Bearing service life prediction method and system based on model training

Abstract

The invention relates to the technical field of bearing service life prediction, in particular to a bearing service life prediction method and system based on model training, and the method comprises the steps: obtaining first hub bearing data during the operation of an automobile, constructing first multi-dimensional hub bearing data, and inputting the first multi-dimensional hub bearing data into a bearing service life prediction model; extracting a first multi-dimensional hub bearing feature through a feature extraction layer of the model; a feature analysis layer of the model obtains a bearing vibration abnormal coefficient, a bearing temperature abnormal coefficient and a bearing sound wave abnormal coefficient according to the first multi-dimensional hub bearing features, and further obtains a bearing comprehensive abnormal coefficient; analyzing according to the comprehensive abnormal coefficient of the bearing and a preset abnormal coefficient threshold value of the bearing to obtain the service life of the hub bearing; according to the method, the service life prediction accuracy of the hub bearing is improved by constructing the bearing service life prediction model, so that the automobile driving safety is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of bearing life prediction, and in particular to a bearing service life prediction method and system based on model training. Background Art

[0002] The wheel hub bearing is a key component installed in the automobile wheel hub, which is used to bear the weight and provide precise guidance for the rotation of the wheel hub. The health of the wheel hub bearing directly affects the safety of automobile driving. Therefore, predicting the service life of the wheel hub bearing is of great significance to improvement.

[0003] Most of the existing wheel hub bearing life prediction methods analyze single vibration data, resulting in a lack of comprehensiveness in the prediction of wheel hub bearing service life, which in turn limits the accuracy of wheel hub bearing service life prediction, thus affecting the safety of vehicle driving.

[0004] Therefore, a bearing service life prediction method and system based on model training is proposed. Summary of the invention

[0005] The purpose of the present invention is to provide a bearing service life prediction method and system based on model training. The present invention relates to the field of bearing life prediction technology, specifically to a bearing service life prediction method and system based on model training. The present invention obtains the first wheel hub bearing data when the car is running, constructs the first multi-dimensional wheel hub bearing data and inputs the bearing life prediction model, extracts the first multi-dimensional wheel hub bearing feature through the feature extraction layer of the model; the feature analysis layer of the model obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing sound wave abnormality coefficient according to the first multi-dimensional wheel hub bearing feature, and then obtains the bearing comprehensive abnormality coefficient; according to the bearing comprehensive abnormality coefficient and the preset bearing abnormality coefficient threshold, the analysis is performed to obtain the wheel hub bearing service life; the present invention improves the wheel hub bearing life prediction accuracy by constructing the bearing life prediction model, thereby improving the driving safety of the car.

[0006] To achieve the above object, the present invention provides the following technical solutions: A bearing service life prediction method based on model training, comprising: S10. Acquire first wheel hub bearing data when the vehicle is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, and construct first multi-dimensional wheel hub bearing data according to the first wheel hub bearing data; S20. Input the first multi-dimensional wheel hub bearing data into the bearing life prediction model, and sequentially pass through the input layer, preprocessing layer, feature extraction layer, feature analysis layer, bearing life prediction layer and output layer of the bearing life prediction model; the preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain the second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive abnormality coefficient, and the bearing comprehensive abnormality coefficient is: ; in, is the comprehensive abnormal coefficient of the bearing; is the influencing factor of bearing vibration abnormality coefficient; is the bearing vibration abnormality coefficient; is the influencing factor of bearing temperature anomaly coefficient; is the bearing temperature anomaly coefficient; is the influencing factor of the bearing acoustic abnormality coefficient; is the bearing acoustic anomaly coefficient; S30. The bearing life prediction layer analyzes the bearing comprehensive abnormality coefficient and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; the output layer is used to output the wheel hub bearing service life.

[0007] Preferably, the bearing vibration data includes outer ring vibration acceleration and inner ring vibration acceleration; the bearing temperature data includes outer ring temperature and inner ring temperature; and the bearing sound wave data includes outer ring sound wave signal.

[0008] Preferably, the first multi-dimensional hub bearing data is: ; in, represents the first multi-dimensional wheel hub bearing data; Indicates bearing vibration data; Indicates bearing temperature data; Represents bearing acoustic wave data.

