CN112729834A - Bearing fault diagnosis method, device and system - Google Patents
Bearing fault diagnosis method, device and system Download PDFInfo
- Publication number
- CN112729834A CN112729834A CN202110077986.2A CN202110077986A CN112729834A CN 112729834 A CN112729834 A CN 112729834A CN 202110077986 A CN202110077986 A CN 202110077986A CN 112729834 A CN112729834 A CN 112729834A
- Authority
- CN
- China
- Prior art keywords
- bearing
- neural network
- convolutional neural
- fault
- dimensional convolutional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 81
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 122
- 238000012549 training Methods 0.000 claims abstract description 45
- 238000005070 sampling Methods 0.000 claims abstract description 8
- 238000012795 verification Methods 0.000 claims description 17
- 238000005096 rolling process Methods 0.000 claims description 16
- 238000012360 testing method Methods 0.000 claims description 15
- 230000001133 acceleration Effects 0.000 claims description 14
- 238000011156 evaluation Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 13
- 239000000284 extract Substances 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 4
- 238000010200 validation analysis Methods 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 12
- 238000003062 neural network model Methods 0.000 abstract description 6
- 238000001514 detection method Methods 0.000 abstract description 2
- 238000013528 artificial neural network Methods 0.000 description 8
- 238000010276 construction Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000011217 control strategy Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
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
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Acoustics & Sound (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: under a preset working condition and a sampling frequency, collecting bearing fault data when a fault bearing operates; constructing a one-dimensional convolution neural network model with a branch structure; training the one-dimensional convolutional neural network model according to the bearing fault data; and automatically extracting the state characteristics of the bearing through the trained one-dimensional convolutional neural network model, and acquiring the fault diagnosis result of the bearing. The scheme of the application has solved present bearing fault diagnosis and has been difficult to directly follow the deep characteristic of extraction in the primary data, need carry out extra preliminary treatment to data, and handle the not enough problem of high dimension data, has realized directly extracting effective characteristic from the one-dimensional data that the collection obtained, and the rate of accuracy is high, convenient operation.
Description
Technical Field
The present disclosure relates to the field of fault detection technologies, and in particular, to a method, an apparatus, and a system for diagnosing a bearing fault.
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. Statistics show that bearing failure accounts for 40 to 50% of all motor failures. 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. Model-based methods simulate an industrial process or an actual system from physical principles or system identification techniques, and compare actual measurements obtained from the system with output values generated by a mathematical model of the system via a correlation algorithm. However, when the model-based method is adopted, the prior information of the system needs to be known in advance, otherwise the model accuracy is greatly influenced. 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.
The traditional intelligent diagnosis algorithm is divided into two parts of feature extraction and classifier, and the key point is the selection of the feature extractor and the classifier, wherein the success of an intelligent fault diagnosis system is judged. The current research of intelligent diagnosis algorithm has the following problems: when the traditional bearing intelligent diagnosis algorithm is adopted for diagnosis, high recognition rate is achieved, different mechanical systems can correspond to combinations of different feature extractors and classifiers, and the universality of the algorithm cannot be guaranteed. And information loss is caused when the signal is subjected to feature extraction by using fast Fourier transform or wavelet transform.
In recent years, machine learning algorithms have been widely used in rolling bearing diagnostics. Machine learning can utilize large amounts of data to address specific trends and patterns not seen by humans. However, the conventional machine learning model still has some disadvantages. For example, it is difficult to extract deep features from raw data and high dimensional data is not processed well.
Disclosure of Invention
The embodiment of the application provides a bearing fault diagnosis method, device and system, and aims to solve the problems that deep features are extracted from original bearing fault diagnosis data and high-dimensional data are not enough to be processed at present.
In order to solve the technical problem, the following technical scheme is adopted in the application:
the embodiment of the application provides a bearing fault diagnosis method, which comprises the following steps:
under a preset working condition and a sampling frequency, collecting bearing fault data when a fault bearing operates;
constructing a one-dimensional convolution neural network model with a branch structure;
training and adjusting the one-dimensional convolutional neural network model according to the bearing fault data;
and extracting the state characteristics of the bearing through the trained one-dimensional convolution neural network model, and acquiring the fault diagnosis result of the bearing.
