A method for evaluating performance degradation of rolling bearings based on electrostatic monitoring

Abstract

The application provides a kind of based on electrostatic monitoring's rolling bearing performance degradation evaluation method, comprising the following steps: S1: fusion electrostatic charge signal characteristics and conventional time domain, frequency domain and time-frequency domain features constitute the electrostatic monitoring system feature parameter set of bearing state characterization set.The application provides a kind of based on electrostatic monitoring's rolling bearing performance degradation evaluation method, and deep denoising self-encoder optimized by L1 sparse constraint and Dropout mechanism can better identify the nonlinear characteristics and different degradation stages of bearing;The adaptive density peak clustering model proposed can well measure the bearing health status, compared with traditional DPC and SVDD, the method proposed in this paper can identify the early degradation of bearing earlier, and has better monotonicity, robustness and trend;Compared with vibration monitoring, electrostatic monitoring has superiority in bearing performance degradation evaluation, and can detect the occurrence of degradation earlier.

Description

Technical Field

[0001] This invention relates to the field of rolling bearings, and more particularly to a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring. Background Technology

[0002] The performance of rolling bearings largely determines whether mechanical equipment operates safely and smoothly, because their degradation or failure will affect the overall performance of the machine. Under complex operating conditions, bearings are prone to performance degradation and failure as operating time accumulates. If the start time of bearing degradation can be detected as early as possible, there will be enough time to arrange maintenance reasonably, reducing the occurrence of accidents and economic losses. Therefore, it is crucial to conduct performance degradation assessment of bearings.

[0003] Conventional methods for online monitoring of rolling bearings include vibration monitoring, oil particle analysis, and temperature monitoring. Among these, vibration signal-based online monitoring technology is widely used in bearing performance degradation research. However, the internal structure of mechanical equipment is complex, and multiple excitation sources are inevitably coupled during operation, resulting in vibration signals that often carry strong path noise. Currently, electrostatic monitoring technology, with its high sensitivity, has become a promising technology for monitoring the condition of rolling bearings. This technology assesses the operating condition of the bearing by monitoring changes in the total charge in the bearing wear area. Its advantage is that it can directly measure the contact degradation of the bearing, rather than the secondary effects caused by degradation. Therefore, compared with vibration monitoring, electrostatic monitoring technology has the advantage of identifying early degradation earlier. Although electrostatic monitoring has high sensitivity, the signal characteristics of the bearing are affected by various factors. Therefore, how to construct an effective health index that reflects the operating condition of rolling bearings has become a research hotspot in the field of equipment condition monitoring.

[0004] Existing methods for assessing the performance degradation of rolling bearings do not analyze the deeper characteristics of electrostatic signal data to construct health indicators that better reflect the degradation process. Single degradation indicators contain limited fault information and are subject to numerous external interferences, and can only perform linear dimensionality reduction on the data. To map the nonlinear characteristics of bearing electrostatic data, autoencoders can be introduced to analyze and process high-dimensional features. However, the selection of parameters still involves considerable subjectivity, which can easily affect the accuracy of the results, and the effect is poor when there are large fluctuations. The method of establishing a linear regression model on a sliding window to repair fluctuations can achieve online monitoring and repair, but the linear regression model lacks the ability to represent the trend of data fluctuations.

[0005] Therefore, it is necessary to provide a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring to solve the above-mentioned technical problems. Summary of the Invention

[0006] This invention provides a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring. It addresses the shortcomings of existing methods, which fail to analyze the deeper characteristics of electrostatic signal data and construct health indicators that better reflect the degradation process. Single degradation indicators contain limited fault information and are susceptible to external interference, only allowing for linear dimensionality reduction. While autoencoders can be introduced to analyze high-dimensional features and map the nonlinear characteristics of bearing electrostatic data, parameter selection remains highly subjective, potentially affecting accuracy and performing poorly with large fluctuations. A method using a sliding window to establish a linear regression model is proposed to mitigate these fluctuations. However, while sliding windows enable online monitoring and repair, linear regression models lack the ability to characterize data fluctuation trends.

[0007] To solve the above-mentioned technical problems, the present invention provides a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring, comprising the following steps S1: fusing electrostatic charge signal characteristics with conventional time-domain, frequency-domain, and time-frequency-domain characteristics to form a set of characteristic parameters of the electrostatic monitoring system characterizing the bearing state;

[0008] S2: The deep denoising autoencoder optimized by introducing L1 regularization and Dropout mechanism performs nonlinear mapping dimensionality reduction on high-dimensional feature sets. IDAE can not only improve the feature selection ability, but also prevent overfitting in deep networks.

