A method, device, equipment and storage medium for monitoring the operation status of a wheel hub bearing

Abstract

The present invention discloses a method, device, equipment and storage medium for monitoring the operation status of a hub bearing, and relates to the technical field of bearing operation status monitoring, and includes the following steps: real-time collection of vibration signals and temperature signals during operation of the hub bearing, extraction of features of the vibration signals and temperature signals, and obtaining a plurality of second features; defining the values ​​corresponding to the plurality of second features at the current moment as first nodes, defining the interaction between the first nodes as a first edge, and constructing first graph data; defining the values ​​corresponding to each second feature at a plurality of moments as second nodes, defining the interaction between the second nodes as a second edge, and constructing second graph data; processing the first graph data and the plurality of second graph data, and monitoring the operation status of the hub bearing according to the processing results. The present invention improves the monitoring sensitivity for different types of faults, and improves the accuracy of online monitoring of hub bearings in practical applications.

Description

Technical Field

[0001] The present invention relates to the technical field of bearing operation status monitoring, and in particular to a method, device, equipment and storage medium for monitoring the operation status of a hub bearing. Background Art

[0002] As a key component for wheel support and rotation, wheel hub bearings are subjected to heavy loads and frequent impacts for a long time, and there is a high risk of failure. Therefore, intelligent monitoring of wheel hub bearings plays an important role in improving wheel performance and reliability.

[0003] The intelligent monitoring of the wheel hub bearing operating status is achieved by integrating various signal acquisition devices on the wheel hub, analyzing the collected signals through signal processing and machine learning technologies, and thus obtaining the wheel hub bearing operating status. Compared with the traditional wheel hub bearing fault monitoring method that relies on manual inspection and regular maintenance, the wheel hub bearing operating status intelligent monitoring system has the advantages of real-time monitoring of the wheel hub bearing working status and timely warning.

[0004] However, the failure modes of wheel hub bearings are diverse, including lubrication failure and mechanical damage. When the wheel hub bearing fails to lubricate, its temperature will increase significantly, and the wheel hub bearing temperature signal can promptly reflect the problem of wheel hub bearing lubrication failure. When the wheel hub bearing is mechanically damaged, its vibration will increase significantly, and the vibration signal is more sensitive to mechanical damage of the wheel hub bearing.

[0005] When collecting fault data, the existing intelligent monitoring methods for wheel hub bearings mostly collect temperature signals or vibration signals, so they are prone to reduced monitoring sensitivity when monitoring different types of faults. At the same time, in the specific data processing process, due to the lengthening of the signal transmission path, the correlation of the collected fault data to the change of the wheel hub bearing state is reduced. In the process of data processing of multiple signals, the existing monitoring methods often only consider the changes of the signal itself, resulting in inaccurate final operating status monitoring results. Summary of the invention

[0006] The present invention provides a wheel hub bearing operating status monitoring method, device, equipment and storage medium, which solves the problem that the existing monitoring method often only considers its own changes during data processing of multiple signals, resulting in inaccurate final operating status monitoring results.

[0007] In a first aspect, the present invention provides a method for monitoring the operating status of a hub bearing, comprising the following steps:

[0008] Real-time collection of vibration and temperature signals of wheel hub bearings during operation;

[0009] Extracting the features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing, and superimposing the corresponding features of the temperature signal with the plurality of first features to obtain a plurality of second features;

[0010] The values ​​corresponding to the multiple second features at the current moment are defined as first nodes, and the interactions between the first nodes are defined as first edges, so as to construct the first graph data; the values ​​corresponding to each second feature at different moments are defined as second nodes, and the interactions between the second nodes are defined as second edges, so as to construct the second graph data;

[0011] The first image data and a plurality of second image data are predicted, reconstructed and superimposed to obtain a processing result, and the operating state of the wheel hub bearing is monitored according to the processing result.