[0009] Preferably, the bearing life prediction model comprises an input layer, a preprocessing layer, a feature extraction layer, a feature analysis layer, a bearing life prediction layer and an output layer; The preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain second multi-dimensional wheel hub bearing data; the preprocessing includes data cleaning and standardization; The feature extraction layer is used to extract the first multidimensional wheel hub bearing feature of the second multidimensional wheel hub bearing data; the feature extraction layer performs feature extraction through LSTM to obtain the first multidimensional wheel hub bearing feature, and the first multidimensional wheel hub bearing feature includes bearing vibration feature, bearing temperature feature and bearing acoustic wave feature; the bearing vibration feature includes the average vibration acceleration of the outer ring, the average vibration acceleration of the inner ring, the maximum vibration acceleration of the outer ring and the maximum vibration acceleration of the inner ring; the bearing temperature feature includes the average temperature of the outer ring and the average temperature of the inner ring; the bearing acoustic wave feature includes the peak value, mean value and standard deviation of the acoustic wave signal; The feature analysis layer is used to obtain the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing feature, and then obtain the bearing comprehensive abnormality coefficient; The bearing life prediction layer compares and analyzes the bearing comprehensive abnormality coefficient with the preset hub abnormality coefficient to obtain the hub bearing service life; The output layer is used to output the service life of the hub bearing.

[0010] Preferably, the bearing vibration abnormal coefficient is: ; in, is the bearing vibration abnormality coefficient; is the influence factor of the average vibration acceleration of the outer ring; is the average vibration acceleration of the outer ring; is the reference outer ring vibration acceleration; is the influence factor of the average vibration acceleration of the inner ring; is the average vibration acceleration of the inner ring; is the reference inner ring vibration acceleration; is the maximum vibration acceleration influence factor of the outer ring; is the maximum vibration acceleration of the outer ring; is the maximum vibration acceleration influence factor of the inner ring; is the maximum vibration acceleration of the inner ring; The bearing temperature anomaly coefficient is: ; in, is the bearing temperature anomaly coefficient; is the influence factor of the average temperature of the outer ring; is the average temperature of the outer ring; is the reference outer ring temperature; is the influence factor of the average temperature of the inner ring; is the average temperature of the inner ring; is the reference inner ring temperature; The bearing acoustic wave anomaly coefficient is: ; in, is the bearing acoustic anomaly coefficient; is the acoustic wave peak impact factor; is the peak value of the sound wave signal; is the peak value of the reference sound wave signal; is the influence factor of the mean value of the acoustic wave signal; is the mean value of the reference sound wave signal; is the influence factor of the standard deviation of the acoustic signal; is the standard deviation of the acoustic signal; is the standard deviation of the reference acoustic wave signal.

[0011] Preferably, the service life of the wheel hub bearing is: ; in, The service life of the wheel hub bearing; For normal wheel hub bearing service life; is the influencing factor of the comprehensive abnormal coefficient of the bearing; is the comprehensive abnormal coefficient of the bearing; It is the preset bearing abnormality coefficient threshold.

[0012] A bearing service life prediction system based on model training, used to execute the bearing service life prediction method based on model training, comprising: A data set construction module is used to obtain first wheel hub bearing data when the vehicle is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, and first multi-dimensional wheel hub bearing data is constructed according to the first wheel hub bearing data; The bearing life prediction model analysis module is used to input the first multi-dimensional wheel hub bearing data into the bearing life prediction model, and sequentially passes through the input layer, preprocessing layer, feature extraction layer, feature analysis layer, bearing life prediction layer and output layer of the bearing life prediction model; the preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive abnormality coefficient, and the bearing comprehensive abnormality coefficient is: ; in, is the comprehensive abnormal coefficient of the bearing; is the influencing factor of bearing vibration abnormality coefficient; is the bearing vibration abnormality coefficient; is the influencing factor of bearing temperature anomaly coefficient; is the bearing temperature anomaly coefficient; is the influencing factor of the bearing acoustic abnormality coefficient; is the bearing acoustic anomaly coefficient; The bearing life output module is used to analyze the bearing comprehensive abnormality coefficient and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; and output the wheel hub bearing service life through the output layer.

[0013] Preferably, the bearing vibration data includes outer ring vibration acceleration and inner ring vibration acceleration; the bearing temperature data includes outer ring temperature and inner ring temperature; and the bearing sound wave data includes outer ring sound wave signal.

[0014] Preferably, the bearing vibration abnormal coefficient is: ; in, is the bearing vibration abnormality coefficient; is the influence factor of the average vibration acceleration of the outer ring; is the average vibration acceleration of the outer ring; is the reference outer ring vibration acceleration; is the influence factor of the average vibration acceleration of the inner ring; is the average vibration acceleration of the inner ring; is the reference inner ring vibration acceleration; is the maximum vibration acceleration influence factor of the outer ring; is the maximum vibration acceleration of the outer ring; is the maximum vibration acceleration influence factor of the inner ring; is the maximum vibration acceleration of the inner ring; The bearing temperature anomaly coefficient is: ; in, is the bearing temperature anomaly coefficient; is the influence factor of the average temperature of the outer ring; is the average temperature of the outer ring; is the reference outer ring temperature; is the influence factor of the average temperature of the inner ring; is the average temperature of the inner ring; is the reference inner ring temperature; The bearing acoustic wave anomaly coefficient is: ; in, is the bearing acoustic anomaly coefficient; is the acoustic wave peak impact factor; is the peak value of the sound wave signal; is the peak value of the reference sound wave signal; is the influence factor of the mean value of the acoustic wave signal; is the mean value of the reference sound wave signal; is the influence factor of the standard deviation of the acoustic signal; is the standard deviation of the acoustic signal; is the standard deviation of the reference acoustic wave signal.

