Rotating machinery fault diagnosis method based on multi-view spatio-temporal graph attention network

Abstract

The application discloses a rotating machinery fault diagnosis method based on a multi-view spatio-temporal graph attention network, and belongs to the technical field of rotating machinery intelligent monitoring and fault diagnosis. The multi-view spatio-temporal graph attention network sequentially comprises a topological graph layer, a double-channel parallel graph attention network, a multi-view feature fusion module, a gated recurrent unit time feature extraction module, a full connection layer and a Softmax function layer. When training, firstly, a distance topological graph and a similarity topological graph of input data are constructed, spatial feature extraction is performed from two view angles of geometric distance and feature similarity through the double-channel parallel graph attention network, dynamic fusion of double-channel features is realized through a learnable attention weight matrix through the multi-view feature fusion module, and finally, the fused spatial features are input into the gated recurrent unit network, so that fault feature information in the time dimension is deeply mined, and the accuracy and robustness of rotating machinery fault diagnosis can be effectively improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent monitoring and fault diagnosis of rotating machinery, and particularly relates to a rotating machinery fault diagnosis method based on a multi-view spatiotemporal graph attention network. BACKGROUND

[0002] With the continuous rise of the intelligent degree of industrial equipment, the state monitoring of key components of rotating machinery is facing new technical challenges and requirements. As the core components of the transmission system, gears and bearings bear complex loads and friction during long-term operation and are prone to damage. Typical fault modes include but are not limited to gear tooth peeling, gear root cracking, rolling element pitting, inner and outer ring wear, etc. These mechanical damage problems not only significantly reduce the operating efficiency of mechanical equipment, but also greatly reduce the stability of equipment operation, and in severe cases, can even cause unexpected production interruptions, thereby causing huge economic losses to enterprises.

[0003] In actual industrial environments, traditional rotating machinery fault diagnosis methods are based on the analysis of vibration signals collected by sensors. Due to signal attenuation caused by transmission path effects, signal-to-noise ratio degradation caused by environmental noise, and feature aliasing caused by multi-source vibration coupling, traditional diagnosis methods have obvious shortcomings: on the one hand, signal analysis methods based on fast Fourier transform have limited sensitivity to minor damage; on the other hand, diagnosis systems relying on expert experience cannot cope with feature changes under different operating conditions. In addition, due to the explosive growth of data volume, traditional signal analysis-based diagnosis methods have been unable to meet the application requirements of rapid diagnosis.

[0004] Under the background of big data analysis, intelligent diagnosis methods represented by convolutional neural networks (CNN) have been widely applied. By constructing an end-to-end fault diagnosis model, not only can the dependence on professional knowledge be reduced, but also deep features in monitoring data can be fully mined. However, existing related network models mostly analyze and process data itself, and do not consider the interaction between multiple data, so the fault detection accuracy needs to be improved. SUMMARY

[0005] The present application is to solve the above technical problems existing in the prior art, and proposes a mechanical fault diagnosis method based on a multi-view spatiotemporal graph attention network.

[0006] The technical solution of the present application is a mechanical fault diagnosis method based on a multi-view spatiotemporal graph attention network model. The collected mechanical vibration signals to be tested are input into the network model for fault diagnosis. The network model is a multi-view spatiotemporal graph attention network model, which is constructed according to the following steps:

[0007] Step 1. Collecting vibration signals containing normal state and typical mechanical fault modes in rotation;

[0008] Step 2. Preprocessing the vibration signals, dividing the training data set and the validation data set;

[0009] Step 3. Establishing a multi-view spatio-temporal graph attention network and setting network training conditions, the multi-view spatio-temporal graph attention network is sequentially provided with a topology graph construction layer, a double-channel parallel graph attention network, a multi-view feature fusion module, a gated recurrent unit time feature extraction module, a full connection layer and a Softmax function layer from input to output, the topology graph construction layer regards each fault data sample as a node in the topology graph, and distance topology graph and similarity topology graph are constructed respectively; the double channels of the double-channel parallel graph attention network have the same architecture but are not parameter shared, the constructed distance topology graph and similarity topology graph are input into one channel of the double-channel parallel graph attention network respectively, and distance view features and similarity view features are obtained respectively; the multi-view feature fusion module dynamically fuses the distance view features and the similarity view features to obtain fusion features; the gated recurrent unit time feature extraction module further extracts features in the time dimension from the obtained fusion features to obtain time dimension features;

[0010] Step 4. Training the multi-view spatio-temporal graph attention network using the training data set and the validation data set to obtain a multi-view spatio-temporal graph attention network model.