[0012] Preferably, the interaction between the first nodes is defined as a first edge, as shown below:

[0013] ;

[0014] In the formula, and Represents the first node a and the first node b , It means taking the maximum value of the two first nodes. Represents the first edge of two first nodes;

[0015] The interaction between the second nodes is defined as the second edge, as follows:

[0016] ;

[0017] In the formula, and Respectively represent the second node c and the second node d , It means taking the maximum value of the two second nodes. Represents the second edge between two second nodes.

[0018] Preferably, the predicting, reconstructing and superimposing the first graph data and the plurality of second graph data to obtain a processing result comprises the following steps:

[0019] Perform feature extraction on the first image data and the plurality of second image data through a time attention layer and a feature attention layer to obtain a plurality of time features and a plurality of space features;

[0020] The first graph data, the plurality of second graph data, the plurality of time features and the plurality of spatial features are fused through a fusion layer to obtain a fusion feature;

[0021] The fused features are processed through the prediction layer to obtain the predicted value; the fused features are reconstructed through the reconstruction layer to obtain the reconstructed value;

[0022] The predicted value and the reconstructed value are superimposed to obtain the score value, which is the processing result.

[0023] Preferably, the monitoring of the operating state of the hub bearing according to the processing result comprises the following steps:

[0024] comparing the score value to a set threshold;

[0025] When the score value is greater than the set threshold, it is determined that the wheel hub bearing operating state is abnormal; when the score value is less than the set threshold, it is determined that the wheel hub bearing operating state is normal.

[0026] Preferably, the step of extracting the features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing comprises the following steps:

[0027] Extract the features of the vibration signal to obtain multiple time domain features and frequency domain features;

[0028] A plurality of time domain features and frequency domain features are optimized through a monotonicity index and a correlation index to obtain a plurality of first features for characterizing the failure trend of the hub bearing.

[0029] Preferably, the real-time collection of vibration signals and temperature signals of the hub bearing during operation specifically includes the following steps:

[0030] The vibration sensor is used to collect the vibration signal of the wheel hub bearing in real time during operation;

[0031] The temperature signal of the wheel hub bearing during operation is collected in real time by the temperature sensor, and the corresponding characteristics of the temperature signal include the maximum value of the temperature signal and the minimum value of the temperature signal.

[0032] In a second aspect, the present invention further provides a hub bearing operating status monitoring device, comprising:

[0033] The acquisition module is used to collect the vibration signal and temperature signal of the wheel hub bearing in real time during operation;

[0034] An extraction module is used to extract features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing, and to superimpose corresponding features of the temperature signal with the plurality of first features to obtain a plurality of second features;

[0035] A construction module, used to define the values ​​corresponding to the plurality of second features at the current moment as first nodes, define the interactions between the first nodes as first edges, and construct first graph data; define the values ​​corresponding to each second feature at different moments as second nodes, define the interactions between the second nodes as second edges, and construct second graph data;

[0036] The processing module is used to predict, reconstruct and superimpose the first image data and multiple second image data to obtain processing results, and monitor the operating status of the wheel hub bearing according to the processing results.

[0037] In a third aspect, the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for monitoring the operating status of a hub bearing when executing the program.

[0038] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method for monitoring the operating status of a hub bearing is implemented.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The present invention collects the vibration signal and temperature signal of the hub bearing during operation in real time, collects data for characterizing different types of faults, and improves the monitoring sensitivity for different types of faults. Then, the features of the vibration signal and the temperature signal are extracted, and the corresponding second features are constructed. The relationship between multiple second features and the temporal association of each second feature are used to construct multiple graph data. The present invention constructs feature-to-feature correlation relationship graph data of multiple second features, improves the correlation between fault data and hub bearing state changes, and constructs feature-time scale correlation relationship graph data within each second feature, deeply mining the weak features of hub bearing state changes. The present invention not only takes into account the changes in the signal itself, but also takes into account the correlation between signals, enhances the generalization performance of hub bearing online monitoring for complex fault modes, and improves the accuracy of hub bearing online monitoring in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 It is a flow chart of a method for monitoring the running status of a hub bearing of the present invention;

[0043] Figure 2 It is a schematic diagram of the graph data of the present invention;

[0044] Figure 3 It is a schematic diagram of the parallel network model structure of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] The present invention provides a method for monitoring the operating status of a hub bearing, referring to Figure 1 , specifically including the following steps:

[0047] Step 1: Collect monitoring data of wheel hub bearing during operation in real time.