[0009] S3: By combining optimization and entropy analysis, the adaptive determination of DPC cluster centers is achieved, thereby avoiding the "chain effect" caused by incorrect cluster center allocation. Therefore, a model is established through the improved ADPC algorithm to measure the health status of the bearing, and the Mahalanobis distance MD between the test set and the cluster center is calculated to quantitatively characterize the performance degradation of the rolling bearing.

[0010] S4: Combining the 3δ principle, a multinomial regression model is established on the sliding window to correct the spurious fluctuations of the MD index and obtain the improved index IMD that can accurately characterize the degradation process of rolling bearings. The bearing performance degradation assessment is achieved by analyzing the relationship curve between the health threshold and IMD.

[0011] Preferably, the algorithm is improved in three ways. First, a 5-layer deep denoising autoencoder is built, and L1 regularization is introduced to improve the data feature extraction capability by utilizing its sparsity. A Dropout mechanism is introduced to prevent overfitting caused by the deepening of the network. Second, to solve the problem that the clustering effect of the traditional density peak clustering algorithm is too different from the expected value due to the manual selection of cluster centers and cutoff distances, an adaptive density peak clustering algorithm is proposed, and the original index calculation method is improved by Mahalanobis distance. Third, a multinomial regression model is established on the sliding window to repair spurious fluctuations.

[0012] Preferably, the method for validating the performance degradation assessment of the rolling bearing involves conducting an accelerated fatigue life test on the bearing using an ABLT-1A testing machine. This monitoring platform includes a loading system, a transmission system, a lubrication system, an electrical control system, a computer monitoring system, a test head, and a test head base. The test head includes a test device, an electrostatic sensor mounting position, and a signal acquisition system. Electrostatic sensors ES1 and ES2 are used to monitor the electrostatic signals of the bearing.

[0013] Preferably, the L1 regularization is the sum of the absolute values ​​of each element in the weight vector. By guiding the model to pay more attention to those weights with larger absolute values ​​and continuously adjusting the feature weights, the weights of some features become 0, thus achieving the effect of sparsity.

[0014] Preferably, the principle by which the L1 algorithm generates a sparse model is as follows:

[0015]

[0016] In the formula, J0 is the loss function of the original model, and λ is the regularization coefficient;

[0017] In machine learning, L1 regularization is commonly used for sparse models. Therefore, in the loss function of a deep denoising autoencoder, adding an L1 norm penalty term to minimize the sum of the L1 norms of the objective function and the features achieves feature selection. This not only improves the model's robustness and generalization ability but also enhances feature extraction for electrostatic signals. The loss function of the deep denoising autoencoder after adding L1 regularization is:

[0018]

[0019] Adding L1 regularization to the loss function makes it easier to sparsify the data, reduces the difficulty of the learning task, facilitates automatic feature selection, makes the model interpretable, and provides good data for degradation evaluation models.

[0020] Preferably, Dropout is a technique that can help reduce overfitting in training neural networks. In this study, the deep denoising autoencoder is composed of multiple layers of denoising autoencoders. As the complexity of the neural network increases, overfitting is likely to occur during model training. Therefore, the Dropout mechanism is introduced to solve the overfitting problem, so as to prevent complex co-adaptation on the training data and avoid repeatedly extracting the same features.

[0021] During the training of a deep learning network, Dropout temporarily and randomly discards hidden neurons with a set probability. In the next training iteration, these neurons are recombined according to the same set probability. After multiple Dropout processes, the original network becomes a combination of several thinner, different neural networks. The calculation process of Dropout in a neural network can be represented as follows:

[0022] y (l+1) =σ(W (l)T x (l) +b (l) )1≤l≤L-1

[0023] x (l) =dropout(y (l) ,p) 2≤l≤L

[0024] x (l+1) =(1-p)σ(W) (l)T x (l) +b (l) )

[0025] Where l represents the number of neuron layers, L is the total number of neuron layers, and p represents the dropout rate;

[0026] By training the same set of data with different neural networks and averaging the outputs of each network, overfitting can be reduced. In practice, the activation output of some hidden neurons can be set to 0, so that these neurons are not included in the forward propagation training process, thereby achieving the optimization of the deep denoising autoencoder through dropout.

[0027] Preferably, the abnormal bearing data accounts for a very small portion of the total lifecycle data. There is an extreme imbalance between normal and abnormal samples. The density peak clustering algorithm efficiently allocates sample points and removes outliers by finding density peak points. It is suitable for cluster analysis of large-scale bearing data. ADPC is used to perform unsupervised learning on the bearing data to construct the health status of the bearing.

