CN117236188A - Simulation data-assisted bearing migration diagnosis method, system and equipment medium - Google Patents
Simulation data-assisted bearing migration diagnosis method, system and equipment medium Download PDFInfo
- Publication number
- CN117236188A CN117236188A CN202311325892.8A CN202311325892A CN117236188A CN 117236188 A CN117236188 A CN 117236188A CN 202311325892 A CN202311325892 A CN 202311325892A CN 117236188 A CN117236188 A CN 117236188A
- Authority
- CN
- China
- Prior art keywords
- bearing
- data
- fault
- domain
- training
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 65
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000004088 simulation Methods 0.000 title claims abstract description 46
- 238000013508 migration Methods 0.000 title claims abstract description 32
- 230000005012 migration Effects 0.000 title claims abstract description 32
- 238000012549 training Methods 0.000 claims abstract description 81
- 238000012360 testing method Methods 0.000 claims abstract description 31
- 230000036541 health Effects 0.000 claims abstract description 25
- 238000005259 measurement Methods 0.000 claims abstract description 23
- 238000005096 rolling process Methods 0.000 claims abstract description 15
- 238000010276 construction Methods 0.000 claims description 29
- 238000000605 extraction Methods 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 17
- 238000003860 storage Methods 0.000 claims description 15
- 230000001133 acceleration Effects 0.000 claims description 8
- 238000002405 diagnostic procedure Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000013480 data collection Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 14
- 238000010586 diagram Methods 0.000 description 6
- 238000009826 distribution Methods 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013016 damping Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The application discloses a simulation data-aided bearing migration diagnosis method, a system and equipment media, which are used for generating bearing simulation signals with different health states as labeled source domain data; collecting actual measurement signals of bearings in different health states, and taking the actual measurement signals as unlabeled target domain data; extracting high-dimensional fault characteristic information from source domain data and target domain data; inputting the high-dimensional fault characteristic information into a fault classification network to perform fault classification training; inputting the high-dimensional fault characteristic information into a domain classification network, and performing classification training of a source domain and a target domain; inputting the high-dimensional fault characteristic information into a weighted domain identification network, and training entropy values of known faults and unknown faults in a target domain; optimizing training parameters of the module by using a reverse error propagation method; and (3) performing repeated training by using the source domain data and the target domain data, selecting a diagnosis model with optimal training accuracy and testing accuracy, and performing fault diagnosis on the rolling bearing equipment.
Description
Technical Field
The application belongs to the field of bearing fault diagnosis, and relates to a simulation data-aided bearing migration diagnosis method, a simulation data-aided bearing migration diagnosis system and equipment media.
Background
The bearing is used as a key part of the rotary machine, and can run in a high-temperature, fatigue and heavy-load complex environment for a long time, and once the bearing fails, serious accidents can be caused, so that huge economic loss and casualties are caused. Therefore, the method has important economic benefits for monitoring and diagnosing the health state of the bearing.
In recent years, intelligent fault diagnosis methods have received a great deal of attention thanks to the rapid development of new generation artificial intelligence technology. The deep learning model is developed and successfully applied to the field of intelligent fault diagnosis of end-to-end mechanical equipment, and is widely researched by students at home and abroad. However, many deep learning models exhibit superior diagnostic performance that is not sufficient, labeled, and uniformly distributed samples of fault data. In practical engineering, it is difficult or impossible to obtain a data sample containing failure tag information for reasons of cost, safety, and the like. Furthermore, since mechanical devices often operate under different conditions of load and rotational speed, etc., the measured data often varies from one distribution to another, and models trained on data (source domain) in one distribution are often difficult to apply directly to data (target domain) in another distribution. Therefore, the unsupervised cross-domain fault diagnosis task has more practical research significance.
Unsupervised domain adaptation is an important migration learning strategy, and can be used for solving the problem of poor model generalization capability caused by domain offset. In recent years, researchers have sequentially developed unsupervised field-adaptive mechanical fault diagnosis studies based on feature transformation and countermeasure learning, respectively.
The migration method based on feature transformation and countermeasure learning shows good learning ability and migration effect in various unsupervised cross-domain fault diagnosis tasks, provides important reference for subsequent research in the field, and still needs to solve some challenging problems for further improving the accuracy, stability and generalization of diagnosis.
