variables and 22 discrete variables, such as speed, voltage, current, temperature, operating mode, brake status, etc.
Data cleaning and conversion, such as filling missing values, data normalization, are required on the raw data collected by the sensor before a new failure sample is generated using a different method. Data normalization is to balance dimensional differences between numerical variables. Furthermore, some variables in the raw data need to be converted to numerical values, depending on the data requirements of the model. After data processing, 43 variables in the input data become numerical variables between [0,1 ].
In order to avoid the influence of random factors, the average precision and the running time of ten-fold cross validation are observed in experiments. And in order to obtain approximate local optimal solution, all models are trained and optimized through gradually increasing structural complexity and iteration times until generalization precision is basically kept stable or reduced. Under the same operating environment, the model outputs the average F1-score, the average G-mean and the average AUC of the fault classes in the test results through 10 times of cross validation.
FIG. 5 is a two-dimensional projection diagram of a processed data set based on t-distribution random neighborhood embedding t-SNE, wherein 1.0 represents a fault sample, and 0.0 represents a normal sample. the t-SNE algorithm is one of the most common and efficient methods for analyzing high-dimensional data visualizations. The method converts the similarity between data points into probability, and the high-dimensional data can be visualized by projecting the high-dimensional data to a two-dimensional or three-dimensional space. Since the faulty sample is completely covered when all data is used, only 10% of the normal sample is shown in the figure. By comparison, the 10% normal samples in the graph are substantially consistent with the t-SNE projection distribution of the entire data set.
Heterogeneous fault samples are generated based on the improved generation countermeasure network, and whether the fault diagnosis accuracy can be effectively improved after the generated samples are added into training data is mainly compared. It can also be seen from fig. 5 that the data is severely unbalanced, and the classification result of the fault diagnosis model is severely biased to a normal state, so that the prediction accuracy of the fault class is very low. Obviously, in an actual scene, only the prediction accuracy of the fault state is concerned, so that the overall generalization accuracy cannot be used as a comparison index, and the accuracy and the recall rate of the fault state should be used. However, it can be seen from fig. 5 that the two types of data have strong intersection, which makes it impossible to achieve a very high level of accuracy and recall rate of the fault state at the same time. Therefore, the comprehensive indexes F1-score, G-means and AUC are used as evaluation indexes of different oversampling methods in the present invention.
Under the same operating environment, a classical artificial neural network model is used as a fault diagnosis model, the influence of different oversampling modes on a fault diagnosis result is compared according to a comparison experiment process, and the experiment result is as follows:
TABLE 1 comparison of different Generation methods
| F1-score | G-means | AUC | |
| Untreated | 0.38095 | 0.81494 | 0.63299 |
| RandomOversampling | 0.58065 | 0.74922 | 0.79879 |
| SMOTE-regular | 0.54054 | 0.67361 | 0.83125 |
| SMOTE-borderline1 | 0.63636 | 0.69469 | 0.96407 |
| SMOTE-borderline2 | 0.52632 | 0.65881 | 0.83108 |
| ADASYN | 0.62222 | 0.68301 | 0.96389 |
| GAN | 0.84404 | 0.85449 | 0.99705 |
| MAE-D2GAN | 0.875 | 0.88192 | 0.99723 |
It can be seen that the category imbalance may seriously affect the result of the fault diagnosis, and in this case, the conventional oversampling method may improve the accuracy of the fault diagnosis. Among other things, the MAE-D2GAN proposed in the present invention, combined with an improved generation countermeasure network with dual discriminators and an improved auto-encoder, can further improve the fault diagnosis results, better than the conventional GAN.
Compared with a typical oversampling method, in a high-speed train braking system, the MAE-D2GAN can generate a fault sample with higher quality, so that the prediction accuracy of the data-driven fault diagnosis model is greatly improved. In addition, for other PHM analysis of high reliability systems, MAE-D2GAN also has potential value for fault prediction and regression scenarios.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements made to the technical solution of the present invention by those skilled in the art without departing from the spirit of the present invention shall fall within the protection scope defined by the claims of the present invention.