[0015] Preferably, the service life of the wheel hub bearing is: ; in, The service life of the wheel hub bearing; For normal wheel hub bearing service life; is the influencing factor of the comprehensive abnormal coefficient of the bearing; is the comprehensive abnormal coefficient of the bearing; It is the preset bearing abnormality coefficient threshold.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention obtains first wheel hub bearing data when the automobile is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, constructs first multi-dimensional wheel hub bearing data according to the first wheel hub bearing data, and inputs the bearing life prediction model. By constructing the first multi-dimensional wheel hub bearing data, a comprehensive data basis is provided for the bearing life prediction model, the comprehensiveness of the bearing life prediction is improved, thereby improving the accuracy of the bearing life prediction, and further improving the safety of automobile driving.

[0017] 2. The present invention identifies the first multi-dimensional wheel hub bearing data by constructing a bearing life prediction model, and the model consists of an input layer, a preprocessing layer, a feature extraction layer, a feature analysis layer, a bearing life prediction layer and an output layer; the preprocessing layer preprocesses the first multi-dimensional wheel hub bearing data to obtain the second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration anomaly coefficient, the bearing temperature anomaly coefficient and the bearing acoustic wave anomaly coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive anomaly coefficient, and analyzes the bearing anomaly through the bearing life prediction model, providing a comprehensive and accurate model basis, thereby improving the accuracy of bearing life prediction, and then improving the safety of vehicle driving.

[0018] 3. The present invention analyzes the bearing comprehensive abnormality coefficient predicted by the bearing life prediction model and the preset bearing abnormality coefficient threshold to obtain the service life of the hub bearing; combining the abnormal analysis of the hub bearing with the hub bearing life prediction, the accuracy of the hub bearing service life prediction is improved, thereby improving the safety of vehicle driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic flow chart of a bearing service life prediction method based on model training provided in Embodiment 1 of the present invention; Figure 2 A schematic diagram of the structure of a bearing service life prediction system based on model training provided in the second embodiment of the present invention; Figure 3 A schematic diagram of the structure of a bearing life prediction model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] 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.

[0021] Embodiment 1 In order to achieve the accuracy of A wheel hub bearing life prediction and improve the safety of vehicle driving, a bearing service life prediction method based on model training was applied; Reference Figure 1 , is a flow chart of a bearing service life prediction method based on model training provided in Embodiment 1 of the present invention, comprising: S10. Acquire first wheel hub bearing data when the vehicle is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, and construct first multi-dimensional wheel hub bearing data according to the first wheel hub bearing data; Further, the bearing vibration data includes outer ring vibration acceleration and inner ring vibration acceleration; the bearing temperature data includes outer ring temperature and inner ring temperature; the bearing sound wave data includes outer ring sound wave signal; the bearing vibration data is acquired through a vibration sensor; the bearing temperature data is acquired through a temperature sensor; the bearing sound wave data is acquired through a sound sensor; Furthermore, the first multi-dimensional wheel hub bearing data is: ; in, represents the first multi-dimensional wheel hub bearing data; Indicates bearing vibration data; Indicates bearing temperature data; Represents bearing acoustic wave data.

[0022] This embodiment obtains first wheel hub bearing data when the car is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, constructs first multi-dimensional wheel hub bearing data according to the first wheel hub bearing data, and inputs the bearing life prediction model. By constructing the first multi-dimensional wheel hub bearing data, a comprehensive data foundation is provided for the bearing life prediction model, the comprehensiveness of the bearing life prediction is improved, thereby improving the accuracy of the bearing life prediction, and further improving the safety of the car driving.