[0028] Preferably, the Density Peak Fast Search Clustering Algorithm (DPC) is a novel density-based clustering algorithm. It is based on the following assumptions: the local density of the cluster center is higher than the local density of other points in the same cluster, and there is a large distance between the cluster center and any point with a higher local density. When modeling the normal state of electrostatic data, the cluster center has a large local density value and a large relative distance value.

[0029] To find cluster centers that simultaneously satisfy both conditions, this algorithm introduces the definitions of local density and relative distance. Traditional density peak clustering algorithms use Euclidean distance to reflect the relationship between two points. However, Euclidean distance is easily affected by feature scale. Therefore, Mahalanobis distance is introduced to support the distance calculation of DPC, which can effectively integrate the characteristics of multiple feature vectors. The formula for calculating the Mahalanobis distance between data points is:

[0030]

[0031] In the formula, S -1 Given the covariance matrix, the local density calculation process is as follows:

[0032]

[0033] When x < 0, χ(x) = 1; otherwise, χ(x) = 0. ij d is the Mahalanobis distance between data points i and j. c To cut off the distance and avoid the situation where the local density of data points is the same, a Gaussian function is used instead of the formula, as shown in the formula;

[0034]

[0035] The relative distance calculation process is as follows:

[0036]

[0037] In the formula, S represents the set of data, which contains n samples; Represents density ratio ρ i A set of large points; Represents the empty set;

[0038] Choose one that can comprehensively consider the local density ρ i and relative distance δ i The parameter μ i :

[0039] μ i =ρ i δ i (3)

[0040] In the DPC algorithm, cluster centers have a larger local density ρ compared to other points. i and relative distance δ i Therefore, by using formula (18), the μ of the cluster centers is increased. i μ of other points i The difference in values ​​is used to calculate μ. i Then, μ i Sort the values ​​in descending order, then calculate μ. iThe formula for calculating the downward trend of the value is:

[0041]

[0042] In the formula, μ i This is the current value of μ; μ i-1 and μ i+1 These represent the μ values ​​of the previous and next time moments, respectively. After obtaining the downward trend, the point with the largest downward trend and the previous few points are taken as the cluster center.

[0043] Traditional DPC algorithms typically obtain d based on experience or experimental verification. c The value of d introduces considerable uncertainty into the clustering results. To improve clustering effectiveness, d can be determined by comprehensively measuring the differences between the data. c To determine the value of d, this paper uses the entropy method to constrain d. c In information theory, the amount of information contained in an event is measured by information entropy. In comprehensive evaluation, if the numerical differences of the same indicator are small, that is, the probability distribution is relatively uniform, it means that the event "a certain value belongs to a certain evaluated object" has greater uncertainty and higher information entropy. Therefore, the indicator is given a lower weight. Conversely, if the numerical differences of the same indicator are large, the indicator has a higher weight.

[0044] The formula for calculating information entropy is:

[0045]

[0046] Where n represents the number of samples in dataset S; p i ρ at point i i ×δ i The proportion of all points, p i The possible values ​​are as follows:

[0047]

[0048] d when taking the minimum value c The value of d when used as the final clustering value c The value is used to replace empirical methods or experimental verification, thereby achieving adaptive selection of parameters.

[0049] Preferably, the performance of the bearing is in a state of continuous degradation during operation. Due to some external factors, the degradation characteristics may fluctuate abnormally, which will seriously affect the description of the degradation state of the bearing by the degradation index. Therefore, if there are false fluctuations, they need to be repaired before they can be used as indicators for subsequent degradation assessment.

[0050] This paper constructs a sliding window based on health indicators, sets the window step size to 1, and the window length to 50. The fluctuation threshold θ is set according to the 3σ principle of the data within the sliding window and polynomial regression. Its expression is:

[0051]

[0052] Among them, the values of ω0, ω1, and ω2 are determined using the least squares method; x i is the time series of health indicators, and MD i is the distance value between the measured data and the clustering center;

[0053] The health indicator IMD after repairing false fluctuations is determined according to the sliding window fluctuation threshold θ i , and its expression is:

[0054]

[0055] Among them, min{MD(i)} and max{MD(i)} represent the upper and lower limits of MD(i) respectively;

[0056]

[0057] In the formula, μ represents the mean of MD(i); σ represents the standard deviation of MD(i);

[0058] It is judged whether there are false fluctuations in the bearing health indicator according to the above formula and repaired. When MD(i) ≤ min{MD(i)}, MD(i) moves to the average value to obtain a new predicted value IMD; when min{MD(i)} < MD(i) < max{MD(i)}, MD(i) is replaced with the value predicted by polynomial regression; when MD(i) ≥ max{MD(i)}, MD(i) moves to max{MD(i)}, and then a new predicted value is calculated. By establishing a regression model on the sliding window, the false fluctuations existing in the health indicator can be effectively repaired, so as to obtain a health indicator with better performance in characterizing the degradation trend.

