CN112729831B - Bearing fault diagnosis method, device and system - Google Patents
Bearing fault diagnosis method, device and system Download PDFInfo
- Publication number
- CN112729831B CN112729831B CN202110078023.4A CN202110078023A CN112729831B CN 112729831 B CN112729831 B CN 112729831B CN 202110078023 A CN202110078023 A CN 202110078023A CN 112729831 B CN112729831 B CN 112729831B
- Authority
- CN
- China
- Prior art keywords
- neural network
- convolutional neural
- network model
- bearing
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The application discloses a bearing fault diagnosis method, a bearing fault diagnosis device and a bearing fault diagnosis system, which relate to the technical field of fault detection, wherein the method comprises the following steps: acquiring bearing data capable of reflecting the working state of a bearing; adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model; inputting the bearing data into the target layered convolution neural network model, and obtaining an output result of the target layered convolution neural network model, wherein the output result comprises at least one of the following items: bearing fault status, fault location, fault severity. The scheme of the application solves the problems that the current bearing fault diagnosis depends on thought factors, deep features cannot be extracted from original data, high-dimensional data cannot be well processed and the like, realizes accurate and quick bearing fault diagnosis, and ensures normal operation of equipment.
Description
Technical Field
The present disclosure relates to the field of fault detection technologies, and in particular, to a bearing fault diagnosis method, device, and system.
Background
Rolling bearings are widely used in rotary machines such as pumps, turbines, gearboxes, compressors, engines, etc., and are very susceptible to failure due to the complexity of the rotating equipment and the nature of the operating environment. Bearing failure accounts for 40% to 50% of all motor failures as shown by relevant statistics. When a bearing fails, it may cause serious economic loss and even be life threatening. Therefore, the automatic and accurate fault diagnosis of the rolling bearing is of great significance for maintaining the safe and stable operation of mechanical equipment.
In the prior art, fault diagnosis methods can be generally classified into model-based, signal-based and intelligent-based methods. The model-based method is to compare the actual measurement value obtained from the system with the output value generated by the system mathematical model, however, when the model-based method is adopted, the prior information of the system needs to be known in advance, otherwise, the model precision is greatly influenced. Signal-based diagnostic methods use fourier transforms, wavelet transforms, etc. to extract time, frequency, or time-frequency domain features from the measured signal, however, signal-based diagnostic methods rely heavily on a priori knowledge of the pattern analysis and monitoring system. In practice, this a priori knowledge is largely influenced by human factors and may not be available even in the case of system nonlinearities or highly complex operating conditions. Intelligent based approaches can utilize large amounts of data to address specific trends and patterns not seen by humans through machine learning. However, the conventional machine learning model still has the disadvantages that it is difficult to extract deep features from the original data and the high dimensional data cannot be processed well.
Disclosure of Invention
The application aims to provide a fault diagnosis detection method, a fault diagnosis detection device and a fault diagnosis detection system, so that the problems that fault diagnosis depends on human factors, deep features cannot be extracted, and high-dimensional data cannot be processed in the prior art are solved.
In order to achieve the above object, the present application provides a bearing fault diagnosis method including:
acquiring bearing data capable of reflecting the working state of a bearing;
adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model;
inputting the bearing data into the target layered convolution neural network model, and obtaining an output result of the target layered convolution neural network model, wherein the output result comprises at least one of the following items: bearing fault status, fault location, fault severity.
Optionally, the pre-constructed hierarchical convolutional neural network model includes a basic construction module of a convolutional neural network fused with a hierarchical structure of the bearing fault, and a plurality of prediction calculation modules;
wherein the prediction calculation module outputs an output result of the target hierarchical convolutional neural network model.
Optionally, the building process of the pre-constructed hierarchical convolutional neural network model includes:
constructing a hierarchy of bearing faults, wherein the hierarchy comprises a fault status layer, a fault location layer and a severity layer;
the basic building module of the convolutional neural network comprises at least three feature extraction modules and a prediction calculation module, wherein the at least three feature extraction modules are sequentially connected, and the prediction calculation module is arranged behind the last feature extraction module;
fusing the hierarchical structure of the bearing fault and the basic construction module based on the hierarchical structure of the convolutional neural network;
selecting two feature extraction modules according to a preset selection rule, and arranging a prediction calculation module behind each selected feature extraction module to construct the pre-constructed hierarchical convolutional neural network model;
wherein each of the prediction calculation modules outputs one parameter of the output result.
Optionally, adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model, including:
adjusting the basic building module according to the corresponding relation between the number of the pre-configured bearing data and the number of the feature extraction modules;
and adjusting the position of the prediction calculation module according to the number of the pre-configured feature extraction modules and the position information of the prediction calculation module.
Optionally, inputting the bearing data into the target hierarchical convolutional neural network model, and obtaining an output result of the target hierarchical convolutional neural network model, including:
dividing the bearing data into training set data, verification set data and test set data;
inputting the training set data into the target layered convolutional neural network model, and performing iterative training on the target layered convolutional neural network model;
inputting the verification set data into the target layered convolutional neural network model after iterative training, and adjusting the model parameters of the target layered convolutional neural network model after iterative training according to the verification result;
and inputting the test set data into the adjusted target hierarchical convolutional neural network model to obtain the output result.