Optionally, the collecting bearing fault data when the faulty bearing operates includes:
adopting a preset bearing data set;
and dividing the bearing data set into a training set, a verification set and a test set according to a preset dividing proportion.
Optionally, the collecting bearing fault data when the fault bearing operates includes:
processing a plurality of bearings with different damage positions and different size faults;
running the bearings on a laboratory table, and collecting vibration acceleration signals corresponding to each rolling bearing;
and storing the fault position and the size information of each rolling bearing and the vibration acceleration signal data corresponding to the fault position and the size information of each rolling bearing, and establishing bearing fault data.
Optionally, the constructing a one-dimensional convolutional neural network model with a branch structure includes:
constructing a one-dimensional convolutional neural network basic model which can extract at least three pieces of characteristic information;
constructing a hierarchical structure of the bearing fault data, wherein the hierarchical structure comprises a fault status layer, a fault location layer and a fault severity layer;
and based on the characteristic of natural layering of a convolutional neural network model, fusing the layering structure of the bearing fault data with the one-dimensional convolutional neural network basic model to construct the one-dimensional convolutional neural network model with the branch structure.
Optionally, the constructing the one-dimensional convolutional neural network base model includes:
determining the size of a first convolution kernel of the structural parameters of the one-dimensional convolution neural network basic model as a first preset value;
determining the sizes of the rest convolution kernels of the structural parameters of the one-dimensional convolution neural network basic model as second preset values;
determining the sizes of a plurality of pool layers of the one-dimensional convolutional neural network basic model as a third preset value;
wherein the first preset value is greater than the second preset value.
Optionally, the training and adjusting the one-dimensional convolutional neural network model according to the bearing fault data includes:
inputting the training set into the one-dimensional convolutional neural network model for training to obtain a loss value of the one-dimensional convolutional neural network model;
and optimizing the one-dimensional convolutional neural network model according to the loss value of the model so as to update the one-dimensional convolutional neural network model.
Optionally, the training and adjusting the one-dimensional convolutional neural network model according to the bearing fault data includes:
inputting the verification set into the trained one-dimensional convolutional neural network model for verification to obtain an evaluation index and a performance index of the one-dimensional convolutional neural network model;
judging whether the evaluation index and the performance index are smaller than a preset threshold value or not;
and if the evaluation index and the performance index are smaller than a preset threshold value, adjusting the one-dimensional convolutional neural network model, otherwise, storing the trained one-dimensional convolutional neural network model.
Optionally, the obtaining a fault diagnosis result of the bearing includes:
inputting the test set into the one-dimensional convolutional neural network model, extracting the bearing state characteristics by the model, and outputting the fault diagnosis result;
wherein the fault diagnosis result comprises at least one of: fault status, fault location and fault severity of the bearing.
Optionally, after the state features of the bearing are extracted through the trained one-dimensional convolutional neural network model and the fault diagnosis result of the bearing is obtained, the method further includes:
matching the adaptive strategy signals from a preset table according to the fault diagnosis result; wherein the preset table includes: at least one fault diagnosis result, and a strategy signal corresponding to the fault diagnosis result.
The embodiment of the present application further provides a diagnostic device for bearing fault, including:
the acquisition module is used for acquiring bearing fault data when a fault bearing operates under the preset working condition and sampling frequency;
the building module is used for building a one-dimensional convolution neural network model with a branch structure;
the processing module is used for training and adjusting the one-dimensional convolutional neural network model according to the bearing fault data;
and the acquisition module is used for extracting the state characteristics of the bearing through the trained one-dimensional convolutional neural network model and acquiring the fault diagnosis result of the bearing.
The embodiment of the present application further provides a bearing fault diagnosis system, including: a processor, a memory and a program stored on said memory and executable on said processor, said program, when executed by said processor, implementing the steps of the method for diagnosing a bearing fault as described above.