[0011] Preferably, the step 2 adopts a sliding window method to segment the vibration signals into equal lengths and performs data normalization processing to form various types of data samples, and then randomly divides the training set and the validation set according to the stratified random sampling principle.

[0012] Preferably, the step 4 is specifically as follows:

[0013] Step 4.1. Through the topology graph construction layer, each fault data sample is regarded as a node in the topology graph, and distance topology graph and similarity topology graph are constructed respectively, and the specific operation is as follows:

[0014] Step 4.1.1. By measuring the Euclidean distance between any two nodes in the physical space, the Euclidean distance matrix between all nodes in this view is obtained ;

[0015] ;

[0016] ;

[0017] wherein, NN represents the number of nodes used to construct the graph structure, is the Euclidean distance between any two nodes , ;

[0018] Step 4.1.2 KNN sparsification is performed on the Euclidean distance matrix , and the top x i shortest distance connections of each node k are retained to form a distance adjacency matrix and a distance topology graph is established X , ;

[0019] ;

[0020] ;

[0021] ;

[0022] wherein, represents the distance metric neighborhood of node , denotes the edge connection weight between any two nodes based on distance metric, is the kernel function width;

[0023] Step 4.1.3 The similarity matrix of all nodes in this perspective is obtained by measuring the cosine similarity between node features ;

[0024] ;

[0025] ;

[0026] wherein, denotes the cosine value of the inner product angle between two nodes , ;

[0027] Step 4.1.4 For each node , the KNN method is used to find its k nearest neighbors in the perspective of similarity, and a similarity adjacency matrix is obtained and a similarity topology graph is constructed X , ;

[0028] ;

[0029] ;

[0030] ;

[0031] wherein, represent nodes of the similarity measure neighborhood, denotes the similarity adjacency matrix;

[0032] Step 4.2 extracts the deep representation of nodes in different metric spaces through a dual-channel parallel graph attention network, and the specific operation is as follows:

[0033] Step 4.2.1 initializes the dual-channel parallel graph attention network, and the distance topology graph (X, ) and the similarity topology graph (X, ) are input into one channel of the dual-channel parallel graph attention network respectively;

[0034] Step 4.2.2: calculate and use Softmax function to normalize to obtain the attention score of each edge in the adjacency matrix;

[0035] ;

[0036] wherein, || represents node feature splicing operation, W is a shared weight matrix, denotes a scoring function, LeakyReLU denotes an activation function, denotes the neighborhood nodes connected with node in the graph;

[0037] Step 4.2.3 aggregates the node features based on the attention score of each edge in the adjacency matrix, and obtains the distance perspective feature and the similarity perspective feature through graph attention convolution operation respectively;

[0038] Step 4.3 performs splicing operation on the distance perspective feature and the similarity perspective feature along the feature dimension through a multi-perspective feature fusion module to generate joint features :

[0039] ;

[0040] Self-attention fusion is performed on the joint features to learn the interaction importance of different feature dimensions to obtain the fusion features ;

[0041] ;

[0042] wherein, is a projection matrix, d is the dimension of the attention space;

[0043] Step 4.4 inputs the fused feature into a gated recurrent unit time feature extraction module to further extract features in the time dimension, and obtains time dimension features .

[0044] Step 4.5 inputs the time dimension features into a fully connected layer and processes them through a Softmax function layer to obtain an output result output .

[0045] .

[0046] wherein, respectively represent the weight and bias of the fully connected layer;

[0047] Step 4.6 calculates the cross-entropy loss of the output result output using a cross-entropy loss function, and determines whether the network has not decreased in cross-entropy loss for consecutive iterations or has reached the maximum number of iterations, and if not, returns to Step 4.1 to continue training; if yes, outputs the trained network model, i.e., the multi-view spatio-temporal graph attention network model.

[0048] The present application constructs a multi-view spatio-temporal graph attention network from input to output, which sequentially includes a topology graph layer, a double-channel parallel graph attention network (GAT), a multi-view feature fusion module, a gated recurrent unit time feature extraction module (GRU), a fully connected layer and a Softmax function layer. During network training, first, distance topology graphs and similarity topology graphs of input data are constructed from different measurement perspectives, and then spatial features are extracted from two perspectives of geometric distance and feature similarity through the double-channel parallel graph attention network; then, through the multi-view feature fusion module, dynamic fusion of double-channel features is realized through a learnable attention weight matrix; finally, the fused spatial features are input into the gated recurrent unit network to deeply mine fault feature information in the time dimension.