(1) Most of bearing fault source domain data in the current research come from a fault simulation test bed in a laboratory, the data are more suitable for researching general rules of fault phenomena, and the specific construction of the simulation test bed with certain precision can be realized in a non-first time, and the requirement of a large amount of resources is required to be durable, so that the fault data requirements under a plurality of working conditions are difficult to flexibly meet.
(2) Conventional migration methods all assume that the target domain data and the source domain data must have a common health state, however, in engineering practice, the target domain data not only includes a health state common to the source domain data, but also may often include a fault state that does not occur in the source domain data, i.e., open set diagnosis. Therefore, when the conventional migration method is used for open set diagnosis, since there is no sign of additional faults at the time of training the method, the faults cannot be recognized, and are misclassified into other health states, so that the diagnosis accuracy is lowered.
Disclosure of Invention
The application aims to overcome the defects of the prior art, and provides a simulation data-aided bearing migration diagnosis method, a simulation data-aided bearing migration diagnosis system and a simulation data-aided equipment medium, which can establish a fault simulation model reflecting the actual running condition of a mechanical system, acquire a large number of bearing fault samples with rich fault information and sufficient tag data, so as to solve the problem of insufficient training samples; in training, the influence of additional faults in the target domain is removed, and the common health state in the target domain is identified by utilizing the fault information of the source domain, so that the diagnosis precision is improved.
In order to achieve the purpose, the application is realized by adopting the following technical scheme:
a simulation data-aided bearing migration diagnosis method comprises the following steps:
step 1), building a bearing virtual simulation model, and generating bearing simulation signals with different health states as labeled source domain data;
step 2), building a bearing fault test bed, and collecting bearing actual measurement signals in different health states to serve as label-free target domain data;
step 3), constructing a feature extraction module according to the bearing source domain data and the target domain data obtained in the step 1) and the step 2), and extracting high-dimensional fault feature information from the source domain data and the target domain data;
step 4), constructing a fault classification network, inputting the extracted high-dimensional fault characteristic information into the fault classification network, and performing fault classification training;
step 5), constructing a domain classification network, inputting the extracted high-dimensional fault characteristic information into the domain classification network, and performing classification training of a source domain and a target domain;
step 6), constructing a weighted domain identification network, inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network, and training entropy values of known faults and unknown faults in a target domain;
step 7), optimizing training parameters of the modules in the steps 3) to 6) by using a reverse error propagation method;
step 8), repeating training of the steps 1) to 7) by utilizing the source domain data of the simulation signals in the step 1) and the target domain data of the actual measurement signals in the step 2), and selecting an optimal diagnosis model according to the statistical training accuracy and the test accuracy;
and 9) performing fault diagnosis on the rolling bearing device by using the optimal diagnosis model.
Preferably, in step 1), the data generated by the bearing virtual simulation model is a vibration acceleration signal.
Preferably, in step 2), the bearing test bed is used to collect target domain data, which is a bearing vibration acceleration signal, all of the source domain data set and part of the target domain data set are used for model training, and the remaining target domain data set is used for model testing.
Preferably, in step 4), the failure classification network comprises a 3-layer fully connected layer, a 2-layer ReLU layer, a 1-layer Dropout layer, and a 1-layer Softmax layer.
Preferably, in step 5), the domain classification network comprises a 2-layer fully connected layer, a 1-layer ReLU layer, a 1-layer Sigmoid layer, and a 1-layer gradient inversion layer.
Preferably, in step 6), the weighted domain discrimination network comprises a layer 2 full connection layer, a layer 1 ReLU layer and a layer 1 Sigmoid layer.
Preferably, in step 7), gradient calculation is performed according to the losses generated by the fault classification network, the domain classification network and the weighted domain discrimination network, and compensation optimization is performed on the training parameters in steps 3) to 6) according to the calculated gradients.