Optionally, inputting the training set data to the target hierarchical convolutional neural network model, and performing iterative training on the target hierarchical convolutional neural network model, including:
inputting the training set data into the target layered convolutional neural network model to obtain a final loss value of the target layered convolutional neural network model;
and optimizing the target layered convolutional neural network model according to the final loss value so as to update the target layered convolutional neural network model.
Optionally, inputting the training set data to the target hierarchical convolutional neural network model to obtain a final loss value of the target hierarchical convolutional neural network model, including:
obtaining a loss value output by each prediction calculation module;
calculating a final loss value of the model based on a preconfigured loss weight and the loss value output by each of the predictive computation modules.
Optionally, the method further comprises:
acquiring diagnosis requirement information input by a user;
adjusting the weight of each layer in the hierarchical structure according to the diagnosis requirement information; and the hierarchical structure is the hierarchical structure of the bearing fault in the pre-constructed hierarchical convolutional neural network model.
Optionally, the method further comprises:
and executing a control strategy corresponding to the output result.
The embodiment of the present application further provides a bearing fault diagnosis device, including:
the first acquisition module is used for acquiring bearing data capable of reflecting the working state of the bearing;
the first adjusting module is used for adjusting a pre-constructed hierarchical convolutional neural network model according to the quantity of the bearing data to obtain a target hierarchical convolutional neural network model;
a second obtaining module, configured to input the bearing data into the target hierarchical convolutional neural network model, and obtain an output result of the target hierarchical convolutional neural network model, where the output result includes at least one of the following: bearing fault status, fault location, fault severity.
Optionally, the pre-constructed hierarchical convolutional neural network model includes a basic construction module of a convolutional neural network fused with a hierarchical structure of the bearing fault, and a plurality of prediction calculation modules;
wherein the prediction calculation module outputs an output result of the target hierarchical convolutional neural network model.
Optionally, the apparatus further comprises:
the building module is used for building the pre-built hierarchical convolutional neural network model;
the establishing module comprises:
the first construction submodule is used for constructing a hierarchical structure of the bearing fault, wherein the hierarchical structure comprises a fault state layer, a fault position layer and a severity layer;
the second construction submodule is used for constructing a basic construction module of the convolutional neural network, the basic construction module of the convolutional neural network comprises at least three feature extraction modules and a prediction calculation module, wherein the at least three feature extraction modules are sequentially connected, and the prediction calculation module is arranged behind the last feature extraction module;
the fusion submodule is used for fusing the hierarchical structure of the bearing fault and the basic construction module based on the hierarchical structure of the convolutional neural network;
the third construction submodule is used for selecting two feature extraction modules according to a preset selection rule, and a prediction calculation module is arranged behind each selected feature extraction module so as to construct the pre-constructed hierarchical convolutional neural network model;
wherein each of the prediction calculation modules outputs one parameter of the output result.
Optionally, the first adjusting module includes:
the first adjusting submodule is used for adjusting the basic building module according to the corresponding relation between the number of the pre-configured bearing data and the number of the feature extraction modules;
and the second adjusting submodule is used for adjusting the position of the prediction calculation module according to the number of the pre-configured feature extraction modules and the position information of the prediction calculation module.
Optionally, the second obtaining module includes:
the dividing submodule is used for dividing the bearing data into training set data, verification set data and test set data;
the training submodule is used for inputting the training set data into the target layered convolutional neural network model and performing iterative training on the target layered convolutional neural network model;
the third adjusting submodule is used for inputting the verification set data into the target layered convolutional neural network model after iterative training and adjusting the model parameters of the target layered convolutional neural network model after iterative training according to the verification result;
and the first obtaining submodule is used for inputting the test set data into the adjusted target hierarchical convolutional neural network model and obtaining the output result.
Optionally, the training submodule includes:
the second obtaining submodule is used for inputting the training set data into the target layered convolutional neural network model to obtain a final loss value of the target layered convolutional neural network model;
and the updating submodule is used for optimizing the target layered convolution neural network model according to the final loss value so as to update the target layered convolution neural network model.
Optionally, the second obtaining sub-module includes:
the obtaining unit is used for obtaining the loss value output by each prediction calculation module;
and the calculation unit is used for calculating the final loss value of the model according to a pre-configured loss weight and the loss value output by each prediction calculation module.
Optionally, the apparatus further comprises:
the third acquisition module is used for acquiring the diagnosis requirement information input by the user;
the second adjusting module is used for adjusting the weight of each layer in the hierarchical structure according to the diagnosis requirement information; and the hierarchical structure is the hierarchical structure of the bearing fault in the pre-constructed hierarchical convolutional neural network model.
Optionally, the apparatus further comprises:
and the execution module is used for executing the control strategy corresponding to the output result.
The embodiment of the present application further provides a bearing fault diagnosis system, including: a processor, a memory and a program stored on the memory and executable on the processor, the program when executed by the processor implementing the steps of the bearing fault diagnosis method as described above.
Embodiments of the present application also provide a readable storage medium, which stores a program, and the program, when executed by a processor, implements the steps of the bearing fault diagnosis method described above.