[0049] Compared with the prior art, the present application has the following advantages:

[0050] (1) By generating distance topology graphs and similarity topology graphs respectively, the limitations of traditional single-view analysis are broken through, so that the network model can comprehensively capture fault feature information from different measurement perspectives such as geometric distance and feature similarity, thereby improving the completeness and accuracy of fault information representation and laying a solid foundation for accurate fault diagnosis;

[0051] (2) The combination of the dual-channel parallel graph attention network (GAT) and the self-attention fusion module can accurately adjust the weight contribution of each visual feature according to the inherent characteristics and mutual correlation of different visual features through an adaptive learning process, thereby improving the adaptability of the fault diagnosis model to different data distributions and fault modes;

[0052] (3) By cascading the gated recurrent unit time feature extraction module (GRU) and the dual-channel graph attention network (GAT), the spatial correlation features under different perspectives are deeply mined from the topological structure by means of the powerful spatial feature extraction capability of the graph GAT, and the time sequence dependence relationship in the fault development process is effectively captured by utilizing the advantage of the GRU in time sequence data processing, thereby realizing the all-around and deep representation of the fault information from the spatial and temporal dimensions, improving the extraction and analysis capability of the model for the fault features, and improving the accuracy and robustness of the rotating machinery fault diagnosis. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 A flowchart for constructing a network model of the embodiment of the present application.

[0054] Figure 2 A structural schematic diagram of the network model of the embodiment of the present application.

[0055] Figure 3 A vibration data waveform diagram under different operating conditions of the embodiment of the present application.

[0056] Figure 4 A confusion matrix of the rotating machinery fault diagnosis result of the embodiment of the present application. DETAILED DESCRIPTION

[0057] The mechanical fault diagnosis method based on the multi-perspective spatio-temporal graph attention network model of the present application is to input the collected vibration signal of the mechanical equipment to be tested into a network model for fault diagnosis. The network model is a multi-perspective spatio-temporal graph attention network model, as shown in Figure 1 The network model is constructed according to the following steps:

[0058] Step 1. Collect the vibration signals of the rotating machinery under normal state and typical fault modes.

[0059] The rotating machinery vibration test bench used in the embodiment of the present application is composed of a driving motor, a transmission belt, a transmission shaft system, a load applicator, and a rolling bearing and the like basic components, an acceleration sensor is arranged directly above the rolling bearing seat, the sampling frequency is set to 12800 Hz, when the equipment is running to the rated speed / load working condition, the vibration signals containing normal state and typical fault modes are collected; the specific typical fault mode is to manufacture crack defects on the outer ring, rolling elements and inner ring of the cylindrical roller bearing of model N205EM, the bearing has a total of 13 rolling elements, the diameter of the rolling element is 7.5 mm , the pitch diameter of the bearing is 38.5 mm , and the contact angle is 0°;

[0060] Step 2. Preprocess the vibration signal, and divide the training data set and the verification data set;

[0061] The vibration signal is segmented by the sliding window method and normalized, and various types of data samples are formed as shown in Figure 3 , then the training set and the verification set are randomly divided according to the stratified random sampling principle, and the distribution of each type of fault data sample is balanced; Figure 3 The data samples from top to bottom in the figure are normal state (no fault), bearing outer ring fault, bearing rolling element fault and bearing inner ring fault, and the present embodiment divides the data in a ratio of 8:2, that is, 80% of the data is randomly selected as the training data set, and the remaining 20% of the data is used as the verification data set;

[0062] Step 3. Establish a multi-view spatio-temporal graph attention network and set the network training conditions (set the model training times, the optimizer, the initial learning rate and the like hyperparameters), the multi-view spatio-temporal graph attention network is provided with a topology graph construction layer, a double-channel parallel graph attention network (GAT), a multi-view feature fusion module, a single-layer gated recurrent unit time feature extraction module (GRU), a 2-layer full connection layer and a Softmax function layer from input to output as shown in Figure 2 , the topology graph construction layer regards each fault data sample as a node in the topology graph, and distance topology graph and similarity topology graph are constructed respectively; the double channels of the double-channel parallel graph attention network have the same architecture but are not parameter shared, the constructed distance topology graph and similarity topology graph are input into one channel of the double-channel parallel graph attention network respectively, and distance view features and similarity view features are obtained respectively; the multi-view feature fusion module fuses the distance view features and the similarity view features to obtain fusion features; the gated recurrent unit time feature extraction module further extracts features in the time dimension from the obtained fusion features to obtain time dimension features;

[0063] Step 4. Train the multi-view spatio-temporal graph attention network using the training dataset and the validation dataset to obtain a multi-view spatio-temporal graph attention network model;