A simulated data aided bearing migration diagnostic system comprising:
the bearing virtual simulation data generation module is used for establishing a bearing virtual simulation model and generating bearing simulation signals with different health states as labeled source domain data;
the actual measurement bearing data acquisition module is used for building a bearing fault test bed and acquiring bearing actual measurement signals in different health states as label-free target domain data;
the feature extraction module construction module is used for constructing a feature extraction module from the source domain data and the target domain data and extracting high-dimensional fault feature information from the source domain data and the target domain data;
the fault classification network construction module is used for inputting the extracted high-dimensional fault characteristic information into a fault classification network to perform fault classification training;
the domain classification network construction module is used for inputting the extracted high-dimensional fault characteristic information into a domain classification network to perform classification training of a source domain and a target domain;
the weighted domain identification network construction module is used for inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network and carrying out entropy training of known faults and unknown faults in the target domain;
the training parameter optimization module is used for optimizing training parameters in the feature extraction module construction module, the fault classification network construction module, the field classification network construction module and the weighting field discrimination network construction module by using a reverse error propagation method;
the diagnosis model selection module is used for carrying out repeated training of the first seven modules by utilizing the simulation source domain data generated by the bearing virtual simulation data generation module and the actual measurement target domain data collected by the actual measurement bearing data collection module, and selecting a diagnosis model with optimal training accuracy and test accuracy;
and the application module is used for carrying out fault diagnosis on the rolling bearing of the test bed to be tested by using the optimal diagnosis model.
A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the simulated data assisted bearing migration diagnostic method when the computer program is executed.
A computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the simulated data aided bearing migration diagnostic method.
Compared with the prior art, the application has the following beneficial effects:
according to the bearing virtual simulation data generation module, the bearing virtual simulation model considering the load characteristic and the random noise characteristic is established to generate high-quality training fault sample data, a large number of bearing fault samples with rich fault information and sufficient label data are obtained, so that the problem of insufficient training samples is solved, and the problem of high-quality sample loss of typical fault diagnosis is solved. The weighted domain identification network adaptively weights known faults and unknown faults in the target domain by establishing a weighted domain prediction matrix and a probability prediction matrix, eliminates the influence of the unknown faults on virtual simulation data training of the source domain, and then utilizes the source domain fault information to realize the identification of the common health state in the target domain, thereby improving the diagnosis precision.
Drawings
FIG. 1 is a flow chart of a simulated data aided bearing migration diagnostic method of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the application; 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.
It should be noted that the words "front", "rear", "left", "right", "upper" and "lower" used in the following description refer to directions in the drawings, and the words "inner" and "outer" refer to directions toward or away from, respectively, the geometric center of a particular component.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the description of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
As shown in FIG. 1, the method for diagnosing the open set migration fault of the bearing driven by virtual simulation data comprises the following steps:
step 1), building a bearing virtual simulation model, and generating bearing simulation signals with different health states (normal state, inner ring fault, outer ring fault and rolling body fault) as labeled source domain data;
step 2), building a bearing fault test bed, and collecting actual measurement signals of bearings in different health states (normal state, inner ring fault, outer ring fault, rolling body fault and unknown fault) as label-free target domain data;
step 3), constructing a feature extraction module according to the bearing source domain data and the target domain data obtained in the step 1) and the step 2), and extracting high-dimensional fault feature information from the source domain data and the target domain data;
step 4), constructing a fault classification network, inputting the extracted high-dimensional fault characteristic information into the fault classification network, and performing fault classification training;
step 5), constructing a domain classification network, inputting the extracted high-dimensional fault characteristic information into the domain classification network, and performing classification training of a source domain and a target domain;
step 6), constructing a weighted domain identification network, inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network, and training entropy values of known faults and unknown faults in a target domain;
step 7), optimizing training parameters of the modules in the steps 3) to 6) by using a reverse error propagation method;
step 8), obtaining simulation source domain data by using the step 1) and obtaining actual measurement target domain data by using the step 2), performing repeated training of the steps 1) to 7), and selecting an optimal diagnosis model according to the statistical training accuracy and the test accuracy;
and 9) performing fault diagnosis on the rolling bearing device by using the optimal diagnosis model.
In the selected step 1), the data generated by the bearing virtual simulation model is a vibration acceleration signal.
In the selected step 2), collecting target domain data by using a bearing test bed, wherein the data is a bearing vibration acceleration signal, all source domain data sets and part of target domain data sets are used for model training, and the rest of target domain data sets are used for model testing.
In the selected step 3), the feature extraction module is composed of a convolutional neural network including 3 convolutional layers, 3 BatchNorm layers, 3 pooling layers, and 3 ReLU layers.
In the selected step 4), the high-dimensional fault characteristic information extracted in the step 3) is input into a fault classification network to complete training of a classification module, wherein the fault classification network comprises a 3-layer full-connection layer, a 2-layer ReLU layer, a 1-layer Dropout layer and a 1-layer Softmax layer.