Claims (7)
1. A method for generating heterogeneous imbalance fault samples based on improved generation of countermeasure networks, comprising the steps of:
step 1: collecting fault states and monitoring data of electromechanical product systems or equipment to obtain historical monitoring data;
step 2: performing data preprocessing on the historical monitoring data obtained in the step 1 to obtain a preprocessed data set [ X, Y ], wherein X is a relevant variable of a fault state of an electromechanical product system or equipment, Y is a fault state of the electromechanical product system or the equipment, X is heterogeneous data, and X comprises a continuous variable and a discrete variable;
and step 3: setting an automatic encoder to realize the interconversion of heterogeneous data and continuous characteristics, adding a normal sample discriminator in the existing classical generation countermeasure network, establishing an improved generation countermeasure network with double discriminators, inputting the preprocessed data set [ X, Y ] obtained in the step 2 into the automatic encoder and the improved generation countermeasure network with double discriminators, and obtaining a heterogeneous fault sample U, wherein the method specifically comprises the following steps:
step 31: the automatic encoder comprises an encoder and a decoder; the encoder functions to encode the heterogeneous monitoring data into a continuum of latent features; the decoder output layer uses differentiable Gumbel-softmax to realize the discretization of continuous characteristics, the function of the decoder is to decode continuous fault characteristic data into heterogeneous fault samples, and input the preprocessed data set [ X, Y ] obtained in the step 2 into the encoder to obtain fault sample characteristic data X and normal sample characteristic data;
step 32: said step 3 improved generation countermeasure network with dual discriminators comprises one generator G and two discriminators D and F; training the discriminator F according to the normal sample characteristic data obtained in the step 31; inputting white noise z into the generator G with the double-discriminator improved generation countermeasure network to obtain a characteristic data set G (z) corresponding to the potential spatial fault sample;
step 33: training the generator G and the discriminator D according to the fault sample feature data x and a feature data set G (z) corresponding to the potential spatial fault sample, wherein the improved generation of the countermeasure network with the double discriminators in the training process has the following loss functions V (D, G):
in the formula: d (x) is the probability that the generated sample comes from the fault sample feature data x;
calculating expected values for all x in the process; d (G (z)) is the probability that discriminator D discriminates whether the set of continuous variables G (z) is true; f (G (z)) is the probability that the discriminator F discriminates whether the set of continuous variables G (z) is normal or not;
calculating expected values for all z set by the pass;
the objective function of the improved generation of the countermeasure network with dual discriminators is
Based on the objective function
Iteratively training the generator G and discriminator D, the iterative training including training the generator G to minimize the loss function V (D, G) and training the discriminator D to maximize the loss function V (D, G); when the difference between the maximum value and the minimum value of the loss function V (D, G) in the model is continuously and iteratively trained for 100 times is less than 0.01, stopping iterative training; obtaining a trained generator G after the iterative training is stopped; inputting a group of white noises z into the trained generator G to obtain a characteristic data set G (z) corresponding to a potential spatial fault sample;
step 34: inputting the feature data set G (z) corresponding to the potential spatial fault sample obtained in the step 33 into a decoder in the automatic encoder to obtain a heterogeneous fault sample U;
and 4, step 4: adding a column of fault state I to the heterogeneous fault sample U obtained in the step 34 to obtain a heterogeneous fault sample set [ U, I ], wherein the fault state I is a column vector with elements of 1, and the dimension of the fault state I is consistent with that of the heterogeneous fault sample U.