[0059] Preferably, the experimental device includes an experimental device main body 100. A base 200 is fixedly installed at the bottom of the experimental device main body 100. Hydraulic cylinders 300 are fixedly installed around the top of the base 200. The bottom end of the output shaft of the hydraulic cylinder 300 is fixedly connected to a support block 400. Movable wheels 500 are arranged around the bottom of the base 200.

[0060] Compared with the related technology, a rolling bearing performance degradation evaluation method based on electrostatic monitoring provided by the present invention has the following beneficial effects:

[0061] This invention provides a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring. It utilizes a deep denoising autoencoder with L1 sparsity constraints to achieve nonlinear dimensionality reduction of a high-dimensional feature set, eliminating data redundancy. A Dropout mechanism is introduced to prevent model overfitting, constructing an improved degradation model, ADPC. Mahalanobis distance is used to calculate the length of data points, and the cutoff distance satisfying the electrostatic data distribution characteristics is found using the entropy method. An index that simultaneously reflects local density and relative distance is constructed, with the point with the largest trend value serving as the cluster center. Compared to traditional DPC, the improved model can adaptively select cluster centers and cutoff distances, reducing errors from empirical selection. Furthermore, Mahalanobis distance, compared to the original Euclidean distance, is unaffected by dimensions and considers the correlation between variables. To correct spurious fluctuations in the constructed degradation index during online monitoring, a [further details are needed]. A sliding window approach, based on fundamental principles, establishes a multinomial regression model within the window. This allows for the addition of new variables to address nonlinear problems. Electrostatic charge signal characteristics are used to reflect the effective value of charge in the signal, serving as a valuable supplement to conventional time-domain and frequency-domain indicators in electrostatic monitoring. Compared to the traditional dimensionality reduction method PCA, the deep denoising autoencoder optimized through L1 sparsity constraints and Dropout mechanisms can better identify the nonlinear characteristics and different degradation stages of bearings. The proposed adaptive density peak clustering model effectively measures bearing health. Compared to traditional DPC and SVDD, the proposed method can identify early bearing degradation earlier and exhibits better monotonicity, robustness, and trend analysis. Compared to vibration monitoring, electrostatic monitoring is superior in bearing performance degradation assessment, enabling earlier detection of degradation. Attached Figure Description

[0062] Figure 1 A flowchart illustrating the first embodiment of a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention;

[0063] Figure 2 A schematic diagram of the DDAE network structure for a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention;

[0064] Figure 3 A schematic diagram of a network using Dropout to improve a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring, provided by the present invention.

[0065] Figure 4 A schematic diagram of the DPC principle for a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention;

[0066] Figure 5 A schematic diagram illustrating the principle of sliding window feature repair in a rolling bearing performance degradation assessment method based on electrostatic monitoring, provided by this invention.

[0067] Figure 6 A schematic diagram of the overall process for a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring, provided by the present invention;

[0068] Figure 7 A schematic diagram of the electrostatic monitoring principle for a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention;

[0069] Figure 8 This is a schematic diagram of the second embodiment of a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention.

[0070] The diagram is labeled as follows: 100, main body of the experimental device; 200, base; 300, hydraulic cylinder; 400, support block; 500, caster wheel. Detailed Implementation

[0071] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0072] First Embodiment

[0073] Please refer to the following: Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 , Figure 6 and Figure 7 ,in, Figure 1 A flowchart illustrating the first embodiment of a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention; Figure 2 A schematic diagram of the DDAE network structure for a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention; Figure 3 A schematic diagram of a network using Dropout to improve a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring, provided by the present invention. Figure 4 A schematic diagram of the DPC principle for a rolling bearing performance degradation assessment method based on electrostatic monitoring provided by the present invention; Figure 5 A schematic diagram illustrating the principle of sliding window feature repair in a rolling bearing performance degradation assessment method based on electrostatic monitoring, provided by this invention. Figure 6 A schematic diagram of the overall process for a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring, provided by the present invention; Figure 7 This invention provides a schematic diagram of the electrostatic monitoring principle for a rolling bearing performance degradation assessment method based on electrostatic monitoring. The method includes the following steps: S1: fusing electrostatic charge signal characteristics with conventional time-domain, frequency-domain, and time-frequency-domain characteristics to construct a set of characteristic parameters for the electrostatic monitoring system characterizing the bearing state;

[0074] S2: The deep denoising autoencoder optimized by introducing L1 regularization and Dropout mechanism performs nonlinear mapping dimensionality reduction on high-dimensional feature sets. IDAE can not only improve the feature selection ability, but also prevent overfitting in deep networks.