In the selected step 5), the high-dimensional fault characteristic information extracted in the step 3) is input into a domain classification network, the domain classification network is trained, and the domain classification network comprises a 2-layer full-connection layer, a 1-layer ReLU layer, a 1-layer Sigmoid layer and a 1-layer gradient inversion layer.
In the selected step 6), the high-dimensional fault characteristic information extracted in the step 3) is input into a weighted domain identification network, and training is carried out on the high-dimensional fault characteristic information, wherein the weighted domain identification network comprises a 2-layer full-connection layer, a 1-layer ReLU layer and a 1-layer Sigmoid layer.
In the selected step 7), gradient calculation is carried out according to the losses generated by the fault classification network, the field classification network and the weighted field discrimination network, and the training parameters in the step 3) and the step 6) are compensated and optimized according to the calculated gradients.
The specific process is as follows:
1) In consideration of load characteristics and random noise characteristics, a bearing virtual simulation model is established to generate bearing simulation data, the bearing simulation data is used as source domain data, and a rolling bearing vibration acceleration signal can be expressed as:
where h (t) is the impulse response of the impact; i is the number of the impact; t is the time between two impacts; t is a discrete time variable; τ is the magnitude of uncertainty in the inter-arrival time; q (t) is the periodic modulation generated by the load distribution, respectively; n (t) is background noise.
n (t) is added random background noise, wherein the added random noise is Gaussian white noise, swedish noise and Poisson noise, and the expression is as follows:
wherein A and B are random variables subject to uniform distribution; f (t) is the frequency of the signal; t is the discretizationA time variable; c is the amplitude of the noise; delta (t) is a Dirac (Dirac) pulse function; t is t 0 The time when the noise appears; d represents the amplitude or standard deviation of the noise; random () is a function that generates random numbers subject to uniform distribution; λ represents a parameter of the poisson distribution, typically representing the number of average event occurrences per unit time.
The simulation signal of the rolling bearing is actually a numerical realization of the vibration signal in the above formula, and in order to take into account the speed variation of the simulation model, the vibration signal can be defined in the angular domain according to the speed distribution and then converted back into the time domain. Firstly, defining a random function theta as an equivalent function of iT in an angle domain, wherein the expected value of the random function theta represents an average angle formed by two continuous impacts:
Θ＝Θ i+1 -Θ i
in the formula Θ i Is the angle of separation between two successive impacts; ΔΘ imp For defining the angular position of equidistant bursts; o (O) imp The failure order of the bearing characteristic is related to the model parameter of the bearing.
The interval Δt can be expressed as:
wherein f r Is frequency conversion.
The impulse excitation excites the resonance of the bearing structure, assuming a single degree of freedom system, the response of the linear single degree of freedom system to each impulse can be expressed in the time domain as:
s(t)＝exp(-2πξf n t)sin(2πf d t)
wherein f n Is natural frequency; and xi is the damping coefficient.
Finally, the bearing virtual simulation model is numerically resolved as follows:
wherein i is the number of the impact; t is a discrete time variable; n is the total number of impacts; n (t) is background noise.
Data generated using a bearing virtual simulation model is defined as In order to simulate the source domain data,for the health state label corresponding to the simulation sample, n s Is the simulated source domain sample total.
2) The method for acquiring the vibration acceleration signal of the bearing by using the bearing test bed as target domain data can be expressed as For actually measuring the target domain data, n t The method is an unsupervised learning method, and therefore no label is contained. In addition, define the source domain class set as C s The target domain class set is C t Since the method is directed to open-set diagnosis, the method is performed by->
3) High-dimensional feature extraction is carried out on the simulation source domain data and the actual measurement target domain data obtained in the step 1) and the step 2) by using a feature extraction module, and high-dimensional fault features are obtained, wherein the high-dimensional fault features are obtainedThe medium feature extraction module is composed of a convolutional neural network, wherein the convolutional neural network comprises a 3-layer convolutional layer, a 3-layer BatchNorm layer, a 3-layer pooling layer and a 3-layer ReLU layer. By the formulaObtaining high-dimensional fault characteristics of simulation source domain data and actual measurement target domain data>G represents the feature extractor.
4) Constructing a fault classification network, and carrying out high-dimensional fault characteristic information extracted in the step 3)And inputting the fault classification network to complete training of the classification module, wherein the fault classification network comprises a 3-layer full-connection layer, a 2-layer ReLU layer, a 1-layer Dropout layer and a 1-layer Softmax layer. By the formula->Obtaining a probability prediction matrix corresponding to the simulation source domain characteristics>C represents a fault classification network. The loss function of the fault classification network is as follows in combination with the real fault labels of the simulated source domain features:
where L is the cross entropy loss function.