Embodiments of the present application also provide a readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the method for diagnosing a bearing fault as described above.
The beneficial effect of this application is:
in the technical scheme, bearing fault data during the operation of a fault bearing is collected under the preset working condition and the sampling frequency; facilitating subsequent diagnosis of the fault of the bearing based on the bearing data; constructing a one-dimensional convolution neural network model with a branch structure; training the one-dimensional convolutional neural network model according to the bearing fault data; and automatically extracting the state characteristics of the bearing through the trained one-dimensional convolutional neural network model, and directly obtaining the fault diagnosis result of the bearing. According to the bearing fault diagnosis method, the original vibration signals are directly used as input, extra data processing is not needed, the difficulty of obtaining training samples is reduced, deeper fault features and more abstract information are extracted by combining the one-dimensional convolutional neural network basic model and the hierarchical structure of bearing fault data, high classification accuracy is achieved, and meanwhile the training difficulty is reduced.
Drawings
FIG. 1 is a schematic flow chart of a method for diagnosing bearing faults according to an embodiment of the present application;
FIG. 2 is a schematic structural diagram of a basic model of a one-dimensional convolutional neural network according to an embodiment of the present application;
FIG. 3 is a diagram illustrating a one-dimensional convolutional neural network model with a branched structure according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of a bearing fault diagnosis device according to an embodiment of the present application.
Detailed Description
To make the technical problems, technical solutions and advantages to be solved by the present application clearer, the following detailed description is made with reference to the accompanying drawings and specific embodiments. In the following description, specific details such as specific configurations and components are provided only to help the embodiments of the present application be fully understood. Accordingly, it will be apparent to those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present application. In addition, descriptions of well-known functions and constructions are omitted for clarity and conciseness.
It should be appreciated that reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
In various embodiments of the present application, it should be understood that the sequence numbers of the following processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.
The method, the device and the system for diagnosing the bearing fault are provided aiming at the problems that deep features are difficult to directly extract from original data in the conventional bearing fault diagnosis, additional preprocessing is required to be carried out on the data, and high-dimensional data are not enough to be processed.
It should be noted that, the present application provides a bearing fault diagnosis method, apparatus and system based on original one-dimensional data, so as to implement "end-to-end" bearing fault intelligent diagnosis by using a convolutional neural network model, in order to solve the problems that the existing model-based diagnosis method and signal-based diagnosis method rely on prior knowledge of the system, information loss is caused when a traditional intelligent diagnosis algorithm performs feature extraction, a traditional machine learning model is difficult to directly extract deep features from original data, and high-dimensional data cannot be well processed. The method does not need any data preprocessing operation, can directly extract effective characteristics from the collected one-dimensional data, and has high accuracy and simple operation.
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:
here, it should be noted that the bearing fault 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 fault data can be extracted from the signal; as another example, the bearing data may be in accordance with a bearing data set provided by the university of paderburn (Paderborn) of germany.
200, constructing a one-dimensional convolution neural network model with a branch structure;
here, a more accurate diagnosis result can be obtained by obtaining a one-dimensional convolutional neural network model having a branch structure; and deeper fault characteristics and more abstract information can be extracted, and the classification precision is high.
it should be noted that, the bearing state characteristics can be automatically learned from the original one-dimensional bearing fault data through the training model, and the optimal model more conforming to the bearing data can be obtained by adjusting the parameters of the model, so as to obtain a more accurate diagnosis result.
And 400, extracting the state characteristics of the bearing through the trained one-dimensional convolutional neural network model, and acquiring the fault diagnosis result of the bearing.
In the embodiment, an original one-dimensional signal, namely bearing fault data, is directly used as input to train a constructed diagnosis model; and inputting fault data into the trained network model, and diagnosing faults according to the output result of the model so as to make corresponding decisions. Therefore, effective characteristics can be directly extracted from the collected one-dimensional data without any data preprocessing operation, the intelligent diagnosis of the bearing fault based on the original one-dimensional data is realized, the accuracy is high, and the operation is simple.