[0075] S3: By combining optimization and entropy analysis, the adaptive determination of DPC cluster centers is achieved, thereby avoiding the "chain effect" caused by incorrect cluster center allocation. Therefore, a model is established through the improved ADPC algorithm to measure the health status of the bearing, and the Mahalanobis distance MD between the test set and the cluster center is calculated to quantitatively characterize the performance degradation of the rolling bearing.

[0076] S4: Combining the 3δ principle, a multinomial regression model is established on the sliding window to correct the spurious fluctuations of the MD index and obtain the improved index IMD that can accurately characterize the degradation process of rolling bearings. The bearing performance degradation assessment is achieved by analyzing the relationship curve between the health threshold and IMD.

[0077] The algorithm is improved in three ways. First, a 5-layer deep denoising autoencoder is built, and L1 regularization is introduced to improve the data feature extraction capability by utilizing its sparsity. The Dropout mechanism is introduced to prevent overfitting caused by the deepening of the network. Second, to solve the problem that the traditional density peak clustering algorithm causes the clustering effect to deviate too much from the expected value due to the manual selection of cluster centers and cutoff distances, an adaptive density peak clustering algorithm is proposed, and the original index calculation method is improved by Mahalanobis distance. Third, a multinomial regression model is built on the sliding window to repair spurious fluctuations.

[0078] The method for validating the performance degradation assessment of the rolling bearing involves conducting accelerated fatigue life tests on the bearing using an ABLT-1A testing machine. This monitoring platform includes a loading system, a transmission system, a lubrication system, an electrical control system, a computer monitoring system, a test head, and a test head base. The test head includes a test apparatus, an electrostatic sensor mounting position, and a signal acquisition system. Electrostatic sensors ES1 and ES2 are used to monitor the electrostatic signals of the bearing.

[0079] The L1 regularization is the sum of the absolute values ​​of each element in the weight vector. By guiding the model to pay more attention to the weights with larger absolute values ​​and continuously adjusting the feature weights, the weights of some features become 0, thus achieving the effect of sparsity.

[0080] The principle behind L1 generating sparse models is as follows:

[0081]

[0082] In the formula, J0 is the loss function of the original model, and λ is the regularization coefficient;

[0083] In machine learning, L1 regularization is commonly used for sparse models. Therefore, in the loss function of a deep denoising autoencoder, adding an L1 norm penalty term to minimize the sum of the L1 norms of the objective function and the features achieves feature selection. This not only improves the model's robustness and generalization ability but also enhances feature extraction for electrostatic signals. The loss function of the deep denoising autoencoder after adding L1 regularization is:

[0084]

[0085] Adding L1 regularization to the loss function makes it easier to sparsify the data, reduces the difficulty of the learning task, facilitates automatic feature selection, makes the model interpretable, and provides good data for degradation evaluation models.

[0086] Dropout is a technique that can help reduce overfitting in training neural networks. In this study, the deep denoising autoencoder is composed of multiple layers of denoising autoencoders. As the complexity of the neural network increases, overfitting is likely to occur during model training. Therefore, the Dropout mechanism is introduced to solve the overfitting problem, so as to prevent complex co-adaptation on the training data and avoid repeatedly extracting the same features.

[0087] During the training of a deep learning network, Dropout temporarily and randomly discards hidden neurons with a set probability. In the next training iteration, these neurons are recombined according to the same set probability. After multiple Dropout processes, the original network becomes a combination of several thinner, different neural networks. The calculation process of Dropout in a neural network can be represented as follows:

[0088] y (l+1) =σ(W (l)T x (l) +b (l) ) 1≤l≤L-1

[0089] x (l) =dropout(y (l) ,p) 2≤l≤L

[0090] x (l+1) =(1-p)σ(W) (l)T x (l) +b (l) )

[0091] Where l represents the number of neuron layers, L is the total number of neuron layers, and p represents the dropout rate;

[0092] By training the same set of data with different neural networks and averaging the outputs of each network, overfitting can be reduced. In practice, the activation output of some hidden neurons can be set to 0, so that these neurons are not included in the forward propagation training process, thereby achieving the optimization of the deep denoising autoencoder through dropout.