5) Constructing a domain classification network, and carrying out high-dimensional fault characteristic information extracted in the step 3)The method is input into a domain classification network, the domain classification network is trained, and the domain classification network comprises a 2-layer full-connection layer, a 1-layer ReLU layer, a 1-layer Sigmoid layer and a 1-layer gradient inversion layer. By the formula->Obtaining a sample domain prediction matrix->D represents a domain classification network. According to the sample domain information, the loss function of the domain classification network is as follows:
wherein n is st ＝min(n s ,n t )；d m The true domain label representing the mth feature is denoted as the emulated source domain when 0 and the measured target domain when 1.
6) Constructing a weighted domain discrimination network, and extracting high-dimensional fault characteristic information in the step 3)The data is input into a weighted domain discrimination network, and training is carried out on the weighted domain discrimination network, wherein the weighted domain discrimination network comprises a 2-layer full-connection layer, a 1-layer ReLU layer and a 1-layer Sigmoid layer. By the formula->Obtaining a weighted domain prediction matrix->E is a weighted domain authentication network. In training, self-adaptive weighting is carried out on the samples, the domain similarity of the target domain unknown set and the source domain is restrained by increasing the domain similarity of the target domain shared set and the source domain, and the domain similarity degree measured by the sample weight is controlled. And then, distinguishing the shared set sample and the unknown set sample in the source domain and the target domain by setting a weight threshold. The loss function of the weighted domain discrimination network is as follows:
wherein n is st ＝min(n s ,n t )；d m The true domain label representing the mth feature is denoted as the emulated source domain when 0 and the measured target domain when 1.
Acquiring weighted domain prediction matrices using weighted domain discrimination networksAcquiring probability prediction matrix corresponding to simulation source domain features by using fault classification network>Sample weights are obtained by the following formula:
wherein T; representing the total number of samples in each training;representing a weighted domain prediction matrix; />And representing the probability prediction matrix corresponding to the simulation source domain features.
7) And (3) through repeated training, selecting an optimal diagnosis model according to the statistical training accuracy and the test accuracy, and performing fault diagnosis on the rolling bearing equipment by using the optimal diagnosis model. By setting a reasonable threshold value, identifying unknown faults of the target domain, eliminating interference of the unknown faults, and identifying known fault bearing samples of the target domain by using a fault classifier obtained by training source domain data, wherein the method is expressed as follows:
wherein y is a predictive failure flagSigning; w (w) 0 A threshold value to distinguish between known faults and unknown faults of the target domain.
And 8) repeating training in the steps 1) to 7) by utilizing the source domain data of the simulation signals in the step 1) and the target domain data of the actually measured signals in the step 2), and selecting an optimal diagnosis model according to the statistical training accuracy and the test accuracy.
And 9) performing fault diagnosis on the rolling bearing device by using the optimal diagnosis model.
The following are device embodiments of the present application that may be used to perform method embodiments of the present application. For details of the device embodiment that are not careless, please refer to the method embodiment of the present application.
In still another embodiment of the present application, a simulation data-aided bearing migration diagnosis system is provided, where the simulation data-aided bearing migration diagnosis system may be used to implement the above-mentioned simulation data-aided bearing migration diagnosis method, and specifically, the simulation data-aided bearing migration diagnosis system includes a bearing virtual simulation data generation module, an actual measurement bearing data acquisition module, a feature extraction module construction module, a fault classification network construction module, a domain classification network construction module, a weighted domain discrimination network construction module, a training parameter optimization module, a diagnostic model selection module, and an application module.
The bearing virtual simulation data generation module is used for establishing a bearing virtual simulation model and generating bearing simulation signals with different health states as labeled source domain data.
The actually measured bearing data acquisition module is used for building a bearing fault test bed and acquiring bearing actually measured signals in different health states to serve as label-free target domain data.
The feature extraction module construction module is used for constructing a feature extraction module from the source domain data and the target domain data, and extracting high-dimensional fault feature information from the source domain data and the target domain data.
The fault classification network construction module is used for inputting the extracted high-dimensional fault characteristic information into a fault classification network to perform fault classification training.
The domain classification network construction module is used for inputting the extracted high-dimensional fault characteristic information into the domain classification network to perform classification training of the source domain and the target domain.