2. The method for generating heterogeneous unbalanced fault samples for the countermeasure network based on the improvement as claimed in claim 1, wherein the step 34 of obtaining the heterogeneous fault samples U comprises the following specific steps: inputting the feature data set g (z) corresponding to the latent space fault sample obtained in the step 33 into the decoder, obtaining a decoded data set, where the decoded data set includes a column corresponding to a discrete variable and a column corresponding to a continuous variable, discretizing the column corresponding to the discrete variable with a differentiable Gumbel-softmax, obtaining a discretized column, and obtaining a heterogeneous fault sample U according to the column corresponding to the continuous variable and the discretized column.
3. The method for generating heterogeneous imbalance fault samples for countermeasure networks based on improvement according to claim 1, wherein the encoding process using the encoder in step 31 is specifically: and (3) carrying out single-hot coding on discrete variables in the preprocessed data set [ X, Y ] obtained in the step (2) to obtain single-hot coded variables, and uniformly coding the single-hot coded variables and continuous variables in the preprocessed data set [ X, Y ] obtained in the step (2) to obtain fault sample characteristic data X and normal sample characteristic data.
4. The method for generating heterogeneous unbalanced fault samples for a countermeasure network based on improvement as recited in claim 1, wherein the fault status in step 1 comprises faulty and non-faulty.
5. The method for generating heterogeneous imbalance fault samples for a countermeasure network based on improvement according to claim 1, wherein the monitoring data in the step 1 is heterogeneous data, and the heterogeneous data comprises a continuous characteristic, a discrete characteristic and a signal characteristic.
6. The method for generating heterogeneous imbalance fault samples for a countermeasure network based on improvement according to claim 1, wherein the data preprocessing in the step 2 includes filling missing values, replacing outliers, processing dimension gaps and digitizing; the missing value filling adopts a local mean filling method; the process dimension gap is a data normalization that takes the z-score method.
7. A method for generating heterogeneous unbalanced fault samples for real-time fault diagnosis based on an improved generation countermeasure network is characterized by specifically comprising the following steps:
step 1: collecting fault states and monitoring data of electromechanical product systems or equipment to obtain historical monitoring data;
step 2: performing data preprocessing on the historical monitoring data obtained in the step 1 to obtain a preprocessed data set [ X, Y ], wherein X is a relevant variable of a fault state of an electromechanical product system or equipment, Y is a fault state of the electromechanical product system or the equipment, X is heterogeneous data, and X comprises a continuous variable and a discrete variable;
and step 3: setting an automatic encoder to realize the interconversion of heterogeneous data and continuous characteristics, adding a normal sample discriminator in the existing classical generation countermeasure network, establishing an improved generation countermeasure network with double discriminators, inputting the preprocessed data set [ X, Y ] obtained in the step 2 into the automatic encoder and the improved generation countermeasure network with double discriminators, and obtaining a heterogeneous fault sample U;
and 4, step 4: adding a column of fault state I to the heterogeneous fault sample U obtained in the step 3 to obtain a heterogeneous fault sample set [ U, I ], wherein the fault state I is a column vector with elements of 1, and the dimension of the fault state I is consistent with that of the heterogeneous fault sample U;
and 5: the heterogeneous fault sample set [ U, I ] obtained in the step 4 is used]Incorporating the preprocessed data set [ X, Y ] obtained in step 2]In (1), obtaining a new data set
Based on the new data set
Training an Artificial Neural Network (ANN) algorithm to obtain a fault diagnosis model;