The above technical scheme of this application has following beneficial effect at least:
according to the bearing fault diagnosis method, firstly, bearing data capable of reflecting the working state of a bearing are obtained; facilitating subsequent diagnosis of the fault of the bearing based on the bearing data; secondly, adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model; the pre-constructed hierarchical convolutional neural network model is flexibly adjusted according to the number of the bearing data; the optimal target layered convolution neural network model can be selected according to specific tasks, so that the target layered convolution neural network model can be adapted to the bearing data; thirdly, inputting the bearing data into the target layered convolution neural network model, and obtaining an output result of the target layered convolution neural network model, wherein the output result comprises at least one of the following items: the bearing fault state, the fault position and the fault severity degree, so that the intelligent layered (on three levels) diagnosis of the bearing data is realized, and the bearing fault diagnosis method has strong adaptability.
Drawings
FIG. 1 is a schematic flow chart of a bearing fault diagnosis method according to an embodiment of the present application;
FIG. 2 is a schematic view of a hierarchy of bearing faults in an embodiment of the present application;
FIG. 3 is a schematic diagram of a base building module according to an embodiment of the present application;
FIG. 4 is a schematic diagram of a pre-constructed hierarchical convolutional neural network model according to an embodiment of the present application;
fig. 5 is a schematic structural diagram of a bearing fault diagnosis device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.
The terms first, second and the like in the description and in the claims of the present application are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application are capable of operation in sequences other than those illustrated or described herein. In addition, "and/or" in the specification and claims means at least one of connected objects, a character "/" generally means that a preceding and succeeding related objects are in an "or" relationship.
The bearing fault diagnosis method provided by the embodiment of the present application is described in detail below with reference to the accompanying drawings through specific embodiments and application scenarios thereof.
As shown in fig. 1, which is one of schematic diagrams of a flow of a bearing fault diagnosis method according to an embodiment of the present application, the method includes:
step 101: acquiring bearing data capable of reflecting the working state of a bearing;
here, it should be noted that the bearing data is specifically a signal capable of reflecting the working condition and/or the working state of the bearing, for example, a displacement, a speed, an acceleration, or the like of the bearing can be obtained, that is, the bearing data can be extracted from the signal; as another example, the bearing data may be data in accordance with a bearing data set provided by the university of kasseik, usa.
Step 102: adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model;
according to the bearing fault diagnosis method, in the bearing fault diagnosis process, firstly, a pre-constructed hierarchical convolutional neural network model is adjusted according to the number of bearing data to obtain a target hierarchical convolutional neural network matched with the currently acquired bearing data, so that one can obtain a more accurate diagnosis result; and the generality of the hierarchical convolutional neural network model is improved.
Step 103: inputting the bearing data into a target layered convolution neural network model, and obtaining an output result of the target layered convolution neural network model, wherein the output result comprises at least one of the following items: bearing fault status, fault location, fault severity.
In the step, the output result of the target hierarchical convolutional neural network is set to three levels including the bearing fault state, the fault position and the fault severity, so that the intelligent hierarchical diagnosis of the bearing fault is realized.
According to the bearing fault diagnosis method, firstly, bearing data capable of reflecting the working state of a bearing are obtained; facilitating subsequent diagnosis of the fault of the bearing based on the bearing data; secondly, adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model; the pre-constructed hierarchical convolutional neural network model is flexibly adjusted according to the number of the bearing data; the optimal target layered convolution neural network model can be selected according to specific tasks, so that the target layered convolution neural network model can be adapted to the bearing data; thirdly, inputting the bearing data into the target layered convolution neural network model, and obtaining an output result of the target layered convolution neural network model, wherein the output result comprises at least one of the following items: the bearing fault state, the fault position and the fault severity degree, so that the intelligent layered (on three levels) diagnosis of the bearing data is realized, and the bearing fault diagnosis method has strong adaptability.
As an optional implementation manner, the pre-constructed layered convolutional neural network model comprises a basic construction module of a convolutional neural network fused with a hierarchical structure of the bearing fault and a plurality of prediction calculation modules; and the prediction calculation module outputs the output result of the target hierarchical convolutional neural network model.
In the optional implementation mode, the hierarchical structure of the bearing fault is the basis for realizing intelligent hierarchical diagnosis, the hierarchical diagnosis of the bearing data is realized by fusing the hierarchical structure of the bearing fault with the basic construction module of the convolutional neural network model, and the hierarchical diagnosis of the bearing data is realized by arranging a plurality of prediction calculation modules, so that the bearing data is divided into different hierarchical structures, and the target hierarchical convolutional neural network model respectively outputs the diagnosis results of each layer to accurately position the bearing fault.
As an optional implementation manner, the building process of the pre-constructed hierarchical convolutional neural network model in step 102 includes:
the method comprises the following steps: constructing a hierarchical structure of the bearing fault, wherein the hierarchical structure comprises a fault state layer, a fault position layer and a severity layer;
here, it should be noted that the hierarchy of bearing failures constructed in this step may be a failure hierarchy of bearing data of the university of kasseiki, usa. As shown in fig. 2, the bearing fault can be divided into three levels according to different diagnosis bases according to the diagnosis requirement of the bearing fault in the actual application process. The first layer is divided into two types, and whether faults exist is judged, if no faults exist or faults exist; the second layer is divided into four types, and the positions where the faults exist are judged, such as: no fault, rolling element fault, inner ring fault and outer ring fault; the third layer is divided into ten categories to further diagnose the severity of the fault. That is, the first layer should be a failure status layer, the second layer should be a failure location layer, and the third layer should be a severity layer; further, according to the hierarchical structure of the bearing fault data, corresponding three labels, such as [1,1,1], are added to the fault data, and respectively correspond to whether a fault exists, the fault location and the severity of the fault. That is, the data output by the first layer is data representing a fault state, the data output by the second layer is data representing a fault position, and the data output by the third layer is data representing a fault severity, so that the bearing data is divided into different hierarchical structures, that is, the hierarchical structures are used for preprocessing the bearing data to obtain the bearing data of the labels corresponding to the respective hierarchical structures.