[0023] S20. Input the first multi-dimensional wheel hub bearing data into the bearing life prediction model, and sequentially pass through the input layer, preprocessing layer, feature extraction layer, feature analysis layer, bearing life prediction layer and output layer of the bearing life prediction model; the preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain the second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive abnormality coefficient, and the bearing comprehensive abnormality coefficient is: ; in, is the comprehensive abnormal coefficient of the bearing; is the influencing factor of bearing vibration abnormality coefficient; is the bearing vibration abnormality coefficient; is the influencing factor of bearing temperature anomaly coefficient; is the bearing temperature anomaly coefficient; is the influencing factor of the bearing acoustic abnormality coefficient; is the bearing acoustic anomaly coefficient; Further, , and They are 0.3, 0.3 and 0.4 respectively; Further, refer to Figure 3 A structural schematic diagram of a bearing life prediction model provided by an embodiment of the present invention; the bearing life prediction model comprises an input layer, a preprocessing layer, a feature extraction layer, a feature analysis layer, a bearing life prediction layer and an output layer; the preprocessing layer is used to preprocess the first multi-dimensional hub bearing data to obtain second multi-dimensional hub bearing data; the preprocessing includes data cleaning and standardization; The feature extraction layer is used to extract the first multidimensional wheel hub bearing feature of the second multidimensional wheel hub bearing data; the feature extraction layer performs feature extraction through LSTM to obtain the first multidimensional wheel hub bearing feature, the first multidimensional wheel hub bearing feature includes bearing vibration feature, bearing temperature feature and bearing acoustic wave feature; the bearing vibration feature includes the average vibration acceleration of the outer ring, the average vibration acceleration of the inner ring, the maximum vibration acceleration of the outer ring and the maximum vibration acceleration of the inner ring; the bearing temperature feature includes the average temperature of the outer ring and the average temperature of the inner ring; the bearing acoustic wave feature includes the peak value, mean value and standard deviation of the acoustic wave signal; the first multidimensional wheel hub bearing feature is shown in Table 1; Table 1 A wheel hub bearing's first multi-dimensional wheel hub bearing characteristic data table

[0024] The feature analysis layer is used to obtain the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing feature, and then obtain the bearing comprehensive abnormality coefficient; The bearing life prediction layer compares and analyzes the bearing comprehensive abnormality coefficient with the preset hub abnormality coefficient to obtain the hub bearing service life; The output layer is used to output the service life of the wheel hub bearing; Furthermore, the bearing vibration abnormality coefficient is: ; in, is the bearing vibration abnormality coefficient; is the influence factor of the average vibration acceleration of the outer ring; is the average vibration acceleration of the outer ring; is the reference outer ring vibration acceleration; is the influence factor of the average vibration acceleration of the inner ring; is the average vibration acceleration of the inner ring; is the reference inner ring vibration acceleration; is the maximum vibration acceleration influence factor of the outer ring; is the maximum vibration acceleration of the outer ring; is the maximum vibration acceleration influence factor of the inner ring; is the maximum vibration acceleration of the inner ring; Further, , , and The historical bearing vibration data was analyzed and obtained by using the multivariate linear regression method; The bearing temperature anomaly coefficient is: ; in, is the bearing temperature anomaly coefficient; is the influence factor of the average temperature of the outer ring; is the average temperature of the outer ring; is the reference outer ring temperature; is the influence factor of the average temperature of the inner ring; is the average temperature of the inner ring; is the reference inner ring temperature; Further, and The historical bearing temperature data was analyzed and obtained by using the multivariate linear regression method; The bearing acoustic wave anomaly coefficient is: ; in, is the bearing acoustic anomaly coefficient; is the acoustic wave peak impact factor; is the peak value of the sound wave signal; is the peak value of the reference sound wave signal; is the influence factor of the mean value of the acoustic wave signal; is the mean value of the reference sound wave signal; is the influence factor of the standard deviation of the acoustic signal; is the standard deviation of the acoustic signal; is the standard deviation of the reference acoustic wave signal.

[0025] Further, , and The historical bearing acoustic wave data were analyzed and obtained by using the multivariate linear regression method;

[0026] This embodiment identifies the first multi-dimensional wheel hub bearing data by constructing a bearing life prediction model, and the model consists of an input layer, a preprocessing layer, a feature extraction layer, a feature analysis layer, a bearing life prediction layer and an output layer; the preprocessing layer preprocesses the first multi-dimensional wheel hub bearing data to obtain the second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration anomaly coefficient, the bearing temperature anomaly coefficient and the bearing sound wave anomaly coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive anomaly coefficient, and analyzes the bearing anomaly through the bearing life prediction model, providing a comprehensive and accurate model foundation, thereby improving the accuracy of bearing life prediction, and then improving the safety of vehicle driving.

[0027] Furthermore, the bearing vibration abnormal coefficient, bearing temperature abnormal coefficient, bearing acoustic wave abnormal coefficient and bearing comprehensive abnormal coefficient of hub A are specifically shown in Table 2; Table 2 Table of various abnormal coefficients of wheel hub bearing A

[0028] S30. The bearing life prediction layer analyzes the bearing comprehensive abnormality coefficient and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; the output layer is used to output the wheel hub bearing service life.

[0029] Furthermore, the service life of the hub bearing is: ; in, The service life of the wheel hub bearing; For normal wheel hub bearing service life; is the influencing factor of the comprehensive abnormal coefficient of the bearing; is the comprehensive abnormal coefficient of the bearing; It is the preset bearing abnormality coefficient threshold.