Through the steps 100 to 400, the original vibration signal is directly used as input, extra data processing is not needed, the difficulty of obtaining a training sample is reduced, deeper fault features and more abstract information are extracted by combining a one-dimensional convolutional neural network basic model and a bearing fault data hierarchical structure, high classification precision is achieved, and meanwhile the training difficulty is reduced.
Optionally, the step 100 includes:
adopting a preset bearing data set;
and dividing the bearing data set into a training set, a verification set and a test set according to a preset dividing proportion.
In this embodiment, a preset bearing data set is adopted, wherein the bearing data set is preferably a bearing data set provided by university of paderburn in germany, and the original time series signal of the bearing data set is divided by the preset division ratio, and the division can be divided into a training set, a verification set and a test set. The training set data is used for training a pre-constructed one-dimensional convolutional neural network model with a branch structure, the verification set data is used for further adjusting the trained model parameters, and the test set data is used for checking the model performance.
In the optional implementation mode, the bearing data is divided into the training set, the verification set and the test set, so that the training, the verification and the test of the pre-constructed one-dimensional convolutional neural network model with the branch structure are sequentially realized, and thus, the one-dimensional convolutional neural network model with the branch structure is the optimal model according with the bearing data, and a more accurate diagnosis result is obtained.
Optionally, the step 100 further includes:
processing a plurality of bearings with different damage positions and different size faults;
running the bearings on a laboratory table, and collecting vibration acceleration signals corresponding to each rolling bearing;
and storing the fault position and the size information of each rolling bearing and the vibration acceleration signal data corresponding to the fault position and the size information of each rolling bearing, and establishing bearing fault data.
In this embodiment, in order to obtain data that can reflect different damage positions and different sizes of faults of the rolling bearing, the rolling bearing with the faults of different damage positions and different sizes needs to be processed, the vibration acceleration signal of the loading area is measured by using the acceleration sensor, and the rotation speed of the rolling bearing is measured by using the eddy current sensor. And measuring vibration acceleration data of each fault bearing in different working conditions, and storing the vibration acceleration data and the data by using the damage position of the bearing and the size of the fault as labels of the collected data. Wherein the different operating conditions include applying different rotational speeds, load torques, and accelerations during the bearing fault data measurement.
In an alternative embodiment, the step 200 includes:
step 210, constructing a one-dimensional convolutional neural network basic model, wherein the one-dimensional convolutional neural network basic model can extract at least three pieces of characteristic information;
here, it should be noted that each one-dimensional convolutional neural network basic model is composed of a convolutional layer, a batch normalization layer, an activation layer, and a maximum pooling layer.
Step 220, constructing a hierarchical structure of the bearing fault data, wherein the hierarchical structure comprises a fault state layer, a fault position layer and a fault severity layer;
taking the fault hierarchy of the bearing data of the university of paderburn (Paderborn) in germany as an example, the bearing fault is classified into three different classification levels according to different classification manners such as the existence position of the bearing fault, the severity of the fault and the like. The first layer is divided into two types, and whether a fault exists is judged; the second layer is divided into three types, and the position of the fault is judged; the third layer is divided into five categories to further diagnose the severity of the fault. After the hierarchical structure of the bearing fault is constructed, the method also comprises the following steps: according to the hierarchical structure of the bearing fault data, corresponding three labels, such as [0,0,0], are added to the bearing fault data, and whether a fault exists in the hierarchical structure of the bearing fault data, a fault position and a fault damage level are respectively corresponding, so that a fault state layer, a fault position layer and a fault severity layer can be reflected.
And step 230, fusing the layered structure of the bearing fault data and the one-dimensional convolutional neural network basic model based on the characteristic of natural layering of a convolutional neural network model, and constructing the one-dimensional convolutional neural network model with the branch structure.
In this step, as shown in fig. 3, taking a one-dimensional convolutional neural network model with a branch structure including five feature extraction modules as an example, the model can extract at least three pieces of feature information and output three different diagnosis results, so in addition to the output layer at the end of the conventional convolutional neural network, an additional branch structure is added after the pooling layers of the first feature extraction module and the third feature extraction module of the one-dimensional convolutional neural network basic model to extract feature information. 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 multilayer diagnosis of the bearing fault.