Step two: the basic building module of the convolutional neural network comprises at least three feature extraction modules and a prediction calculation module, wherein the at least three feature extraction modules are sequentially connected, and the prediction calculation module is arranged behind the last feature extraction module;
here, it should be noted that, as shown in fig. 3, each feature extraction module is composed of a convolution layer, a batch normalization layer, an activation layer, and a max-pooling layer. The convolution kernel size of the first layer is 32 × 1, and except for the first layer, the convolution kernels are all 3 × 1. The area size of all pool layers is 2 × 1.
Step three: fusing the hierarchical structure of the bearing fault and a basic construction module based on the hierarchical structure of the convolutional neural network;
the step can specifically fuse the hierarchical structure of the bearing fault with the basic building module according to the natural layering characteristic of the convolutional neural network.
Step four: selecting two feature extraction modules according to a preset selection rule, and setting a prediction calculation module behind each selected feature extraction module to construct the pre-constructed hierarchical convolutional neural network model; wherein each of the prediction calculation modules outputs one parameter of the output result.
In this step, as shown in fig. 4, taking the example that the basic building block includes five feature extraction modules, in addition to the prediction calculation module disposed after the fifth feature extraction module of the basic building block, an additional prediction calculation module is added after the pooling layers of the first feature extraction module and the third feature extraction module of the basic building block. Extracting simple features from the lower layer of the convolutional neural network, outputting a thicker diagnosis result, and performing simple diagnosis; and extracting more complex characteristics by a higher layer, outputting a finer diagnosis result, and performing precise diagnosis, thereby realizing the layered diagnosis of the bearing fault.
In this optional embodiment, the pre-constructed hierarchical convolutional neural network model is constructed by fusing the hierarchical structure and the basic construction module and arranging the plurality of prediction calculation modules behind the pooling layers of the different feature extraction modules, so that the hierarchical convolutional neural network model is constructed, the intelligent hierarchical diagnosis of the bearing data is realized, the output results including three parameters of the bearing fault state, the fault position and the fault severity are output, and the corresponding decision is conveniently made according to the output results and the development trend thereof in the following process.
As a specific implementation manner, step 102, adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data, and obtaining a target hierarchical convolutional neural network model, includes:
adjusting a basic construction module according to the corresponding relation between the number of the pre-configured bearing data and the number of the feature extraction modules;
here, it should be noted that the correspondence is a pre-configured correspondence, that is, the structure of the hierarchical convolutional neural network model used for different amounts of bearing data is different, such as: under the condition that the quantity of the bearing data is small, the structure of the layered convolutional neural network model is relatively simple, if the basic construction module only comprises three feature extraction modules, and under the condition that the quantity of the bearing data is large, the structure of the layered convolutional neural network model is relatively complex, if the basic construction module comprises seven feature extraction modules, of course, a user can input corresponding parameters according to the requirement of the user to adjust the pre-constructed layered convolutional neural network model.
And adjusting the position of the prediction calculation module according to the number of the pre-configured feature extraction modules and the position information of the prediction calculation module.
That is to say, in the target hierarchical convolutional neural network models with different structures, the positions of the prediction calculation modules are different, and after the basic construction module is adjusted according to the number of the bearing data, the positions of the prediction calculation modules can be adjusted according to the relationship between the number of the pre-configured feature extraction modules and the positions of the prediction calculation modules, so as to obtain the optimal hierarchical convolutional neural network model meeting the user requirements.
As an optional implementation manner, step 103, inputting the bearing data into the target hierarchical convolutional neural network model, and obtaining an output result of the target hierarchical convolutional neural network model, including:
the method comprises the following steps: dividing bearing data into training set data, verification set data and test set data;
in this step, the training set data is used to train the target hierarchical convolutional neural network model, the validation set data is used to adjust the model parameters, and the test set data is used to verify the model performance.
Step two: inputting training set data into a target layered convolutional neural network model, and performing iterative training on the target layered convolutional neural network model;
step three: inputting the verification set data into the target layered convolutional neural network model after iterative training, and adjusting the model parameters of the target layered convolutional neural network model after iterative training according to the verification result to enable the target layered convolutional neural network model to achieve the best diagnosis performance;
step four: and inputting the test set data into the adjusted target hierarchical convolutional neural network model to obtain an output result.
In the step, bearing data needing to be diagnosed is input into a trained model, three predicted values are sequentially output by the data in the flowing process of the model and respectively represent the bearing state, the fault position and the severity of the fault, and therefore the bearing state is diagnosed according to the output of the model.
In the optional implementation mode, the bearing data is divided into the training set data, the verification set data and the test set data, so that the training, the verification and the test of the pre-constructed hierarchical convolutional neural network model are sequentially realized, and thus, the target hierarchical convolutional neural network model is the optimal model according with the bearing data, so that a more accurate diagnosis result is obtained.