The weighted domain identification network construction module is used for inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network and carrying out entropy training of known faults and unknown faults in the target domain.
The training parameter optimization module is used for optimizing training parameters in the feature extraction module construction module, the fault classification network construction module, the domain classification network construction module and the weighted domain discrimination network construction module by using a reverse error propagation method.
The diagnosis model selection module is used for carrying out repeated training of the first seven modules by utilizing the simulation source domain data generated by the bearing virtual simulation data generation module and the actual measurement target domain data collected by the actual measurement bearing data collection module, and selecting a diagnosis model with optimal training accuracy and testing accuracy.
The application module is used for carrying out fault diagnosis on the rolling bearing of the test bed to be tested by using the optimal diagnosis model.
In yet another embodiment of the present application, a terminal device is provided, the terminal device including a processor and a memory, the memory for storing a computer program, the computer program including program instructions, the processor for executing the program instructions stored by the computer storage medium. The processor may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable gate arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., which are the computational core and control core of the terminal adapted to implement one or more instructions, in particular adapted to load and execute one or more instructions to implement a corresponding method flow or a corresponding function; the processor according to the embodiment of the application can be used for simulating the operation of a data-aided bearing migration diagnosis method, and comprises the following steps: step 1), building a bearing virtual simulation model, and generating bearing simulation signals with different health states as labeled source domain data; step 2), building a bearing fault test bed, and collecting bearing actual measurement signals in different health states to serve as label-free target domain data; step 3), constructing a feature extraction module according to the bearing source domain data and the target domain data obtained in the step 1) and the step 2), and extracting high-dimensional fault feature information from the source domain data and the target domain data; step 4), constructing a fault classification network, inputting the extracted high-dimensional fault characteristic information into the fault classification network, and performing fault classification training; step 5), constructing a domain classification network, inputting the extracted high-dimensional fault characteristic information into the domain classification network, and performing classification training of a source domain and a target domain; step 6), constructing a weighted domain identification network, inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network, and training entropy values of known faults and unknown faults in a target domain; step 7), optimizing training parameters of the modules in the steps 3) to 6) by using a reverse error propagation method; step 8), repeating training of the steps 1) to 7) by utilizing the source domain data of the simulation signals in the step 1) and the target domain data of the actually measured signals in the step 2), and selecting a diagnosis model with optimal training accuracy and testing accuracy; and 9) performing fault diagnosis on the rolling bearing device by using the optimal diagnosis model.
In still another embodiment, the present application also provides a computer-readable storage medium (Memory) that is a Memory device in a terminal device for storing programs and data. It will be appreciated that the computer readable storage medium herein may include both a built-in storage medium in the terminal device and an extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space storing an operating system of the terminal. Also stored in the memory space are one or more instructions, which may be one or more computer programs (including program code), adapted to be loaded and executed by the processor. The computer readable storage medium herein may be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as at least one magnetic disk memory.
One or more instructions stored in a computer-readable storage medium may be loaded and executed by a processor to implement the respective steps of the method for simulated data assisted bearing migration diagnosis of the above-described embodiments; one or more instructions in a computer-readable storage medium are loaded by a processor and perform the steps of: step 1), building a bearing virtual simulation model, and generating bearing simulation signals with different health states as labeled source domain data; step 2), building a bearing fault test bed, and collecting bearing actual measurement signals in different health states to serve as label-free target domain data; step 3), constructing a feature extraction module according to the bearing source domain data and the target domain data obtained in the step 1) and the step 2), and extracting high-dimensional fault feature information from the source domain data and the target domain data; step 4), constructing a fault classification network, inputting the extracted high-dimensional fault characteristic information into the fault classification network, and performing fault classification training; step 5), constructing a domain classification network, inputting the extracted high-dimensional fault characteristic information into the domain classification network, and performing classification training of a source domain and a target domain; step 6), constructing a weighted domain identification network, inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network, and training entropy values of known faults and unknown faults in a target domain; step 7), optimizing training parameters of the modules in the steps 3) to 6) by using a reverse error propagation method; step 8), repeating training of the steps 1) to 7) by utilizing the source domain data of the simulation signals in the step 1) and the target domain data of the actually measured signals in the step 2), and selecting a diagnosis model with optimal training accuracy and testing accuracy; and 9) performing fault diagnosis on the rolling bearing device by using the optimal diagnosis model.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
It is to be understood that the above description is intended to be illustrative, and not restrictive. Many embodiments and many applications other than the examples provided will be apparent to those of skill in the art upon reading the above description. The scope of the present teachings should, therefore, be determined not with reference to the above description, but instead should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled. The disclosures of all articles and references, including patent applications and publications, are incorporated herein by reference for the purpose of completeness. The omission of any aspect of the subject matter disclosed herein in the preceding claims is not intended to forego such subject matter, nor should the applicant not be considered to be a part of the disclosed subject matter.