In this optional embodiment, the one-dimensional convolutional neural network model with a branch structure, which is constructed in advance, is constructed by fusing the hierarchical structure of step 220 and the one-dimensional convolutional neural network basic model constructed in step 210, so that the one-dimensional convolutional neural network model with a branch structure is constructed, the intelligent hierarchical diagnosis of the bearing data is realized, and the output results including three predicted values of the bearing fault state, the fault position and the fault severity are output, so that corresponding decisions can be made subsequently according to the output results and the development trend thereof.
Of course, when the acquired data amount of the bearing fault is small, the one-dimensional convolutional neural network model with the branch structure may only include three feature extraction modules, and when the number of the bearing data is large, the number of the feature extraction modules of the one-dimensional convolutional neural network basic model constructed in step 210 may be increased to enhance the feature extraction capability of the model, and of course, the user may also input corresponding parameters to adjust the one-dimensional convolutional neural network model with the branch structure according to the own requirements.
Specifically, as shown in fig. 2, the constructing a one-dimensional convolutional neural network base model includes:
determining a first convolution kernel size of the structural parameters of the one-dimensional convolution neural network basic model as a first preset value,
determining the sizes of the rest convolution kernels of the structural parameters of the one-dimensional convolution neural network basic model as second preset values;
determining the sizes of a plurality of pool layers of the one-dimensional convolutional neural network basic model as a third preset value;
wherein the first preset value is greater than the second preset value.
In this embodiment, a one-dimensional convolutional neural network base model is constructed, where the one-dimensional convolutional neural network base model includes structure parameters including but not limited to convolutional kernel size and training hyper-parameters including but not limited to number of rounds of training and learning rate; here, constructing the one-dimensional convolutional neural network base model includes five feature extraction modules. Each module is composed of a convolution layer, a batch normalization layer, an activation layer and a maximum pooling layer. Determining the first convolution kernel size of the structural parameters of the one-dimensional convolution neural network basic model as a first preset value, namely, the first layer convolution kernel size is 32 multiplied by 1; determining the sizes of the rest convolution kernels of the structural parameters of the one-dimensional convolution neural network basic model to be second preset values, namely, the sizes of the rest convolution kernels except the first layer are all 3 multiplied by 1; and determining the sizes of a plurality of pool layers of the one-dimensional convolutional neural network basic model as a third preset value, namely the area sizes of all the pool layers are 2 multiplied by 1.
Optionally, the step 300 includes:
step 310, inputting the training set into the one-dimensional convolutional neural network model with the branch structure for training to obtain a loss value of the one-dimensional convolutional neural network model;
in this step, as shown in fig. 3, in the model training process, the final loss value of the model is obtained by weighting and summing the loss values of each branch structure according to the loss weight.
And secondly, optimizing the one-dimensional convolutional neural network model with the branch structure according to the loss value of the model so as to update the one-dimensional convolutional neural network model with the branch structure.
In the implementation process of the step, one-dimensional bearing data is directly used as model input without any operations of wavelet transformation, fast Fourier transformation and other feature extraction, the training set of the bearing fault data is used for training a one-dimensional convolutional neural network model with branch structures, the final loss value of the model is obtained by weighting and summing the loss value of each branch structure according to the loss weight, and optimization is carried out in the training process. Updating the model after training iteration; the validation set of bearing fault data is used to adjust model parameters, and the test set of bearing fault data verifies the diagnostic performance of the model.
Step 320, inputting the verification set into the trained one-dimensional convolutional neural network model for verification to obtain an evaluation index and a performance index of the one-dimensional convolutional neural network model;
judging whether the evaluation index and the performance index are smaller than a preset threshold value or not;
and if the evaluation index and the performance index are smaller than a preset threshold value, adjusting the one-dimensional convolutional neural network model, otherwise, storing the trained one-dimensional convolutional neural network model.