As a specific implementation manner, inputting training set data to a target hierarchical convolutional neural network model, and performing iterative training on the target hierarchical convolutional neural network model, including:
firstly, inputting training set data into a target layered convolutional neural network model to obtain a final loss value of the target layered convolutional neural network model;
in this step, as shown in fig. 5, in the model training process, the final loss value of the model is obtained by weighting and summing the loss values of each prediction calculation module according to the loss weight.
And secondly, optimizing the target layered convolutional neural network model according to the final loss value so as to update the target layered convolutional neural network model.
Here, it should be noted that the model parameters are optimized in the model training process, and after the training iteration, the model parameters may be adjusted.
As an optional implementation manner, the bearing fault detection method of the present application further includes:
acquiring diagnosis requirement information input by a user;
the diagnosis requirement information in this step may be simple diagnosis or precise diagnosis of the degree of detection of the bearing fault, which is input by the user according to the requirement.
Adjusting the weight of each layer in the hierarchical structure according to the diagnosis requirement information; the hierarchical structure is a hierarchical structure of bearing faults in a pre-constructed hierarchical convolutional neural network model.
In this step, if the diagnosis requirement information input by the user is simple diagnosis, the weights of the fault location layer and the fault severity layer in the hierarchical structure may be set to be smaller, for example, 0; if the diagnosis requirement information input by the user is precision diagnosis, the weights of the fault location layer and the fault severity layer in the hierarchical structure can be set to be larger.
In the optional implementation mode, the weights of all layers in the hierarchical structure are adjusted according to the diagnosis requirement information input by the user, so that the user can select simple diagnosis or precise diagnosis of the bearing data according to the requirement, and therefore the applicability and the universality of the bearing fault diagnosis method in the embodiment of the application are improved.
As an optional implementation manner, the bearing fault detection method of the present application further includes:
and executing a control strategy corresponding to the output result.
As mentioned above, the output result includes a fault state, a fault location and a fault severity, and the embodiment of the application may diagnose the state and the development trend of the bearing according to the output result, and make a corresponding control strategy, where the control strategy may include adjustment, control, maintenance or continuous monitoring, etc.
According to the fault diagnosis and detection method, a hierarchical structure of the bearing fault can be constructed according to the actual diagnosis requirement, and the flexibility is high; and the target layered convolutional neural network model comprises a basic construction module of the convolutional neural network fused with the hierarchical structure of the bearing fault and a plurality of prediction calculation modules, so that the intelligent layered diagnosis of the bearing data is realized, and three predicted values corresponding to the hierarchical structure of the bearing fault are finally output, namely: and diagnosing the bearing state and the fault state according to the three predicted values so as to further make a corresponding decision. Therefore, on one hand, bearing fault diagnosis does not depend on prior knowledge of a system, and the problems that the traditional learning model is difficult to extract deep features from original data and cannot well process high-dimensional data and the like are solved; on the other hand, the diagnosis precision is relatively high, and the operation is relatively convenient; on the other hand, the bearing state can be detected in real time, and a decision can be made in time according to a detection result, so that the normal operation of the equipment is ensured.
As shown in fig. 5, an embodiment of the present application further provides a bearing fault diagnosis device, including:
a first obtaining module 501, configured to obtain bearing data that can reflect a working state of a bearing;
a first adjusting module 502, configured to adjust a pre-constructed hierarchical convolutional neural network model according to the number of bearing data, to obtain a target hierarchical convolutional neural network model;
a second obtaining module 503, configured to input the bearing data into the target hierarchical convolutional neural network model, and obtain an output result of the target hierarchical convolutional neural network model, where the output result includes at least one of the following: bearing fault status, fault location, fault severity.
In the bearing fault diagnosis device according to the embodiment of the application, first, the first obtaining module 501 obtains bearing data capable of reflecting the working state of a bearing; facilitating subsequent diagnosis of the fault of the bearing based on the bearing data; secondly, the first adjusting module 502 adjusts the pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model; the pre-constructed hierarchical convolutional neural network model is flexibly adjusted according to the number of the bearing data; the optimal target layered convolution neural network model can be selected according to specific tasks, so that the target layered convolution neural network model can be adapted to the bearing data; thirdly, the second obtaining module 503 inputs the bearing data into the target layered convolutional neural network model, and obtains an output result of the target layered convolutional neural network model, where the output result includes at least one of the following: the bearing fault state, the fault position and the fault severity degree, so that the intelligent layered (on three levels) diagnosis of the bearing data is realized, and the bearing fault diagnosis method has strong adaptability.
Optionally, the pre-constructed hierarchical convolutional neural network model comprises a basic construction module of the convolutional neural network fused with the hierarchical structure of the bearing fault, and a plurality of prediction calculation modules;
wherein, the prediction calculation module outputs the output result of the target hierarchical convolutional neural network model.
Optionally, the apparatus further comprises:
the building module is used for building a pre-built layered convolutional neural network model;
the establishing module comprises:
the first construction submodule is used for constructing a hierarchical structure of the bearing fault, wherein the hierarchical structure comprises a fault state layer, a fault position layer and a severity layer;
the second construction submodule is used for constructing a basic construction module of the convolutional neural network, the basic construction module of the convolutional neural network comprises at least three feature extraction modules and a prediction calculation module, wherein the at least three feature extraction modules are sequentially connected, and the prediction calculation module is arranged behind the last feature extraction module;
the fusion submodule is used for fusing the hierarchical structure of the bearing fault and the basic construction module based on the hierarchical structure of the convolutional neural network;
the third construction submodule is used for selecting two feature extraction modules according to a preset selection rule, and a prediction calculation module is arranged behind each selected feature extraction module so as to construct the pre-constructed hierarchical convolutional neural network model;
wherein each of the prediction calculation modules outputs one parameter of the output result.