[0093] The abnormal bearing data constitutes only a small portion of the total lifecycle data. There is an extreme imbalance between normal and abnormal samples. The density peak clustering algorithm efficiently allocates sample points and removes outliers by finding density peak points. It is suitable for cluster analysis of large-scale bearing data. ADPC is used to perform unsupervised learning on the bearing data to construct the health status of the bearing.

[0094] The Density Peak Fast Search Clustering Algorithm (DPC) is a novel density-based clustering algorithm. It is based on the following assumptions: the local density of cluster centers is higher than that of other points in the same cluster, and there is a large distance between cluster centers and any points with higher local density. When modeling electrostatic data under normal conditions, cluster centers have large local density values ​​and relative distance values.

[0095] To find cluster centers that simultaneously satisfy both conditions, this algorithm introduces the definitions of local density and relative distance. Traditional density peak clustering algorithms use Euclidean distance to reflect the relationship between two points. However, Euclidean distance is easily affected by feature scale. Therefore, Mahalanobis distance is introduced to support the distance calculation of DPC, which can effectively integrate the characteristics of multiple feature vectors. The formula for calculating the Mahalanobis distance between data points is:

[0096]

[0097] In the formula, S -1 Given the covariance matrix, the local density calculation process is as follows:

[0098]

[0099] When x < 0, χ(x) = 1; otherwise, χ(x) = 0. ij d is the Mahalanobis distance between data points i and j. c To cut off the distance and avoid the situation where the local density of data points is the same, a Gaussian function is used instead of the formula, as shown in the formula;

[0100]

[0101] The relative distance calculation process is as follows:

[0102]

[0103] In the formula, S represents the set of data, which contains n samples; Represents density ratio ρ i A set of large points; Represents the empty set;

[0104] Choose one that can comprehensively consider the local density ρ i and relative distance δ i The parameter μ i :

[0105] μ i =ρ i δ i (5)

[0106] In the DPC algorithm, cluster centers have a larger local density ρ compared to other points. i and relative distance δ i Therefore, by using formula (18), the μ of the cluster centers is increased. i μ of other points i The difference in values ​​is used to calculate μ. i Then, μ i Sort the values ​​in descending order, then calculate μ. i The formula for calculating the downward trend of the value is:

[0107]

[0108] In the formula, μ i This is the current value of μ; μ i-1 and μ i+1 These represent the μ values ​​of the previous and next time moments, respectively. After obtaining the downward trend, the point with the largest downward trend and the previous few points are taken as the cluster center.

[0109] Traditional DPC algorithms typically obtain d based on experience or experimental verification. c The value of d introduces considerable uncertainty into the clustering results. To improve clustering effectiveness, d can be determined by comprehensively measuring the differences between the data. c To determine the value of d, this paper uses the entropy method to constrain d. c In information theory, the amount of information contained in an event is measured by information entropy. In comprehensive evaluation, if the numerical differences of the same indicator are small, that is, the probability distribution is relatively uniform, it means that the event "a certain value belongs to a certain evaluated object" has greater uncertainty and higher information entropy. Therefore, the indicator is given a lower weight. Conversely, if the numerical differences of the same indicator are large, the indicator has a higher weight.

[0110] The formula for calculating information entropy is:

[0111]

[0112] Where n represents the number of samples in dataset S; p i ρ at point i i ×δ i The proportion of all points, p i The possible values ​​are as follows:

[0113]

[0114] d when taking the minimum value c The value of d when used as the final clustering value c The value is used to replace empirical methods or experimental verification, thereby achieving adaptive selection of parameters.

[0115] During operation, the performance of the bearing is in a state of continuous decline. Due to some external factors, the decline characteristics may fluctuate abnormally, which will seriously affect the description of the bearing's degradation state by the degradation index. Therefore, if there are false fluctuations, they need to be repaired before they can be used as indicators for subsequent degradation assessment.

[0116] This paper constructs a sliding window based on health indicators, setting the window step size to 1 and the window length to 50. The fluctuation threshold θ is set using the 3σ principle of the data within the sliding window and multinomial regression, and its expression is:

[0117]

[0118] The values ​​of ω0, ω1, and ω2 are determined using the least squares method; x i For time series of health indicators, MD i This represents the distance between the data to be tested and the cluster centers.