[0048] The vibration sensor is used to collect the vibration signal of the wheel hub bearing in real time during operation, and the temperature sensor is used to collect the temperature signal of the wheel hub bearing in real time during operation. In this embodiment, the vibration signal is a three-way vibration signal.

[0049] Step 2: Extract the features of the vibration signal to obtain multiple time domain features and frequency domain features.

[0050] The time domain features include mean, standard deviation and skewness, and the frequency domain features include spectrum energy, center frequency and spectrum flatness.

[0051] Step 3: Optimize multiple time domain features and frequency domain features to obtain multiple first features for characterizing the failure trend of the hub bearing.

[0052] The vibration signal contains complex information, but some of its characteristic information does not contain information about the failure of the hub bearing. Therefore, effective feature extraction and screening of the vibration signal, removal of features without degradation trends, and mining of features that can reflect the bearing failure process are important foundations for subsequent good state monitoring. Good characteristic parameters should have monotonicity and time correlation, so the present invention selects a comprehensive evaluation index of monotonicity and time correlation as the criterion for feature screening.

[0053] The monotonicity index is mainly used to evaluate whether the data shows a consistent upward or downward trend over time. The main steps to calculate the monotonicity index include: (1) performing a difference operation on the data. (2) calculating the number of positive and negative values ​​in the difference result, and then calculating the absolute value of the difference between the two numbers. (3) averaging this absolute value to obtain the monotonicity index. The formula for the monotonicity index is as follows:

[0054] ;

[0055] In the formula, num is the number, d is the data after difference, K is the total length of the time series, Mon is the monotonicity index, 0≤ Mon ≤1, Mon The larger it is, the better the data monotonicity is.

[0056] In natural science, the Pearson coefficient is a common correlation coefficient. The Pearson correlation coefficient is suitable for measuring the linear correlation between two variables. Its calculation result is between -1 and 1. It was proposed by Karl Pearson. When the Pearson coefficient is -1, it indicates that the variables are completely negatively correlated, when it is 0, it indicates that there is no correlation between the variables, and when it is 1, it indicates that the variables are completely positively correlated.

[0057] When constructing a bearing health index, because the remaining useful life (RUL) tends to decrease with increasing vibration amplitude and intensity, it is very important to select features that have a strong correlation with RUL, so that such features can better reflect the degradation process of the bearing. Based on this, the present invention uses the Pearson correlation coefficient to screen features and evaluate the performance of the health index. The Pearson correlation coefficient measures the linear relationship between the feature and RUL, and its calculation formula is as follows:

[0058] ;

[0059] In the formula, f k Characterized by time point The value of K is the total length of the time series, f t Characterized by time point t The value of i For the i A point in time, For parameters f The average value of Remaining service life R The average value of R k For k Remaining service life RThe value of Corr is the correlation index, , Corr The closer it is to 1, the stronger the correlation between the feature and RUL.

[0060] The present invention calculates the correlation and monotonicity of the extracted multiple time domain features and frequency domain features to evaluate their relevance and change trend with the vibration signal. By calculating the correlation, it can be understood whether there is a linear correlation between the time domain features and the frequency domain features, while the monotonicity can reveal whether the time domain features and the frequency domain features increase or decrease monotonically with the change of time or frequency. By analyzing the three-way vibration signal in detail, the present invention screened out a total of 45 first features that can characterize the failure trend of the hub bearing.

[0061] Step 4: Integrate the temperature signal into the first feature to obtain multiple second features.