Claims (10)
1. The simulation data assisted bearing migration diagnosis method is characterized by comprising the following steps of:
step 1), building a bearing virtual simulation model, and generating bearing simulation signals with different health states as labeled source domain data;
step 2), building a bearing fault test bed, and collecting bearing actual measurement signals in different health states to serve as label-free target domain data;
step 3), constructing a feature extraction module according to the bearing source domain data and the target domain data obtained in the step 1) and the step 2), and extracting high-dimensional fault feature information from the source domain data and the target domain data;
step 4), constructing a fault classification network, inputting the extracted high-dimensional fault characteristic information into the fault classification network, and performing fault classification training;
step 5), constructing a domain classification network, inputting the extracted high-dimensional fault characteristic information into the domain classification network, and performing classification training of a source domain and a target domain;
step 6), constructing a weighted domain identification network, inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network, and training entropy values of known faults and unknown faults in a target domain;
step 7), optimizing training parameters of the modules in the steps 3) to 6) by using a reverse error propagation method;
step 8), repeating training of the steps 1) to 7) by utilizing the source domain data of the simulation signals in the step 1) and the target domain data of the actual measurement signals in the step 2), and selecting an optimal diagnosis model according to the statistical training accuracy and the test accuracy;
and 9) performing fault diagnosis on the rolling bearing device by using the optimal diagnosis model.
2. The method of claim 1, wherein in step 1), the data generated by the virtual simulation model of the bearing is a vibration acceleration signal.
3. The method of claim 1, wherein in step 2), the bearing test bed is used to collect target domain data, which is a bearing vibration acceleration signal, and all of the source domain data set and part of the target domain data set are used for model training and the remaining target domain data set is used for model testing.
4. The simulated data aided bearing migration diagnostic method of claim 1 wherein in step 4) the fault classification network comprises 3 fully connected layers, 2 ReLU layers, 1 Dropout layers and 1 Softmax layers.
5. The simulated data aided bearing migration diagnostic method of claim 1 wherein in step 5) the domain classification network comprises a 2-layer fully connected layer, a 1-layer ReLU layer, a 1-layer Sigmoid layer and a 1-layer gradient inversion layer.
6. The simulated data aided bearing migration diagnostic method of claim 1 wherein in step 6), the weighted domain discrimination network comprises a 2-layer fully connected layer, a 1-layer ReLU layer and a 1-layer Sigmoid layer.
7. The method for simulation data aided bearing migration diagnosis of claim 1, wherein in step 7), gradient calculation is performed according to losses generated by the fault classification network, the domain classification network and the weighted domain discrimination network, and the training parameters in steps 3) to 6) are compensated and optimized according to gradients obtained by the gradient calculation.
8. A simulated data aided bearing migration diagnostic system comprising:
the bearing virtual simulation data generation module is used for establishing a bearing virtual simulation model and generating bearing simulation signals with different health states as labeled source domain data;
the actual measurement bearing data acquisition module is used for building a bearing fault test bed and acquiring bearing actual measurement signals in different health states as label-free target domain data;
the feature extraction module construction module is used for constructing a feature extraction module from the source domain data and the target domain data and extracting high-dimensional fault feature information from the source domain data and the target domain data;
the fault classification network construction module is used for inputting the extracted high-dimensional fault characteristic information into a fault classification network to perform fault classification training;
the domain classification network construction module is used for inputting the extracted high-dimensional fault characteristic information into a domain classification network to perform classification training of a source domain and a target domain;
the weighted domain identification network construction module is used for inputting the extracted high-dimensional fault characteristic information into the weighted domain identification network and carrying out entropy training of known faults and unknown faults in the target domain;
the training parameter optimization module is used for optimizing training parameters in the feature extraction module construction module, the fault classification network construction module, the field classification network construction module and the weighting field discrimination network construction module by using a reverse error propagation method;
the diagnosis model selection module is used for carrying out repeated training of the first seven modules by utilizing the simulation source domain data generated by the bearing virtual simulation data generation module and the actual measurement target domain data collected by the actual measurement bearing data collection module, and selecting a diagnosis model with optimal training accuracy and test accuracy;
and the application module is used for carrying out fault diagnosis on the rolling bearing of the test bed to be tested by using the optimal diagnosis model.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the simulated data assisted bearing migration diagnostic method according to any one of claims 1 to 7 when the computer program is executed.
10. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the simulated data assisted bearing migration diagnostic method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311325892.8A CN117236188A (en) | 2023-10-12 | 2023-10-12 | Simulation data-assisted bearing migration diagnosis method, system and equipment medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311325892.8A CN117236188A (en) | 2023-10-12 | 2023-10-12 | Simulation data-assisted bearing migration diagnosis method, system and equipment medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117236188A true CN117236188A (en) | 2023-12-15 |
Family
ID=89087927
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311325892.8A Pending CN117236188A (en) | 2023-10-12 | 2023-10-12 | Simulation data-assisted bearing migration diagnosis method, system and equipment medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117236188A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117723782A (en) * | 2024-02-07 | 2024-03-19 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
| CN117911852A (en) * | 2024-03-20 | 2024-04-19 | 西北工业大学 | Underwater target distance prediction method based on self-adaption in part of unsupervised field |
-
2023
- 2023-10-12 CN CN202311325892.8A patent/CN117236188A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117723782A (en) * | 2024-02-07 | 2024-03-19 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
| CN117723782B (en) * | 2024-02-07 | 2024-05-03 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
| CN117911852A (en) * | 2024-03-20 | 2024-04-19 | 西北工业大学 | Underwater target distance prediction method based on self-adaption in part of unsupervised field |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109580215B (en) | Wind power transmission system fault diagnosis method based on deep generation countermeasure network | |
| CN117236188A (en) | Simulation data-assisted bearing migration diagnosis method, system and equipment medium | |
| CN112161784B (en) | Mechanical fault diagnosis method based on multi-sensor information fusion migration network | |
| Li et al. | Self-attention ConvLSTM and its application in RUL prediction of rolling bearings | |
| CN114357594B (en) | Bridge abnormity monitoring method, system, equipment and storage medium based on SCA-GRU | |
| Pan et al. | A deep learning network via shunt-wound restricted Boltzmann machines using raw data for fault detection | |
| CN113065581B (en) | Vibration fault migration diagnosis method for reactance domain self-adaptive network based on parameter sharing | |
| Luleci et al. | CycleGAN for undamaged-to-damaged domain translation for structural health monitoring and damage detection | |
| CN111459144A (en) | Airplane flight control system fault prediction method based on deep cycle neural network | |
| CN111595541A (en) | Multi-dimensional structure damage identification method based on mass vibration transmissibility data convolutional neural network processing | |
| CN112414715B (en) | Bearing fault diagnosis method based on mixed feature and improved gray level symbiosis algorithm | |
| Sun et al. | Fault diagnosis for bearing based on 1DCNN and LSTM | |
| CN117708656B (en) | Rolling bearing cross-domain fault diagnosis method for single source domain | |
| CN116592993A (en) | Mechanical vibration fault diagnosis method based on deep learning | |
| CN114936575A (en) | Motor bearing fault diagnosis method based on sample generation and deep migration learning | |
| Chamangard et al. | Transfer learning for CNN‐based damage detection in civil structures with insufficient data | |
| Li et al. | Early gear pitting fault diagnosis based on bi-directional LSTM | |
| CN116049937A (en) | Cross-domain bridge damage identification method based on deep learning | |
| Wang et al. | Residual life prediction of bearings based on SENet-TCN and transfer learning | |
| CN117034430B (en) | Gate health monitoring method, system and computer readable storage medium based on deep learning and digital twin | |
| CN116700206B (en) | Industrial control system anomaly detection method and device based on multi-modal neural network | |
| CN110779722B (en) | Rolling bearing fault diagnosis method based on encoder signal local weighting | |
| CN108548669B (en) | Fault diagnosis method and system for transmission system of engineering equipment | |
| CN116401603A (en) | Multi-mode bearing fault intelligent diagnosis method based on transfer learning | |
| Bui-Ngoc et al. | Structural health monitoring using handcrafted features and convolution neural network |
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 |