In this embodiment, the verification set is input into the trained one-dimensional convolutional neural network model for verification, so as to obtain an evaluation index and a performance index of the one-dimensional convolutional neural network model; judging whether the evaluation index and the performance index are smaller than a preset threshold value, if so, adjusting the structural parameters and the training parameters, inputting the training set into the one-dimensional convolutional neural network model again for training, and if not, storing the trained one-dimensional convolutional neural network model; through the trained one-dimensional convolutional neural network model, deeper fault characteristics and more abstract information are extracted, and the training difficulty is reduced while the classification precision is high.
Optionally, the step 400 includes:
inputting the test set into the one-dimensional convolutional neural network model, automatically extracting the bearing state characteristics by the model, and directly outputting the fault diagnosis result;
wherein the fault diagnosis result comprises at least one of: fault status, fault location and fault severity of the bearing.
In this embodiment, the bearing fault data to be diagnosed, that is, the test set, is directly input into the trained one-dimensional convolutional neural network model, the three predicted values are sequentially output in the flowing process of the model, which respectively represent the bearing state, the fault position, and the severity of the fault, and the bearing state can be diagnosed according to the output of the model.
Optionally, after the state features of the bearing are automatically extracted according to the trained one-dimensional convolutional neural network model and the fault diagnosis result of the bearing is directly obtained, the method further includes:
matching the adaptive strategy signals from a preset table according to the fault diagnosis result; wherein the preset table includes: at least one fault diagnosis result, and a strategy signal corresponding to the fault diagnosis result.
In this embodiment, the fault diagnosis result includes a fault state, a fault location, and a fault severity, and the embodiment of the present 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, and the like.
In summary, the method for diagnosing the bearing fault based on the original one-dimensional data provided by the application constructs the one-dimensional convolutional neural network model with the branch structure, and directly inputs the one-dimensional bearing data as the model, so that the effective characteristics can be directly extracted from the collected one-dimensional data without any data preprocessing operation, the intelligent diagnosis of the bearing fault based on the original one-dimensional data is realized, the accuracy is high, and the operation is simple.
As shown in fig. 4, an embodiment of the present application further provides a bearing fault diagnosis device, including:
the acquisition module 10 is used for acquiring bearing fault data when a fault bearing operates under a preset working condition and a sampling frequency;
a building module 20, configured to build a one-dimensional convolutional neural network model with a branch structure;
the processing module 30 is configured to train and adjust the one-dimensional convolutional neural network model according to the bearing fault data;
and the obtaining module 40 is used for automatically extracting the bearing state characteristics through the trained one-dimensional convolutional neural network model and obtaining the fault diagnosis result of the bearing.
Optionally, the acquisition module 10 includes:
the acquisition unit is used for adopting a preset bearing data set;
and the dividing unit is used for dividing the bearing data set into a training set, a verification set and a test set according to a preset dividing proportion.
Optionally, the acquisition module 10 further includes:
processing a plurality of bearings with different damage positions and different size faults;
the second acquisition unit is used for operating the bearing on the experiment table and acquiring a vibration acceleration signal corresponding to each rolling bearing;
and the storage unit is used for storing the fault position and the size information of each rolling bearing and the corresponding vibration acceleration signal data thereof and establishing bearing fault data.
Optionally, the building module 20 includes:
the device comprises a first construction unit, a second construction unit and a third construction unit, wherein the first construction unit is used for constructing a one-dimensional convolutional neural network basic model which can extract at least three pieces of characteristic information;
the second construction unit is used for constructing a layered structure of the bearing fault data, wherein the layered structure comprises a fault state layer, a fault position layer and a fault severity layer;
and the third construction unit is used for fusing the hierarchical structure of the bearing fault data and the one-dimensional convolutional neural network basic model based on the characteristic of natural layering of a convolutional neural network model to construct the one-dimensional convolutional neural network model with the branch structure.