[0119] The health indicator IMD after correcting for spurious fluctuations is determined based on the sliding window fluctuation threshold θ. i Its expression is:

[0120]

[0121] Where min{MD(i)} and max{MD(i)} represent the upper and lower limits of MD(i), respectively;

[0122]

[0123] In the formula, μ represents the mean of MD(i); σ represents the standard deviation of MD(i);

[0124] Judge whether there is false fluctuation in the bearing health index according to the above formula and repair it. When MD(i) ≤ min{MD(i)}, MD(i) moves to the average value to obtain a new predicted value IMD; when min{MD(i)} < MD(i) < max{MD(i)}, MD(i) is replaced by the value predicted by polynomial regression; when MD(i) ≥ max{MD(i)}, MD(i) moves to max{MD(i)}, and then a new predicted value is calculated. By establishing a regression model on the sliding window, the false fluctuation existing in the health index can be effectively repaired, so as to obtain a health index with better performance in characterizing the degradation trend.

[0125] Performance degradation assessment based on deep denoising autoencoder and adaptive density peak clustering:

[0126] A method of deep denoising autoencoder and adaptive peak density clustering is proposed for bearing performance degradation assessment. The electrostatic signals of the bearing are input into the ADPC for feature fusion after dimensionality reduction by IDDAE, and the output result is used as the health index for performance degradation assessment. The health index is the distance value IMD between the待测样本 (to-be-tested sample) and the clustering center after ADPC clustering. By judging the relationship between the IMD index and the threshold, it is determined whether the bearing is in a degraded state. If it exceeds the threshold, it means the bearing has degraded, otherwise the bearing is in a normal state.

[0127] The bearing performance degradation assessment is divided into an offline stage and an online stage. Combining the method proposed in the paper, the specific implementation steps of these two stages are as follows:

[0128] (1) Offline stage

[0129] A. Collect the electrostatic signals of the rolling bearing in the normal working state, and extract the original features from the bearing electrostatic signals, including 12 time-domain features, 3 frequency-domain features, 8 time-frequency domain features and 2 electrostatic charge signal features. Among them, the electrostatic charge signal features are specific parameters of electrostatic monitoring, which are divided into the activity rate level (Active level, AL) and the event rate (EventRate, ER).

[0130] The definition of the activity rate level is:

[0131]

[0132] In the formula, Q f is the electrostatic charge signal, T is the intercepted signal length, usually taken as 1s, N is the number of sampling points of Q f (t) within T after discrete sampling, and Q ... (It seems the content is incomplete here. Please check and provide the full text if needed.) fn is Q f(t) The electrostatic charge signal after discrete sampling, the physical meaning of the activity level AL is the effective value of the amount of induced charge excited on the sensor by charged particles in the air path of the electrostatic sensor in the detection space within a certain time interval.

[0133] The event rate ER is defined as:

[0134]

[0135] In the formula, M is the number of points exceeding K times the value of AL(t) in the electrostatic signal interval (1s signal) corresponding to each point A, K is an empirical multiple, N is the total number of samples in the signal segment, and the physical meaning of the event rate can be understood as the percentage of the number of amplitude anomalies to the total number of sampling points within a certain time interval.

[0136] B. The initially extracted raw features are input into IDDAE for greedy training, learning the nonlinear features of the bearing layer by layer. In the IDDAE network used for feature dimensionality reduction, a network with a hidden layer structure of 5 layers is set up, and the neurons in the hidden layer are configured as 128, 64, 32, 16, and 3 respectively. The learning rate is set to 0.001, the Gaussian noise coefficient is set to 0.2, the sparsity coefficient is set to 0.05, and the Dropout rate is set to 0.2.

[0137] C. The feature vectors of rolling bearings under normal conditions extracted from the IDDAE model are used as training samples for the ADPC model. The data distribution shape and cluster center of the normal samples are found by calculating the local density and relative distance.

[0138] Online phase:

[0139] A. Similar to the offline method, extract the original features of the rolling bearing electrostatic signal.

[0140] B. Input the original features into IDDAE to reconstruct the feature vector.

[0141] C. In the offline stage, an ADPC model for normal samples has been trained, which can learn the distribution of data under normal conditions and use it as a benchmark. Therefore, in the online stage, the difference in data distribution between the test data and the benchmark is displayed by Mahalanobis distance, thereby obtaining the index MD that measures the degree of bearing degradation.

[0142] D. Because the measurement environment of online data is subject to a lot of external interference, abnormal fluctuations in the data are inevitable. In order to eliminate the impact of data fluctuations on the accuracy of the results, a regression model is established on the sliding window to correct the spurious fluctuations in the degradation trend curve of the rolling bearing and obtain a new health indicator, IMD.

[0143] E. Performance degradation assessment of rolling bearings is achieved by analyzing the relationship curve between health threshold and health index IMD.