step 6: acquiring real-time monitoring data, performing data preprocessing on the real-time monitoring data, which is the same as the data preprocessing in the step 2, obtaining the preprocessed real-time monitoring data, inputting the preprocessed real-time monitoring data to the fault diagnosis model obtained in the step 5, obtaining a real-time fault state, and completing fault diagnosis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110796968.XA CN113485863B (en) | 2021-07-14 | 2021-07-14 | Method for generating heterogeneous imbalance fault samples based on improved generation of countermeasure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110796968.XA CN113485863B (en) | 2021-07-14 | 2021-07-14 | Method for generating heterogeneous imbalance fault samples based on improved generation of countermeasure network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113485863A true CN113485863A (en) | 2021-10-08 |
| CN113485863B CN113485863B (en) | 2023-05-05 |
Family
ID=77938762
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110796968.XA Active CN113485863B (en) | 2021-07-14 | 2021-07-14 | Method for generating heterogeneous imbalance fault samples based on improved generation of countermeasure network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113485863B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116821697A (en) * | 2023-08-30 | 2023-09-29 | 聊城莱柯智能机器人有限公司 | Mechanical equipment fault diagnosis method based on small sample learning |
| CN117854716A (en) * | 2024-03-08 | 2024-04-09 | 长春师凯科技产业有限责任公司 | Method and system for filling heart disease diagnosis missing data based on HF-GAN |
| CN118013443A (en) * | 2024-04-08 | 2024-05-10 | 华侨大学 | Online real-time vacuum dry pump abnormality detection method based on generation model algorithm |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948117A (en) * | 2019-03-13 | 2019-06-28 | 南京航空航天大学 | A Satellite Anomaly Detection Method for Adversarial Network Autoencoders |
| CN110428004A (en) * | 2019-07-31 | 2019-11-08 | 中南大学 | Component of machine method for diagnosing faults under data are unbalance based on deep learning |
| CN110567720A (en) * | 2019-08-07 | 2019-12-13 | 东北电力大学 | In-depth adversarial diagnosis method for fan bearing faults in unbalanced small sample scenarios |
| CN110647923A (en) * | 2019-09-04 | 2020-01-03 | 西安交通大学 | Variable working condition mechanical fault intelligent diagnosis method based on self-learning under small sample |
| CN111006865A (en) * | 2019-11-15 | 2020-04-14 | 上海电机学院 | Motor bearing fault diagnosis method |
| CN111767861A (en) * | 2020-06-30 | 2020-10-13 | 苏州兴钊防务研究院有限公司 | SAR image target identification method based on multi-discriminator generation countermeasure network |
| CN112039687A (en) * | 2020-07-14 | 2020-12-04 | 南京邮电大学 | A fault diagnosis method based on improved generative adversarial network for small sample features |
| CN113032917A (en) * | 2021-03-03 | 2021-06-25 | 安徽大学 | Electromechanical bearing fault detection method based on generation countermeasure and convolution cyclic neural network and application system |
| CN113095402A (en) * | 2021-04-12 | 2021-07-09 | 江南大学 | Method and system for detecting fault of generation countermeasure network based on coding input |
-
2021
- 2021-07-14 CN CN202110796968.XA patent/CN113485863B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948117A (en) * | 2019-03-13 | 2019-06-28 | 南京航空航天大学 | A Satellite Anomaly Detection Method for Adversarial Network Autoencoders |
| CN110428004A (en) * | 2019-07-31 | 2019-11-08 | 中南大学 | Component of machine method for diagnosing faults under data are unbalance based on deep learning |
| CN110567720A (en) * | 2019-08-07 | 2019-12-13 | 东北电力大学 | In-depth adversarial diagnosis method for fan bearing faults in unbalanced small sample scenarios |
| CN110647923A (en) * | 2019-09-04 | 2020-01-03 | 西安交通大学 | Variable working condition mechanical fault intelligent diagnosis method based on self-learning under small sample |
| CN111006865A (en) * | 2019-11-15 | 2020-04-14 | 上海电机学院 | Motor bearing fault diagnosis method |
| CN111767861A (en) * | 2020-06-30 | 2020-10-13 | 苏州兴钊防务研究院有限公司 | SAR image target identification method based on multi-discriminator generation countermeasure network |
| CN112039687A (en) * | 2020-07-14 | 2020-12-04 | 南京邮电大学 | A fault diagnosis method based on improved generative adversarial network for small sample features |