[0064] Step 4.1 For each fault type, randomly select 10 fault data samples, and for each fault data sample, construct a distance topology graph and a similarity topology graph by constructing a topology graph layer, as follows:

[0065] Step 4.1.1 Obtain the Euclidean distance matrix between all nodes in this view by measuring the Euclidean distance between any two nodes in the physical space ;

[0066] ;

[0067] ;

[0068] wherein, N represents the number of nodes used to construct the graph structure, is the Euclidean distance between any two nodes , ;

[0069] Step 4.1.2 Perform KNN sparsification on the Euclidean distance matrix , and retain the first shortest distance connections of each node k to form a distance adjacency matrix and establish a distance topology graph X , ;

[0070] ;

[0071] ;

[0072] ;

[0073] wherein, represents the distance measurement neighborhood of the node , represents the edge connection weight between any two nodes based on distance measurement, is the width of the heat kernel function;

[0074] Step 4.1.3 Obtain the similarity matrix of all nodes in this view by measuring the cosine similarity between node features ;

[0075] ;

[0076] ;

[0077] wherein, denotes the cosine value of the inner product of vectors between two nodes , ;

[0078] Step 4.1.4, for each node , find its k nearest neighbors in the perspective of similarity using the KNN method, obtain a similarity adjacency matrix and construct a similarity topology graph X , ;

[0079] ;

[0080] ;

[0081] ;

[0082] wherein, represents the similarity metric neighborhood of node , denotes the similarity adjacency matrix;

[0083] Step 4.2, by a double-channel parallel graph attention network, extract the deep representation of nodes in different metric spaces, the specific operation is as follows:

[0084] Step 4.2.1, initialize the double-channel parallel graph attention network, input the distance topology graph (X, ) and the similarity topology graph (X, ) into one channel of the double-channel parallel graph attention network respectively;

[0085] Step 4.2.2: calculate and use Softmax function for normalization processing, to obtain the attention score of each edge in the adjacency matrix;

[0086] ;

[0087] wherein, || represents node feature concatenation operation, W is a shared weight matrix, denotes a scoring function, LeakyReLU denotes an activation function, denotes the neighborhood nodes connected with node in the graph;

[0088] Step 4.2.3, based on the attention score of each edge in the adjacency matrix, the aggregation of node features is carried out, and the distance perspective feature and the similarity perspective feature are obtained through the graph attention convolution operation, respectively and similarity perspective features ;

[0089] Step 4.3, through the multi-perspective feature fusion module, the distance perspective feature and similarity perspective feature are spliced along the feature dimension to generate joint features :

[0090] ;

[0091] The joint features are self-attention fused to learn the interaction importance of different feature dimensions to obtain fused features ;

[0092] ;

[0093] wherein, is a projection matrix, d is the dimension of the attention space;

[0094] Step 4.4, the fused features are input into a single-layer gated recurrent unit time feature extraction module to further extract features in the time dimension to obtain time dimension features ;

[0095] Step 4.5, the time dimension features are input into a fully connected layer and processed by a Softmax function layer after reshaping to obtain an output result output ;

[0096] ;

[0097] wherein, respectively represent the weight and bias of the fully connected layer;

[0098] Step 4.6, the cross-entropy loss of the output result output is calculated by using a cross-entropy loss function, whether the network is continuously iterated for 10 times without the cross-entropy loss decreasing or reaching the maximum iteration number is judged, if not, the step 4.1 is returned to continue training; if yes, the trained network model, i.e., the multi-perspective spatio-temporal graph attention network model, is output.

[0099] The final diagnosis result confusion matrix of the embodiment of the application is shown in Figure 4 . Figure 4In the figure, 1 is normal state (no fault), 2 is bearing outer ring fault, 3 is bearing rolling element fault, and 4 is bearing inner ring fault. Figure 4 It can be seen that, in the diagnosis result of the embodiment of the application, all data are correctly classified except that one rolling element fault is incorrectly identified as inner ring fault, the overall accuracy of the diagnosis model reaches 99.75%, and the effectiveness of the method proposed in the application for rotating machinery fault diagnosis is verified.