Optionally, the first adjusting module 502 includes:
the first adjusting submodule is used for adjusting the basic building module according to the corresponding relation between the number of the pre-configured bearing data and the number of the feature extraction modules;
and the second adjusting submodule is used for adjusting the position of the prediction calculation module according to the number of the pre-configured feature extraction modules and the position information of the prediction calculation module.
Optionally, the second obtaining module 503 includes:
the dividing submodule is used for dividing the bearing data into training set data, verification set data and test set data;
the training submodule is used for inputting training set data to the target layered convolutional neural network model and performing iterative training on the target layered convolutional neural network model;
the third adjusting submodule is used for inputting the verification set data into the target layered convolutional neural network model after iterative training and adjusting the model parameters of the target layered convolutional neural network model after iterative training according to the verification result;
and the first acquisition submodule is used for inputting the test set data into the adjusted target hierarchical convolutional neural network model and acquiring an output result.
Optionally, the training submodule comprises:
the second obtaining submodule is used for inputting the training set data into the target layered convolutional neural network model to obtain a final loss value of the target layered convolutional neural network model;
and the updating submodule is used for optimizing the target layered convolution neural network model according to the final loss value so as to update the target layered convolution neural network model.
Optionally, the second obtaining sub-module includes:
the obtaining unit is used for obtaining the loss value output by each prediction calculation module;
and the calculation unit is used for calculating the final loss value of the model according to a pre-configured loss weight and the loss value output by each prediction calculation module.
Optionally, the apparatus further comprises:
the third acquisition module is used for acquiring the diagnosis requirement information input by the user;
the second adjusting module is used for adjusting the weight of each layer in the hierarchical structure according to the diagnosis demand information; the hierarchical structure is a bearing fault hierarchical structure in a pre-constructed hierarchical convolutional neural network model.
Optionally, the apparatus further comprises:
and the execution module is used for executing the control strategy corresponding to the output result.
The embodiment of the present application further provides a bearing fault diagnosis system, including: the processor, the memory and the program stored in the memory and capable of running on the processor, when executed by the processor, implement each process of the embodiment of the bearing fault diagnosis method described above, and can achieve the same technical effect, and are not described herein again to avoid repetition.
The embodiment of the present application further provides a readable storage medium, where a program is stored on the readable storage medium, and when the program is executed by a processor, the program implements the processes of the bearing fault diagnosis method embodiment described above, and can achieve the same technical effects, and details are not repeated here to avoid repetition. The readable storage medium may be a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be 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. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal 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. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The foregoing is a preferred embodiment of the present application, and it should be noted that, for those skilled in the art, several modifications and refinements can be made without departing from the principle described in the present application, and these modifications and refinements should be regarded as the protection scope of the present application.
Claims (18)
1. A bearing fault diagnosis method, comprising:
acquiring bearing data capable of reflecting the working state of a bearing;
adjusting a pre-constructed hierarchical convolutional neural network model according to the number of the bearing data to obtain a target hierarchical convolutional neural network model;
inputting the bearing data into the target layered convolution neural network model, and obtaining an output result of the target layered convolution neural network model, wherein the output result comprises at least one of the following items: bearing fault state, fault location, fault severity;
the building process of the pre-built hierarchical convolutional neural network model comprises the following steps:
constructing a hierarchical structure of bearing faults, wherein the hierarchical structure comprises a fault state layer, a fault position layer and a severity layer;
the basic building module of the convolutional neural network comprises at least three feature extraction modules and a prediction calculation module, wherein the at least three feature extraction modules are sequentially connected, and the prediction calculation module is arranged behind the last feature extraction module;
fusing the hierarchical structure of the bearing fault and the basic construction module based on the hierarchical structure of the convolutional neural network;
selecting two feature extraction modules according to a preset selection rule, and arranging a prediction calculation module behind each selected feature extraction module to construct the pre-constructed hierarchical convolutional neural network model; wherein each of the prediction calculation modules outputs one parameter of the output result.
2. The method of claim 1, wherein the pre-constructed layered convolutional neural network model comprises a base building module of a convolutional neural network fused to a hierarchy of bearing faults, and a plurality of predictive computation modules;
wherein the prediction calculation module outputs an output result of the target hierarchical convolutional neural network model.
3. The method of claim 1, wherein adjusting a pre-constructed hierarchical convolutional neural network model according to the amount of the bearing data to obtain a target hierarchical convolutional neural network model comprises:
adjusting the basic building module according to the corresponding relation between the number of the pre-configured bearing data and the number of the feature extraction modules;
and adjusting the position of the prediction calculation module according to the number of the pre-configured feature extraction modules and the position information of the prediction calculation module.
4. The method of claim 2, wherein inputting the bearing data into the target hierarchical convolutional neural network model to obtain an output of the target hierarchical convolutional neural network model comprises:
dividing the bearing data into training set data, verification set data and test set data;
inputting the training set data into the target layered convolutional neural network model, and performing iterative training on the target layered convolutional neural network model;
inputting the verification set data into the target layered convolutional neural network model after iterative training, and adjusting the model parameters of the target layered convolutional neural network model after iterative training according to the verification result;
and inputting the test set data into the adjusted target hierarchical convolutional neural network model to obtain the output result.