[0030] Furthermore, the normal wheel hub bearing service life is represented according to the normal usable time of the wheel hub; the preset bearing abnormality coefficient threshold is obtained according to the historical data training analysis of the A wheel hub bearing; In order to verify the effectiveness of the bearing life prediction model provided in this embodiment in the wheel hub bearing service life prediction task, the life prediction accuracy of multiple wheel hub bearings is compared and analyzed using Model 1, Model 2 and Model 3; Model 1 is the bearing life prediction model provided in this embodiment; Model 2 is based on Model 1 without considering temperature data and acoustic wave data; Model 3 is based on Model 1 without considering the bearing comprehensive abnormality coefficient of the wheel hub bearing. The average values ​​of the life prediction accuracy of multiple wheel hub bearings using different models are compared and analyzed, as shown in Table 3. Table 3 Comparison of the average accuracy of life prediction of multiple wheel hub bearings by different models

[0031] It can be seen from Table 3 that the bearing life prediction model of this embodiment has certain effectiveness in the task of predicting the service life of the hub bearing.

[0032] This embodiment performs analysis based on the bearing comprehensive abnormality coefficient predicted by the bearing life prediction model and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; combining the abnormality analysis of the wheel hub bearing with the wheel hub bearing life prediction improves the accuracy of the wheel hub bearing service life prediction, thereby improving the safety of vehicle driving.

[0033] This embodiment acquires the first wheel hub bearing data when the car is running, constructs the first multi-dimensional wheel hub bearing data and inputs the bearing life prediction model, and extracts the first multi-dimensional wheel hub bearing feature through the feature extraction layer of the model; the feature analysis layer of the model obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing sound wave abnormality coefficient according to the first multi-dimensional wheel hub bearing feature, and then obtains the comprehensive bearing abnormality coefficient; the comprehensive bearing abnormality coefficient and the preset bearing abnormality coefficient threshold are analyzed to obtain the service life of the wheel hub bearing; the present invention improves the accuracy of wheel hub bearing life prediction by constructing a bearing life prediction model, thereby improving the driving safety of the car.

[0034] Embodiment 2 In order to achieve the accuracy of B wheel hub bearing life prediction and improve the safety of automobile driving, a bearing service life prediction system based on model training was applied; Reference Figure 2 , is a schematic diagram of a bearing service life prediction system based on model training provided in Embodiment 2 of the present invention, comprising: A data set construction module is used to obtain first wheel hub bearing data when the vehicle is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, and first multi-dimensional wheel hub bearing data is constructed according to the first wheel hub bearing data; Furthermore, the first multi-dimensional wheel hub bearing data is: ; in, represents the first multi-dimensional wheel hub bearing data; Indicates bearing vibration data; Indicates bearing temperature data; Represents bearing acoustic wave data.

[0035] This embodiment obtains first wheel hub bearing data when the car is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, constructs first multi-dimensional wheel hub bearing data according to the first wheel hub bearing data, and inputs the bearing life prediction model. By constructing the first multi-dimensional wheel hub bearing data, a comprehensive data foundation is provided for the bearing life prediction model, the comprehensiveness of the bearing life prediction is improved, thereby improving the accuracy of the bearing life prediction, and further improving the safety of the car driving.

[0036] The bearing life prediction model analysis module is used to input the first multi-dimensional wheel hub bearing data into the bearing life prediction model, and sequentially passes through the input layer, preprocessing layer, feature extraction layer, feature analysis layer, bearing life prediction layer and output layer of the bearing life prediction model; the preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive abnormality coefficient, and the bearing comprehensive abnormality coefficient is: ; in, is the comprehensive abnormal coefficient of the bearing; is the influencing factor of bearing vibration abnormality coefficient; is the bearing vibration abnormality coefficient; is the influencing factor of bearing temperature anomaly coefficient; is the bearing temperature anomaly coefficient; is the influencing factor of the bearing acoustic abnormality coefficient; is the bearing acoustic anomaly coefficient; Further, , and They are 0.3, 0.3 and 0.4 respectively; Further, refer to Figure 3 A structural schematic diagram of a bearing life prediction model provided by an embodiment of the present invention; the bearing life prediction model comprises an input layer, a preprocessing layer, a feature extraction layer, a feature analysis layer, a bearing life prediction layer and an output layer; the preprocessing layer is used to preprocess the first multi-dimensional hub bearing data to obtain second multi-dimensional hub bearing data; the preprocessing includes data cleaning and standardization; The feature extraction layer is used to extract the first multidimensional hub bearing feature of the second multidimensional hub bearing data; the feature extraction layer performs feature extraction through LSTM to obtain the first multidimensional hub bearing feature, the first multidimensional hub bearing feature includes bearing vibration feature, bearing temperature feature and bearing acoustic wave feature; the bearing vibration feature includes the average vibration acceleration of the outer ring, the average vibration acceleration of the inner ring, the maximum vibration acceleration of the outer ring and the maximum vibration acceleration of the inner ring; the bearing temperature feature includes the average temperature of the outer ring and the average temperature of the inner ring; the bearing acoustic wave feature includes the peak value, mean value and standard deviation of the acoustic wave signal; the first multidimensional hub bearing feature is shown in Table 4; Table 4 The first multi-dimensional wheel hub bearing characteristic data table of B wheel hub bearing