Claims

1. A mechanical fault diagnosis method based on a multi-view spatiotemporal graph attention network model, wherein the collected mechanical vibration signal to be tested is input into the network model for fault diagnosis, characterized in that: The network model is a multi-view spatio-temporal graph attention network model, which is constructed according to the following steps: Step 1. Collecting vibration signals of rotating machinery including normal state and typical fault modes; Step 2. Preprocessing the vibration signals, dividing the training data set and the verification data set; Step 3. Establishing a multi-view spatio-temporal graph attention network and setting network training conditions, the multi-view spatio-temporal graph attention network is sequentially provided with a topology graph layer, a double-channel parallel graph attention network, a multi-view feature fusion module, a gated recurrent unit time feature extraction module, a full connection layer and a Softmax function layer from input to output, the topology graph layer is to regard each fault data sample as a node in the topology graph, and distance topology graph and similarity topology graph are constructed respectively; the double-channel parallel graph attention network has the same architecture but non-parameter sharing, the constructed distance topology graph and similarity topology graph are input into one channel of the double-channel parallel graph attention network respectively, and distance view features and similarity view features are obtained respectively; the multi-view feature fusion module dynamically fuses the distance view features and the similarity view features to obtain fusion features; the gated recurrent unit time feature extraction module further extracts features in the time dimension from the obtained fusion features to obtain time dimension features; Step 4. Training the multi-view spatio-temporal graph attention network using the training data set and the verification data set to obtain the multi-view spatio-temporal graph attention network model.

2. The mechanical fault diagnosis method of the multi-view spatio-temporal graph attention network model according to claim 1, characterized in that The step 2 is to segment the vibration signals into equal length by using the sliding window method and to perform data normalization processing to form various types of data samples, and then to randomly divide the training set and the verification set according to the stratified random sampling principle.

3. The mechanical fault diagnosis method of the multi-view spatio-temporal graph attention network model according to claim 2, characterized in that The step 4 is specifically as follows: Step 4.1, by constructing a topology graph layer, regarding each fault data sample as a node in the topology graph, respectively constructing a distance topology graph and a similarity topology graph, the specific operation is as follows: Step 4.1.1 Obtain the Euclidean distance matrix between all nodes in this view by measuring the Euclidean distance between any two nodes in the physical space ; ； ； wherein, N represents the number of nodes used to construct the graph structure, is the Euclidean distance between any two nodes , . Step 4.1.2 Euclidean distance matrix Perform KNN sparsification, keep the top K shortest distance connections of each node k to form a distance adjacency matrix and build a distance topology graph X , ​ ； ； ； wherein, representative node distance metric neighborhood, denotes the edge connection weight between any two nodes based on the distance metric, is the kernel width; Step 4.1.3 Obtain the similarity matrix of all nodes in this view by measuring the cosine similarity between node features ; ； ； wherein represents the cosine of the angle between the inner product of the vectors of two nodes , . Step 4.1.4 For each node , find its k nearest neighbors by KNN method in the view of similarity , get the similarity adjacency matrix X , , and construct the similarity topology graph ； ； ； wherein, representative nodes of the similarity measure neighborhood, denotes the similarity adjacency matrix; Step 4.2, by the double-channel parallel graph attention network, extracting deep representations of nodes in different metric spaces, the specific operation is as follows: Step 4.2.1 initializes a dual-channel parallel graph attention network, inputs the distance topology graph (X, ) and the similarity topology graph (X, ) into one channel of the dual-channel parallel graph attention network respectively; Step 4.2.2: Calculate according to the following formula Use Softmax The function is normalized to obtain the attention score for each edge in the adjacency matrix; ； wherein || denotes a node feature concatenation operation, W is a shared weight matrix, denotes a scoring function, LeakyReLU denotes an activation function, denotes a neighborhood node connected to the node in the graph. Step 4.2.

3. The aggregation of node features based on the attention score of each edge in the adjacency matrix, through the graph attention convolution operation, respectively obtains the distance perspective feature and the similarity perspective feature ; Step 4.3, the multi-view feature fusion module performs a stitching operation on the distance view feature and the similarity view feature along the feature dimension to generate a joint feature . . ​ ； Joint features Self-attention fusion is performed to learn the interaction importance of different feature dimensions to obtain fused features ; ； wherein, is a projection matrix, d is the dimension of the attention space; Step 4.4 fuse features Input to the gated recurrent unit time feature extraction module, further perform feature extraction in the time dimension, obtain time dimension features ; Step 4.5: Time dimension feature Input to the fully connected layer and pass through Softmax After the function layer processing, the output result is obtained output ; ； wherein, respectively denote the weights and biases of the fully connected layer. Step 4.6 calculates the output result by using a cross-entropy loss function output If the cross-entropy loss of the network is not decreased for several times or reaches the maximum iteration number, the method returns to step 4.1 to continue training; otherwise, the trained network model, i.e., the multi-view spatio-temporal graph attention network model, is output.

Images