5. The method of claim 4, wherein inputting the training set data to the target hierarchical convolutional neural network model, iteratively training the target hierarchical convolutional neural network model, comprises:
inputting the training set data into the target layered convolutional neural network model to obtain a final loss value of the target layered convolutional neural network model;
and optimizing the target layered convolutional neural network model according to the final loss value so as to update the target layered convolutional neural network model.
6. The method of claim 5, wherein inputting the training set data to the target hierarchical convolutional neural network model to obtain a final loss value of the target hierarchical convolutional neural network model comprises:
obtaining a loss value output by each prediction calculation module;
calculating a final loss value of the model based on a preconfigured loss weight and the loss value output by each of the predictive computation modules.
7. The method of claim 1, further comprising:
acquiring diagnosis requirement information input by a user;
adjusting the weight of each layer in the hierarchical structure according to the diagnosis requirement information; and the hierarchical structure is the hierarchical structure of the bearing fault in the pre-constructed hierarchical convolutional neural network model.
8. The method of claim 1, further comprising:
and executing a control strategy corresponding to the output result.
9. A bearing failure diagnosis device characterized by comprising:
the first acquisition module is used for acquiring bearing data capable of reflecting the working state of the bearing;
the first adjusting module is used for adjusting a pre-constructed hierarchical convolutional neural network model according to the quantity of the bearing data to obtain a target hierarchical convolutional neural network model;
a second obtaining module, configured to input the bearing data into the target hierarchical convolutional neural network model, and obtain an output result of the target hierarchical convolutional neural network model, where the output result includes at least one of the following: bearing fault state, fault location, fault severity;
the building module is used for building the pre-built hierarchical convolutional neural network model;
the establishing module comprises:
the first construction submodule is used for constructing a hierarchical structure of the bearing fault, wherein the hierarchical structure comprises a fault state layer, a fault position layer and a severity layer;
the second construction submodule is used for constructing a basic construction module of the convolutional neural network, the basic construction module of the convolutional neural network comprises at least three feature extraction modules and a prediction calculation module, wherein the at least three feature extraction modules are sequentially connected, and the prediction calculation module is arranged behind the last feature extraction module;
the fusion submodule is used for fusing the hierarchical structure of the bearing fault and the basic construction module based on the hierarchical structure of the convolutional neural network;
the third construction submodule is used for selecting two feature extraction modules according to a preset selection rule, and a prediction calculation module is arranged behind each selected feature extraction module so as to construct the pre-constructed hierarchical convolutional neural network model; wherein each of the prediction calculation modules outputs one parameter of the output result.
10. The apparatus of claim 9, wherein the pre-constructed hierarchical convolutional neural network model comprises a base construction module of a convolutional neural network fused with a hierarchy of bearing faults, and a plurality of predictive computation modules;
wherein the prediction calculation module outputs an output result of the target hierarchical convolutional neural network model.
11. The apparatus of claim 9, wherein the first adjusting module comprises:
the first adjusting submodule is used for adjusting the basic building module according to the corresponding relation between the number of the pre-configured bearing data and the number of the feature extraction modules;
and the second adjusting submodule is used for adjusting the position of the prediction calculation module according to the number of the pre-configured feature extraction modules and the position information of the prediction calculation module.
12. The apparatus of claim 10, wherein the second obtaining module comprises:
the dividing submodule is used for dividing the bearing data into training set data, verification set data and test set data;
the training submodule is used for inputting the training set data into the target layered convolutional neural network model and performing iterative training on the target layered convolutional neural network model;
the third adjusting submodule is used for inputting the verification set data into the target layered convolutional neural network model after iterative training and adjusting the model parameters of the target layered convolutional neural network model after iterative training according to the verification result;
and the first obtaining submodule is used for inputting the test set data into the adjusted target hierarchical convolutional neural network model and obtaining the output result.
13. The apparatus of claim 12, wherein the training submodule comprises:
the second acquisition sub-module is used for inputting the training set data into the target hierarchical convolutional neural network model to obtain a final loss value of the target hierarchical convolutional neural network model;
and the updating submodule is used for optimizing the target layered convolution neural network model according to the final loss value so as to update the target layered convolution neural network model.
14. The apparatus of claim 13, wherein the second acquisition submodule comprises:
the obtaining unit is used for obtaining the loss value output by each prediction calculation module;
and the calculation unit is used for calculating the final loss value of the model according to a pre-configured loss weight and the loss value output by each prediction calculation module.
15. The apparatus of claim 9, further comprising:
the third acquisition module is used for acquiring the diagnosis requirement information input by the user;
the second adjusting module is used for adjusting the weight of each layer in the hierarchical structure according to the diagnosis requirement information; and the hierarchical structure is the hierarchical structure of the bearing fault in the pre-constructed hierarchical convolutional neural network model.
16. The apparatus of claim 9, further comprising:
and the execution module is used for executing the control strategy corresponding to the output result.
17. A bearing fault diagnostic system, comprising: a processor, a memory and a program stored on the memory and executable on the processor, the program when executed by the processor implementing the steps of the bearing fault diagnosis method as claimed in any one of claims 1 to 8.