Optionally, the first building unit includes:
the determining unit is used for determining the size of a first convolution kernel of the structural parameters of the one-dimensional convolution neural network basic model as a first preset value;
the second determining unit is used for determining the sizes of the rest convolution kernels of the structural parameters of the one-dimensional convolution neural network basic model to be second preset values;
the third determining unit is used for determining the sizes of a plurality of pool layers of the one-dimensional convolutional neural network basic model as a third preset value;
wherein the first preset value is greater than the second preset value.
Optionally, the processing module 30 includes:
the training unit is used for inputting the training set into the one-dimensional convolutional neural network model for training to obtain a loss value of the one-dimensional convolutional neural network model;
and the updating unit is used for optimizing the one-dimensional convolutional neural network model according to the loss value of the model so as to update the one-dimensional convolutional neural network model.
Optionally, the processing module 30 further includes:
inputting the verification set into the trained one-dimensional convolutional neural network model for verification to obtain an evaluation index and a performance index of the one-dimensional convolutional neural network model;
judging whether the evaluation index and the performance index are smaller than a preset threshold value or not;
and if the evaluation index and the performance index are smaller than a preset threshold value, adjusting the one-dimensional convolutional neural network model, otherwise, storing the trained one-dimensional convolutional neural network model.
Optionally, the obtaining module 40 includes:
the output unit is used for inputting the test set into the one-dimensional convolutional neural network model, and the model automatically extracts the bearing state characteristics and outputs the fault diagnosis result;
wherein the fault diagnosis result comprises at least one of: fault status, fault location and fault severity of the bearing.
Optionally, the apparatus further comprises:
the matching module is used for matching the adaptive strategy signals from a preset table according to the fault diagnosis result; wherein the preset table includes: at least one fault diagnosis result, and a strategy signal corresponding to the fault diagnosis 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 (12)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110077986.2A CN112729834B (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 |
|---|---|---|---|
| CN202110077986.2A CN112729834B (en) | 2021-01-20 | 2021-01-20 | Bearing fault diagnosis method, device and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112729834A true CN112729834A (en) | 2021-04-30 |
| CN112729834B CN112729834B (en) | 2022-05-10 |
Family
ID=75594374
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110077986.2A Active CN112729834B (en) | 2021-01-20 | 2021-01-20 | Bearing fault diagnosis method, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112729834B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114330439A (en) * | 2021-12-28 | 2022-04-12 | 盐城工学院 | Bearing diagnosis method based on convolutional neural network |
| CN114970044A (en) * | 2022-06-20 | 2022-08-30 | 华北电力大学 | Rolling bearing fault diagnosis method and system based on threshold convolution neural network |
| CN115017945A (en) * | 2022-05-24 | 2022-09-06 | 南京林业大学 | Mechanical fault diagnosis method and system based on enhanced convolutional neural network |
| CN115493844A (en) * | 2022-09-23 | 2022-12-20 | 华北电力科学研究院有限责任公司 | Steam flow excitation processing method and device for steam turbine generator unit |
| CN117686225A (en) * | 2024-02-02 | 2024-03-12 | 浙江大学 | Permanent magnet synchronous motor bearing fault degree diagnosis method and system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160117587A1 (en) * | 2014-10-27 | 2016-04-28 | Zhicheng Yan | Hierarchical deep convolutional neural network for image classification |
| 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 |
| CN109886604A (en) * | 2019-03-13 | 2019-06-14 | 华北电力大学 | A Transient Stability Evaluation Method of Power System Based on One-dimensional Convolutional Neural Network |
| 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 |
| CN111626361A (en) * | 2020-05-28 | 2020-09-04 | 辽宁大学 | Bearing sub-health identification method for improving capsule network optimization layered convolution |