[0062] Considering the influence of temperature signal on the performance degradation of wheel hub bearing, the present invention adds temperature signal (including the maximum value and the minimum value of temperature signal) as an additional feature, so that the total number of second features used to describe the failure trend of bearing reaches 47.

[0063] Step 5: Define the values ​​corresponding to multiple second features at the current moment as first nodes, define the interaction between the first nodes as the first edge, and construct the first graph data.

[0064] Graph data construction is the process of converting sensor data into graph data. Traditional graph network models mostly use prior knowledge to predetermine the structure of the graph. During the model training process, the graph structure is no longer updated. The bearing failure data is a dynamic time series that contains different features of different sensors. The data type is complex, and the relationship between the data is unknown. The graph structure may change over time. Traditional methods cannot reflect the dynamic characteristics of time. Therefore, the present invention selects a graph adaptive learning method and uses a graph attention mechanism to better learn the topological structure relationship between nodes. The importance of each node is calculated by learning the similarity and connection relationship between nodes, and the weights between edges are adaptively updated during the training process.

[0065] The present invention defines the values ​​corresponding to the multiple second features at the current moment as the first nodes, and the dependencies between the nodes are used as edges to construct the graph structure. Since the dependency pattern between sensors does not need to be symmetrical, a directed graph is selected for construction, using the adjacency matrix A To represent the graph structure, A mn Represents a slave node m To Node n There are directed edge connections.

[0066] A graph attention-based feature extractor is introduced to capture the relationship between sensors and aggregate node information with the information of its surrounding neighboring nodes in the form of a graph structure. The feature vectors of neighboring nodes are weighted using attention weights to facilitate updating the state of the current first node.

[0067] like Figure 2 All the first nodes (1, 2, 3, 4, 5, 6, 7, 8, 9, etc.) are designed to be fully connected in the initial state, that is, each node is connected to all other nodes.

[0068] The calculation formula for the first edge is as follows:

[0069] ;

[0070] In the formula, and Represents the first node a and the first node b , It means taking the maximum value of the two first nodes. Represents the first edge between two first nodes.

[0071] Step 6: Define the values ​​corresponding to each second feature at different times as second nodes, define the interactions between the second nodes as second edges, and construct the second graph data.

[0072] The value of each second feature at N moments before the current moment (including the current moment) is defined as the second node of the graph data, and the interaction relationship between the second nodes is represented by edges, thereby obtaining multiple second graph data. All second nodes are designed to be fully connected in the initial state, that is, each node is connected to all other nodes.

[0073] The calculation formula for the second edge is as follows:

[0074] ;

[0075] In the formula, and Respectively represent the second node c and the second node d , It means taking the maximum value of the two second nodes. Represents the second edge between two second nodes.

[0076] Step 7: Input the graph data into the graph attention network model to obtain the inference score value. Compare the inference score value with the set threshold value. When the inference score value is greater than the threshold value, it is determined that the wheel hub bearing state is abnormal.

[0077] The topological structure of graph data is complex and has the dual characteristics of high-dimensional space and time. Traditional neural network models may encounter difficulties in processing high-dimensional data with dual characteristics of space and time. Therefore, the present invention introduces a parallel graph attention network model to learn vibration signals from the two dimensions of time and space. The model can mine the complex coupling relationship of bearing multimodal monitoring signals and represent the dynamic changes of the time dimension at the same time, so as to more accurately realize the abnormal state monitoring of bearings. The parallel model is divided into two parts: reconstruction and prediction. The results of the two are combined as a judgment indicator to detect the operating status of the rotor bearing. The model structure is as follows: Figure 3 As shown, it includes a temporal attention layer, a feature attention layer, a fusion layer (gated recurrent unit), a prediction layer (prediction model), a reconstruction layer (reconstruction model), and a computation layer.

[0078] Feature extraction is performed on the first graph data and multiple second graph data through the time attention layer and the feature attention layer to obtain multiple time features and multiple spatial features.