18. A readable storage medium, characterized in that the readable storage medium has stored thereon a program which, when executed by a processor, implements the steps of the bearing fault diagnosis method according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110078023.4A CN112729831B (en) | 2021-01-20 | 2021-01-20 | Bearing fault diagnosis method, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110078023.4A CN112729831B (en) | 2021-01-20 | 2021-01-20 | Bearing fault diagnosis method, device and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112729831A CN112729831A (en) | 2021-04-30 |
| CN112729831B true CN112729831B (en) | 2022-05-10 |
Family
ID=75594371
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110078023.4A Active CN112729831B (en) | 2021-01-20 | 2021-01-20 | Bearing fault diagnosis method, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112729831B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109406118A (en) * | 2018-10-15 | 2019-03-01 | 华中科技大学 | A kind of mechanical failure prediction method based on level convolutional neural networks |
| CN110031226A (en) * | 2019-04-12 | 2019-07-19 | 佛山科学技术学院 | A kind of diagnostic method and device of bearing fault |
| CN111562105A (en) * | 2020-03-25 | 2020-08-21 | 浙江工业大学 | Wind turbine generator gearbox fault diagnosis method based on wavelet packet decomposition and convolutional neural network |
| CN111626361A (en) * | 2020-05-28 | 2020-09-04 | 辽宁大学 | Bearing sub-health identification method for improving capsule network optimization layered convolution |
| CN111865702A (en) * | 2020-08-04 | 2020-10-30 | 北京航空航天大学 | Network fault diagnosis method and device based on convolutional neural network |
| WO2020244134A1 (en) * | 2019-06-05 | 2020-12-10 | 华南理工大学 | Multi-task feature sharing neural network-based intelligent fault diagnosis method |
-
2021
- 2021-01-20 CN CN202110078023.4A patent/CN112729831B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109406118A (en) * | 2018-10-15 | 2019-03-01 | 华中科技大学 | A kind of mechanical failure prediction method based on level convolutional neural networks |
| CN110031226A (en) * | 2019-04-12 | 2019-07-19 | 佛山科学技术学院 | A kind of diagnostic method and device of bearing fault |
| WO2020244134A1 (en) * | 2019-06-05 | 2020-12-10 | 华南理工大学 | Multi-task feature sharing neural network-based intelligent fault diagnosis method |
| CN111562105A (en) * | 2020-03-25 | 2020-08-21 | 浙江工业大学 | Wind turbine generator gearbox fault diagnosis method based on wavelet packet decomposition and convolutional neural network |
| CN111626361A (en) * | 2020-05-28 | 2020-09-04 | 辽宁大学 | Bearing sub-health identification method for improving capsule network optimization layered convolution |
| CN111865702A (en) * | 2020-08-04 | 2020-10-30 | 北京航空航天大学 | Network fault diagnosis method and device based on convolutional neural network |
Non-Patent Citations (2)
| Title |
|---|
| Hierarchical adaptive deep convolution neural network;Xiaojie Guo 等;《Measurement》;20160717(第93期);全文 * |
| 基于改进一维卷积神经网络的汽轮发电机组轴系扭振模态参数辨识;何成兵;《中国电机工程学报》;20200831;第40卷;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112729831A (en) | 2021-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7483774B2 (en) | Method and system for intelligent maintenance | |
| US7499777B2 (en) | Diagnostic and prognostic method and system | |
| CN112729834B (en) | Bearing fault diagnosis method, device and system | |
| CN109978134B (en) | Fault prediction method based on fast integrated convolutional neural network | |
| CN112765890B (en) | Dynamic domain adaptive network-based multi-working-condition rotating machine residual life prediction method | |
| CN103842924A (en) | Method of sequential kernel regression modeling for forecasting and prognostics | |
| CN113570138B (en) | Method and device for predicting residual service life of equipment of time convolution network | |
| CN111878320B (en) | Monitoring method and system of wind generating set and computer readable storage medium | |
| CN110515781B (en) | Complex system state monitoring and fault diagnosis method | |
| CN111881617A (en) | Data processing method, and performance evaluation method and system of wind generating set | |
| CN110119787A (en) | A kind of rotary-type operating condition of mechanical equipment detection method and equipment | |
| CN108830335A (en) | Photovoltaic plant fault early warning method and system | |
| CN114021784A (en) | Method and device for determining residual service life of equipment and electronic equipment | |
| CN115329986A (en) | Wind turbine generator anomaly detection and positioning method based on interpretable graph neural network | |
| CN115809596A (en) | Digital twin fault diagnosis method and device | |
| CN116384224A (en) | Aero-engine life prediction method based on conditional parameter dynamic convolutional neural network | |
| De Martini et al. | Fault detection of electromechanical actuators via automatic generation of a fuzzy index | |
| CN112729831B (en) | Bearing fault diagnosis method, device and system | |
| CN102788955B (en) | Remaining lifetime prediction method of ESN (echo state network) turbine generator classification submodel based on Kalman filtering | |
| CN116222753A (en) | Rotor system fault sensitivity feature extraction method and system | |
| CN115859791A (en) | Method and device for estimating residual life of aero-engine | |
| US20180087489A1 (en) | Method for windmill farm monitoring | |
| CN114781577A (en) | Buck circuit fault diagnosis method based on VMD-DCNN-SVM | |
| CN112560252A (en) | Prediction method for residual life of aircraft engine | |
| CN113586554A (en) | Hydraulic motor detection system and method |
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 |