| 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 CN202110077986.2A patent/CN112729834B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160117587A1 (en) * | 2014-10-27 | 2016-04-28 | Zhicheng Yan | Hierarchical deep convolutional neural network for image classification |
| 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 |
| CN109886604A (en) * | 2019-03-13 | 2019-06-14 | 华北电力大学 | A Transient Stability Evaluation Method of Power System Based on One-dimensional Convolutional Neural Network |
| 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 |
| CN111626361A (en) * | 2020-05-28 | 2020-09-04 | 辽宁大学 | Bearing sub-health identification method for improving capsule network optimization layered convolution |
Non-Patent Citations (2)
| Title |
|---|
| XIAOJIE GUO 等: "Hierarchical adaptive deep convolution neural network", 《MEASUREMENT》 * |
| 何成兵: "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", 《中国电机工程学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114330439A (en) * | 2021-12-28 | 2022-04-12 | 盐城工学院 | Bearing diagnosis method based on convolutional neural network |
| CN114330439B (en) * | 2021-12-28 | 2023-04-07 | 盐城工学院 | Bearing diagnosis method based on convolutional neural network |
| CN115017945A (en) * | 2022-05-24 | 2022-09-06 | 南京林业大学 | Mechanical fault diagnosis method and system based on enhanced convolutional neural network |
| CN114970044A (en) * | 2022-06-20 | 2022-08-30 | 华北电力大学 | Rolling bearing fault diagnosis method and system based on threshold convolution neural network |
| CN115493844A (en) * | 2022-09-23 | 2022-12-20 | 华北电力科学研究院有限责任公司 | Steam flow excitation processing method and device for steam turbine generator unit |
| CN117686225A (en) * | 2024-02-02 | 2024-03-12 | 浙江大学 | Permanent magnet synchronous motor bearing fault degree diagnosis method and system |
| CN117686225B (en) * | 2024-02-02 | 2024-04-12 | 浙江大学 | Permanent magnet synchronous motor bearing fault degree diagnosis method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112729834B (en) | 2022-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112729834A (en) | Bearing fault diagnosis method, device and system | |
| CN111046945B (en) | Fault type and damage degree diagnosis method based on combined convolutional neural network | |
| CN112257530B (en) | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine | |
| CN112067916A (en) | Intelligent fault diagnosis method for time series data based on deep learning | |
| CN110598851A (en) | Time series data abnormity detection method fusing LSTM and GAN | |
| CN109782603A (en) | Detection method and monitoring system of rotating machinery coupling fault | |
| CN110132554B (en) | A deep Laplacian self-encoding method for fault diagnosis of rotating machinery | |
| WO2024065777A1 (en) | Method, apparatus, electronic device, and storage medium for diagnosing industrial fault | |
| CN110823576A (en) | Generative Adversarial Network-Based Mechanical Anomaly Detection Method | |
| CN113485302A (en) | Vehicle operation process fault diagnosis method and system based on multivariate time sequence data | |
| CN113408068A (en) | Random forest classification machine pump fault diagnosis method and device | |
| CN109932174A (en) | A fault diagnosis method for gearboxes based on multi-task deep learning | |
| CN111680788A (en) | Equipment fault diagnosis method based on deep learning | |
| CN115859077A (en) | Multi-feature fusion motor small sample fault diagnosis method under variable working conditions | |
| CN111562110A (en) | Fault diagnosis model and cross-component fault diagnosis method based on convolutional neural network | |
| CN115374811A (en) | Novel fault state diagnosis method for rolling bearing | |
| CN112699597A (en) | Nuclear power starting water pump rolling bearing fault detection method and system | |
| CN116361723A (en) | A classification method for bearing fault diagnosis based on multi-scale features and attention | |
| CN115221973A (en) | Aviation bearing fault diagnosis method based on enhanced weighted heterogeneous ensemble learning | |
| CN116625686A (en) | On-line diagnosis method for bearing faults of aero-engine | |
| Saha et al. | Enhancing bearing fault diagnosis using transfer learning and random forest classification: A comparative study on variable working conditions | |
| CN114491823B (en) | A fault diagnosis method for train bearings based on improved generative adversarial network | |
| CN116070134A (en) | Intelligent equipment fault diagnosis method and system based on prototype learning | |
| CN115345255A (en) | Fault diagnosis method, control device, terminal and storage medium | |
| CN114818811A (en) | Aircraft engine rolling bearing fault diagnosis method based on twin network metric learning |
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 |