[0037] The feature analysis layer is used to obtain the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing feature, and then obtain the bearing comprehensive abnormality coefficient; The bearing life prediction layer compares and analyzes the bearing comprehensive abnormality coefficient with the preset hub abnormality coefficient to obtain the hub bearing service life; the output layer is used to output the hub bearing service life; Furthermore, the bearing vibration abnormality coefficient is: ; in, is the bearing vibration abnormality coefficient; is the influence factor of the average vibration acceleration of the outer ring; is the average vibration acceleration of the outer ring; is the reference outer ring vibration acceleration; is the influence factor of the average vibration acceleration of the inner ring; is the average vibration acceleration of the inner ring; is the reference inner ring vibration acceleration; is the maximum vibration acceleration influence factor of the outer ring; is the maximum vibration acceleration of the outer ring; is the maximum vibration acceleration influence factor of the inner ring; is the maximum vibration acceleration of the inner ring; Further, , , and The historical bearing vibration data was analyzed and obtained by using the multivariate linear regression method; The bearing temperature anomaly coefficient is: ; in, is the bearing temperature anomaly coefficient; is the influence factor of the average temperature of the outer ring; is the average temperature of the outer ring; is the reference outer ring temperature; is the influence factor of the average temperature of the inner ring; is the average temperature of the inner ring; is the reference inner ring temperature; Further, and The historical bearing temperature data was analyzed and obtained by using the multivariate linear regression method; The bearing acoustic wave anomaly coefficient is: ; in, is the bearing acoustic anomaly coefficient; is the acoustic wave peak impact factor; is the peak value of the sound wave signal; is the peak value of the reference sound wave signal; is the influence factor of the mean value of the acoustic wave signal; is the mean value of the reference sound wave signal; is the influence factor of the standard deviation of the acoustic signal; is the standard deviation of the acoustic signal; is the standard deviation of the reference acoustic wave signal.

[0038] Further, , and The historical bearing acoustic wave data were analyzed and obtained by using the multivariate linear regression method;

[0039] This embodiment identifies the first multi-dimensional wheel hub bearing data by constructing a bearing life prediction model, and the model consists of an input layer, a preprocessing layer, a feature extraction layer, a feature analysis layer, a bearing life prediction layer and an output layer; the preprocessing layer preprocesses the first multi-dimensional wheel hub bearing data to obtain the second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration anomaly coefficient, the bearing temperature anomaly coefficient and the bearing sound wave anomaly coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive anomaly coefficient, and analyzes the bearing anomaly through the bearing life prediction model, providing a comprehensive and accurate model foundation, thereby improving the accuracy of bearing life prediction, and then improving the safety of vehicle driving.

[0040] Furthermore, the bearing vibration abnormal coefficient, bearing temperature abnormal coefficient, bearing acoustic wave abnormal coefficient and bearing comprehensive abnormal coefficient of hub B are specifically shown in Table 5; Table 5 Abnormal coefficients of B wheel hub bearings

[0041] The bearing life output module is used to analyze the bearing comprehensive abnormality coefficient and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; and output the wheel hub bearing service life through the output layer.

[0042] Furthermore, the service life of the hub bearing is: ; in, The service life of the wheel hub bearing; For normal wheel hub bearing service life; is the influencing factor of the comprehensive abnormal coefficient of the bearing; is the comprehensive abnormal coefficient of the bearing; It is the preset bearing abnormality coefficient threshold.

[0043] Furthermore, the normal wheel hub bearing service life is represented according to the normal usable time of the wheel hub; the preset bearing abnormality coefficient threshold is obtained according to the historical data training analysis of the A wheel hub bearing; In order to verify the effectiveness of the bearing life prediction model provided in this embodiment in the wheel hub bearing service life prediction task, the life prediction accuracy of multiple wheel hub bearings is compared and analyzed using Model 1, Model 2 and Model 3; Model 1 is the bearing life prediction model provided in this embodiment; Model 2 is based on Model 1 without considering temperature data and acoustic wave data; Model 3 is based on Model 1 without considering the bearing comprehensive abnormality coefficient of the wheel hub bearing. The average values ​​of the life prediction accuracy of multiple wheel hub bearings using different models are compared and analyzed, as shown in Table 6. Table 6 Comparison of the average accuracy of life prediction of multiple wheel hub bearings by different models

[0044] It can be seen from Table 6 that the bearing life prediction model proposed in this embodiment has certain effectiveness in the wheel hub bearing service life prediction task.