| CN113032917A (en) * | 2021-03-03 | 2021-06-25 | 安徽大学 | Electromechanical bearing fault detection method based on generation countermeasure and convolution cyclic neural network and application system |
| CN113095402A (en) * | 2021-04-12 | 2021-07-09 | 江南大学 | Method and system for detecting fault of generation countermeasure network based on coding input |
Non-Patent Citations (2)
| Title |
|---|
| 刘杰等: ""改进的生成对抗网络在轴承故障诊断中的应用"", 《噪声与振动控制》 * |
| 邢红杰等: ""基于全局和局部判别对抗自编码器的异常检测方法"", 《计算机科学》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116821697A (en) * | 2023-08-30 | 2023-09-29 | 聊城莱柯智能机器人有限公司 | Mechanical equipment fault diagnosis method based on small sample learning |
| CN116821697B (en) * | 2023-08-30 | 2024-05-28 | 聊城莱柯智能机器人有限公司 | Mechanical equipment fault diagnosis method based on small sample learning |
| CN117854716A (en) * | 2024-03-08 | 2024-04-09 | 长春师凯科技产业有限责任公司 | Method and system for filling heart disease diagnosis missing data based on HF-GAN |
| CN117854716B (en) * | 2024-03-08 | 2024-06-21 | 长春师凯科技产业有限责任公司 | Method and system for filling heart disease diagnosis missing data based on HF-GAN |
| CN118013443A (en) * | 2024-04-08 | 2024-05-10 | 华侨大学 | Online real-time vacuum dry pump abnormality detection method based on generation model algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113485863B (en) | 2023-05-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111914873B (en) | Two-stage cloud server unsupervised anomaly prediction method | |
| CN112784965B (en) | Large-scale multi-element time series data anomaly detection method oriented to cloud environment | |
| CN113485863A (en) | Method for generating heterogeneous unbalanced fault samples based on improved generation countermeasure network | |
| CN107561997B (en) | A kind of power equipment state monitoring method based on big data decision tree | |
| CN107877262A (en) | A kind of numerical control machine tool wear monitoring method based on deep learning | |
| CN118094437B (en) | Method for enhancing factory automation and digital engineering environment | |
| Mathew et al. | Regression kernel for prognostics with support vector machines | |
| CN114091549B (en) | A device fault diagnosis method based on deep residual network | |
| CN114266301A (en) | Intelligent prediction method of power equipment fault based on graph convolutional neural network | |
| CN116842379A (en) | A method for predicting the remaining service life of mechanical bearings based on DRSN-CS and BiGRU+MLP models | |
| CN116467652A (en) | Bearing fault prediction method and system based on convolution noise reduction self-encoder | |
| CN116398418A (en) | Nuclear power circulating water pump online abnormality monitoring and identifying method | |
| CN116910559A (en) | Index anomaly detection method for intelligent operation and maintenance application of power grid supercomputer center | |
| CN115659135A (en) | Anomaly detection method for multi-source heterogeneous industrial sensor data | |
| Liu et al. | Information-based Gradient enhanced Causal Learning Graph Neural Network for fault diagnosis of complex industrial processes | |
| CN118228130B (en) | Monitoring method, system and storage medium based on equipment health state | |
| CN112380763A (en) | System and method for analyzing reliability of in-pile component based on data mining | |
| CN117614680A (en) | A method for detecting abnormal traffic in industrial control networks based on data enhancement and deep learning | |
| CN114547796B (en) | Ball mill feature fusion fault diagnosis method based on optimized BN network | |
| CN116304835A (en) | AI-based dynamic equipment working condition monitoring management method, system, equipment and medium | |
| Ceylan et al. | Siamese inception time network for remaining useful life estimation | |
| CN112348275A (en) | Regional ecological environment change prediction method based on online incremental learning | |
| CN114530163A (en) | Method and system for recognizing life cycle of equipment by adopting voice based on density clustering | |
| CN115249059A (en) | Model training and abnormal data analysis method and device and computer storage medium | |
| Chen et al. | An aero–engine remaining useful life prediction model based on clustering analysis and the improved GRU–TCN |
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 |