[0079] The first graph data, the plurality of second graph data, the plurality of time features and the plurality of space features are fused through a fusion layer to obtain a fused feature.

[0080] The fused features are processed through the prediction layer to obtain the predicted value (prediction loss); the fused features are reconstructed through the reconstruction layer to obtain the reconstructed value (reconstruction loss).

[0081] The predicted value and the reconstructed value are superimposed to obtain the score value (abnormal score), which is the processing result.

[0082] In this embodiment, the temporal attention layer, the feature attention layer, the gated recurrent unit, the prediction model and the reconstruction model are all existing technical means.

[0083] The prediction model and the reconstruction model have obvious complementary advantages. The prediction model is used to predict the value at the next time point, and the reconstruction model is used to describe the data distribution of the entire time series. The parameters of these two models are updated simultaneously during the model training phase, and the loss function is defined as the sum of the two optimization objectives, namely Loss = Loss for + Loss rec ,in Loss for represents the loss function of the prediction model, Loss rec Represents the loss function of the reconstruction model.

[0084] Model status monitoring is mainly divided into two parts: training and monitoring. After the original data is normalized, only normal data is used for model training. During the training process, the threshold of the normal state is determined based on the performance of the normal data, that is, the threshold is set. In the monitoring state, the same model parameters are used to train the test data, and the training results are compared with the set normal threshold. When the score value is less than the set threshold, the wheel hub bearing is judged to be in normal operation, and the next set of data is monitored. When the score value is greater than the set threshold, the wheel hub bearing is judged to be abnormal, and the monitoring ends.

[0085] Based on the same concept, the present invention also provides a wheel hub bearing operating status monitoring device, including a collection module, an extraction module, a construction module and a processing module.

[0086] The acquisition module is used to collect the vibration signal and temperature signal of the wheel hub bearing during operation in real time.

[0087] The extraction module is used to extract the features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing, and to superimpose the corresponding features of the temperature signal with the plurality of first features to obtain a plurality of second features.

[0088] The construction module is used to define the values ​​corresponding to multiple second features at the current moment as first nodes, define the interactions between the first nodes as first edges, and construct first graph data; define the values ​​corresponding to each second feature at different moments as second nodes, define the interactions between the second nodes as second edges, and construct second graph data.

[0089] The processing module is used to predict, reconstruct and superimpose the first image data and multiple second image data to obtain processing results, and monitor the operating status of the wheel hub bearing according to the processing results.

[0090] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the above-mentioned wheel hub bearing operating status monitoring method when executing the program.

[0091] The present invention also provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned wheel hub bearing operating status monitoring method is implemented.

[0092] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0093] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for monitoring the operating status of a hub bearing, characterized in that: The following steps are involved: Real-time collection of vibration and temperature signals of wheel hub bearings during operation; Extracting the features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing, and superimposing the corresponding features of the temperature signal with the plurality of first features to obtain a plurality of second features; The values ​​corresponding to the multiple second features at the current moment are respectively defined as first nodes, the interactions between the first nodes are defined as first edges, and the first graph data is constructed; The values ​​corresponding to each second feature at different times are respectively defined as second nodes, and the interactions between the second nodes are defined as second edges, so as to construct second graph data; Predicting, reconstructing and superimposing the first image data and the plurality of second image data to obtain a processing result, and monitoring the running state of the wheel hub bearing according to the processing result; The step of extracting the features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing comprises the following steps: Extract the features of the vibration signal to obtain multiple time domain features and frequency domain features; Optimizing multiple time domain features and frequency domain features through a monotonicity index and a correlation index to obtain multiple first features for characterizing the failure trend of the hub bearing; The interaction between the first nodes is defined as the first edge, which is as follows: In the formula, x a and x b represents the first node a and the first node b, max(x a , x b ) means taking the maximum value of the two first nodes, y ab Represents the first edge of two first nodes; The interaction between the second nodes is defined as the second edge, as follows: In the formula, x c and x d represents the second node c and the second node d, max(x c , x d ) means taking the maximum value of the two second nodes, y cd The second edge representing the two second nodes; The method of predicting, reconstructing and superimposing the first image data and the plurality of second image data to obtain a processing result comprises the following steps: Perform feature extraction on the first image data and the plurality of second image data through a time attention layer and a feature attention layer to obtain a plurality of time features and a plurality of space features; The first graph data, the plurality of second graph data, the plurality of time features and the plurality of spatial features are fused through a fusion layer to obtain a fusion feature; The fused features are processed through the prediction layer to obtain the predicted value; the fused features are reconstructed through the reconstruction layer to obtain the reconstructed value; The predicted value and the reconstructed value are superimposed to obtain the score value, which is the processing result.