[0045] This embodiment performs analysis based on the bearing comprehensive abnormality coefficient predicted by the bearing life prediction model and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; combining the abnormality analysis of the wheel hub bearing with the wheel hub bearing life prediction improves the accuracy of the wheel hub bearing service life prediction, thereby improving the safety of vehicle driving.

[0046] This embodiment acquires the first wheel hub bearing data when the car is running, constructs the first multi-dimensional wheel hub bearing data and inputs the bearing life prediction model, and extracts the first multi-dimensional wheel hub bearing feature through the feature extraction layer of the model; the feature analysis layer of the model obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing sound wave abnormality coefficient according to the first multi-dimensional wheel hub bearing feature, and then obtains the comprehensive bearing abnormality coefficient; the comprehensive bearing abnormality coefficient and the preset bearing abnormality coefficient threshold are analyzed to obtain the service life of the wheel hub bearing; the present invention improves the accuracy of wheel hub bearing life prediction by constructing a bearing life prediction model, thereby improving the driving safety of the car.

[0047] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A bearing service life prediction method based on model training, characterized in that: include: S10. Acquire first wheel hub bearing data when the vehicle is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, and construct first multi-dimensional wheel hub bearing data according to the first wheel hub bearing data; S20. Input the first multi-dimensional wheel hub bearing data into the bearing life prediction model, and sequentially pass through the input layer, preprocessing layer, feature extraction layer, feature analysis layer, bearing life prediction layer and output layer of the bearing life prediction model; the preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain the second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive abnormality coefficient, and the bearing comprehensive abnormality coefficient is: ； in, is the comprehensive abnormal coefficient of the bearing; is the influencing factor of bearing vibration abnormality coefficient; is the bearing vibration abnormality coefficient; is the influencing factor of bearing temperature anomaly coefficient; is the bearing temperature anomaly coefficient; is the influencing factor of the bearing acoustic wave anomaly coefficient; is the bearing acoustic anomaly coefficient; S30. The bearing life prediction layer analyzes the bearing comprehensive abnormality coefficient and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; the output layer is used to output the wheel hub bearing service life.

2. The bearing service life prediction method based on model training according to claim 1 is characterized in that: The bearing vibration data includes outer ring vibration acceleration and inner ring vibration acceleration; the bearing temperature data includes outer ring temperature and inner ring temperature; and the bearing sound wave data includes outer ring sound wave signal.

3. The method for predicting bearing service life based on model training according to claim 2 is characterized in that: The first multi-dimensional wheel hub bearing data is: ; in, represents the first multi-dimensional wheel hub bearing data; Indicates bearing vibration data; Indicates bearing temperature data; Represents bearing acoustic wave data.

4. The method for predicting bearing service life based on model training according to claim 3 is characterized in that: The bearing life prediction model includes an input layer, a preprocessing layer, a feature extraction layer, a feature analysis layer, a bearing life prediction layer and an output layer; The preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain second multi-dimensional wheel hub bearing data; the preprocessing includes data cleaning and standardization; The feature extraction layer is used to extract the first multidimensional wheel hub bearing feature of the second multidimensional wheel hub bearing data; the feature extraction layer performs feature extraction through LSTM to obtain the first multidimensional wheel hub bearing feature, and the first multidimensional wheel hub bearing feature includes bearing vibration feature, bearing temperature feature and bearing acoustic wave feature; the bearing vibration feature includes the average vibration acceleration of the outer ring, the average vibration acceleration of the inner ring, the maximum vibration acceleration of the outer ring and the maximum vibration acceleration of the inner ring; the bearing temperature feature includes the average temperature of the outer ring and the average temperature of the inner ring; the bearing acoustic wave feature includes the peak value, mean value and standard deviation of the acoustic wave signal; The feature analysis layer is used to obtain the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing feature, and then obtain the bearing comprehensive abnormality coefficient; The bearing life prediction layer compares and analyzes the bearing comprehensive abnormality coefficient with the preset hub abnormality coefficient to obtain the hub bearing service life; The output layer is used to output the service life of the wheel hub bearing.

5. The method for predicting bearing service life based on model training according to claim 1, characterized in that: The bearing vibration abnormal coefficient is: ； in, is the bearing vibration abnormality coefficient; is the influence factor of the average vibration acceleration of the outer ring; is the average vibration acceleration of the outer ring; is the reference outer ring vibration acceleration; is the influence factor of the average vibration acceleration of the inner ring; is the average vibration acceleration of the inner ring; is the reference inner ring vibration acceleration; is the maximum vibration acceleration influence factor of the outer ring; is the maximum vibration acceleration of the outer ring; is the maximum vibration acceleration influence factor of the inner ring; is the maximum vibration acceleration of the inner ring; The bearing temperature anomaly coefficient is: ； in, is the bearing temperature anomaly coefficient; is the influence factor of the average temperature of the outer ring; is the average temperature of the outer ring; is the reference outer ring temperature; is the influence factor of the average temperature of the inner ring; is the average temperature of the inner ring; is the reference inner ring temperature; The bearing acoustic wave anomaly coefficient is: ； in, is the bearing acoustic anomaly coefficient; is the acoustic wave peak impact factor; is the peak value of the sound wave signal; is the peak value of the reference sound wave signal; is the influence factor of the mean value of the acoustic wave signal; is the mean value of the reference sound wave signal; is the influence factor of the standard deviation of the acoustic signal; is the standard deviation of the acoustic signal; is the standard deviation of the reference acoustic wave signal.