2. A method for monitoring the operating status of a hub bearing according to claim 1, characterized in that: The step of monitoring the operating state of the hub bearing according to the processing result comprises the following steps: comparing the score value with a set threshold; When the score value is greater than the set threshold, it is determined that the wheel hub bearing operating state is abnormal; when the score value is less than the set threshold, it is determined that the wheel hub bearing operating state is normal.

3. A method for monitoring the operating status of a hub bearing according to claim 1, characterized in that: The real-time collection of vibration signals and temperature signals of the hub bearing during operation specifically includes the following steps: The vibration sensor is used to collect the vibration signal of the wheel hub bearing in real time during operation; The temperature signal of the wheel hub bearing during operation is collected in real time by the temperature sensor, and the corresponding characteristics of the temperature signal include the maximum value of the temperature signal and the minimum value of the temperature signal.

4. A wheel hub bearing operating status monitoring device, characterized in that: include: The acquisition module is used to collect the vibration signal and temperature signal of the wheel hub bearing in real time during operation; An extraction module is used to extract features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing, and to superimpose corresponding features of the temperature signal with the plurality of first features to obtain a plurality of second features; A construction module, used to define the values ​​corresponding to the plurality of second features at the current moment as first nodes, define the interactions between the first nodes as first edges, and construct first graph data; define the values ​​corresponding to each second feature at different moments as second nodes, define the interactions between the second nodes as second edges, and construct second graph data; A processing module, used for predicting, reconstructing and superimposing the first image data and the plurality of second image data to obtain a processing result, and monitoring the running state of the wheel hub bearing according to the processing result; The step of extracting the features of the vibration signal to obtain a plurality of first features for characterizing the failure trend of the hub bearing comprises the following steps: Extract the features of the vibration signal to obtain multiple time domain features and frequency domain features; Optimizing multiple time domain features and frequency domain features through a monotonicity index and a correlation index to obtain multiple first features for characterizing the failure trend of the hub bearing; The interaction between the first nodes is defined as the first edge, which is as follows: In the formula, x a and x b represents the first node a and the first node b, max(x a , x b ) means taking the maximum value of the two first nodes, y ab Represents the first edge of two first nodes; The interaction between the second nodes is defined as the second edge, as follows: In the formula, x c and x d represents the second node c and the second node d, max(x c , x d ) means taking the maximum value of the two second nodes, y cd The second edge representing the two second nodes; The method of predicting, reconstructing and superimposing the first image data and the plurality of second image data to obtain a processing result comprises the following steps: Perform feature extraction on the first image data and the plurality of second image data through a time attention layer and a feature attention layer to obtain a plurality of time features and a plurality of space features; The first graph data, the plurality of second graph data, the plurality of time features and the plurality of spatial features are fused through a fusion layer to obtain a fusion feature; The fused features are processed through the prediction layer to obtain the predicted value; the fused features are reconstructed through the reconstruction layer to obtain the reconstructed value; The predicted value and the reconstructed value are superimposed to obtain the score value, which is the processing result.

5. A computer device, characterized in that: It comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for monitoring the operating status of the hub bearing as described in any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method for monitoring the operating status of a hub bearing as described in any one of claims 1 to 3 is implemented.

Images