6. A bearing service life prediction method based on model training according to claim 1, characterized in that: The service life of the wheel hub bearing is: ； in, The service life of the wheel hub bearing; For normal wheel hub bearing service life; is the influencing factor of the comprehensive abnormal coefficient of the bearing; is the comprehensive abnormal coefficient of the bearing; It is the preset bearing abnormality coefficient threshold.

7. A bearing service life prediction system based on model training, used to execute any one of the bearing service life prediction methods based on model training, characterized in that: include: A data set construction module is used to obtain first wheel hub bearing data when the vehicle is running; the first wheel hub bearing data includes bearing vibration data, bearing temperature data and bearing sound wave data, and first multi-dimensional wheel hub bearing data is constructed according to the first wheel hub bearing data; The bearing life prediction model analysis module is used to input the first multi-dimensional wheel hub bearing data into the bearing life prediction model, and sequentially passes through the input layer, preprocessing layer, feature extraction layer, feature analysis layer, bearing life prediction layer and output layer of the bearing life prediction model; the preprocessing layer is used to preprocess the first multi-dimensional wheel hub bearing data to obtain second multi-dimensional wheel hub bearing data; the feature extraction layer extracts the first multi-dimensional wheel hub bearing features of the second multi-dimensional wheel hub bearing data through LSTM; the feature analysis layer obtains the bearing vibration abnormality coefficient, the bearing temperature abnormality coefficient and the bearing acoustic wave abnormality coefficient according to the first multi-dimensional wheel hub bearing features, and then obtains the bearing comprehensive abnormality coefficient, and the bearing comprehensive abnormality coefficient is: ； in, is the comprehensive abnormal coefficient of the bearing; is the influencing factor of bearing vibration abnormality coefficient; is the bearing vibration abnormality coefficient; is the influencing factor of bearing temperature anomaly coefficient; is the bearing temperature anomaly coefficient; is the influencing factor of the bearing acoustic abnormality coefficient; is the bearing acoustic anomaly coefficient; The bearing life output module is used to analyze the bearing comprehensive abnormality coefficient and the preset bearing abnormality coefficient threshold to obtain the wheel hub bearing service life; and output the wheel hub bearing service life through the output layer.

8. The bearing service life prediction system based on model training according to claim 7 is characterized in that: The bearing vibration data includes outer ring vibration acceleration and inner ring vibration acceleration; the bearing temperature data includes outer ring temperature and inner ring temperature; and the bearing sound wave data includes outer ring sound wave signal.

9. The bearing service life prediction system based on model training according to claim 7, characterized in that: The bearing vibration abnormal coefficient is: ； in, is the bearing vibration abnormality coefficient; is the influence factor of the average vibration acceleration of the outer ring; is the average vibration acceleration of the outer ring; is the reference outer ring vibration acceleration; is the influence factor of the average vibration acceleration of the inner ring; is the average vibration acceleration of the inner ring; is the reference inner ring vibration acceleration; is the maximum vibration acceleration influence factor of the outer ring; is the maximum vibration acceleration of the outer ring; is the maximum vibration acceleration influence factor of the inner ring; is the maximum vibration acceleration of the inner ring; The bearing temperature anomaly coefficient is: ； in, is the bearing temperature anomaly coefficient; is the influence factor of the average temperature of the outer ring; is the average temperature of the outer ring; is the reference outer ring temperature; is the influence factor of the average temperature of the inner ring; is the average temperature of the inner ring; is the reference inner ring temperature; The bearing acoustic wave anomaly coefficient is: ； in, is the bearing acoustic anomaly coefficient; is the acoustic wave peak impact factor; is the peak value of the sound wave signal; is the peak value of the reference sound wave signal; is the influence factor of the mean value of the acoustic wave signal; is the mean value of the reference sound wave signal; is the influence factor of the standard deviation of the acoustic signal; is the standard deviation of the acoustic signal; is the standard deviation of the reference acoustic wave signal.

10. The bearing service life prediction system based on model training according to claim 7, characterized in that: The service life of the wheel hub bearing is: ； in, The service life of the wheel hub bearing; For normal wheel hub bearing service life; is the influencing factor of the comprehensive abnormal coefficient of the bearing; is the comprehensive abnormal coefficient of the bearing; It is the preset bearing abnormality coefficient threshold.

Images