[0144] Principle of electrostatic monitoring for rolling bearings:

[0145] Static electricity occurs in the wear area of ​​rolling bearings, and electrostatic monitoring technology is used to monitor the total amount of static charge in the wear area in real time. Figure 6 The diagram illustrates the principle of electrostatic monitoring in the wear area. When a charged particle passes through the effective area in front of the probe of the electrostatic sensor, the electric field lines terminate on the probe surface. Due to electrostatic induction, the electrostatic charge source on the probe surface attracts charges of different polarities. At this time, charges of the same polarity will accumulate at the other end of the probe surface. A signal conditioning circuit connected to the other end of the probe can convert the induced charge into a measurable voltage signal. Therefore, the operating status of the rolling bearing can be monitored in real time through the electrostatic signal.

[0146] When performing electrostatic monitoring on the wear area of ​​a rolling bearing, the induced charge of charged particles on the sensor probe is denoted as Q(t), and the output induced voltage is denoted as U(t). The default initial system state is 0. After Laplace transformation, the charge Q(t) and voltage U(t) are Q(s) and U(s), respectively, and they satisfy the following relationship:

[0147] U(s)=R·s·Q(s) / (R·C·s+1)

[0148] In the formula, R is the equivalent resistance of the detection circuit; s is the complex frequency at time t; and C is the equivalent capacitance of the circuit.

[0149] When R·C·s≤1, the electrostatic signal monitoring model can be equivalent to:

[0150]

[0151] The original voltage acquired by the rolling bearing acquisition system is proportional to the first derivative of the surface charge induced by the electrostatic sensor, indicating that the electrostatic signal can directly reflect the operating status of the rolling bearing.

[0152] The working principle of the rolling bearing performance degradation assessment method based on electrostatic monitoring provided by this invention is as follows:

[0153] During operation, a high-dimensional feature set is first constructed based on the features in the time domain, frequency domain, and time-frequency domain, combined with the features of electrostatic charge signal. The feature set is then subjected to dimensionality reduction processing through IDDAE.

[0154] Second, the optimal cluster center is selected by using comprehensive index optimization and entropy method, a health status model is constructed to measure the bearing operating status, and Mahalanobis distance is designed to represent the degree of difference between health data and test data.

[0155] Third, by integrating regression models and principles within a sliding window to correct spurious fluctuations, we obtain the IMD index, which characterizes the degree of bearing degradation. Finally, we analyze the relationship between IMD and a threshold to achieve bearing performance degradation assessment.

[0156] The experimental results show that: (1) The electrostatic charge signal characteristics are used to reflect the effective value of the charge in the signal. In electrostatic monitoring, it can be used as a good supplement to the conventional time domain and frequency domain indicators; (2) Compared with the traditional dimensionality reduction method PCA, the deep noise reduction autoencoder optimized by L1 sparsity constraint and Dropout mechanism can better identify the nonlinear characteristics and different degradation stages of the bearing; (3) The proposed adaptive density peak clustering model can well measure the health status of the bearing. Compared with the traditional DPC and SVDD, the method proposed in this paper can identify the early degradation of the bearing earlier and has better monotonicity, robustness and trend; (4) Compared with vibration monitoring, electrostatic monitoring has the advantage in bearing performance degradation assessment and can detect the occurrence of degradation earlier.

[0157] Compared with related technologies, the rolling bearing performance degradation assessment method based on electrostatic monitoring provided by this invention has the following beneficial effects:

[0158] This invention provides a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring. It utilizes a deep denoising autoencoder with L1 sparsity constraints to achieve nonlinear dimensionality reduction of a high-dimensional feature set, eliminating data redundancy. A Dropout mechanism is introduced to prevent model overfitting, constructing an improved degradation model, ADPC. Mahalanobis distance is used to calculate the length of data points, and the cutoff distance satisfying the electrostatic data distribution characteristics is found using the entropy method. An index that simultaneously reflects local density and relative distance is constructed, with the point with the largest trend value serving as the cluster center. Compared to traditional DPC, the improved model can adaptively select cluster centers and cutoff distances, reducing errors from empirical selection. Furthermore, Mahalanobis distance, compared to the original Euclidean distance, is unaffected by dimensions and considers the correlation between variables. To correct spurious fluctuations in the constructed degradation index during online monitoring, a [further details are needed]. A sliding window approach, based on fundamental principles, establishes a multinomial regression model within the window. This allows for the addition of new variables to address nonlinear problems. Electrostatic charge signal characteristics are used to reflect the effective value of charge in the signal, serving as a valuable supplement to conventional time-domain and frequency-domain indicators in electrostatic monitoring. Compared to the traditional dimensionality reduction method PCA, the deep denoising autoencoder optimized through L1 sparsity constraints and Dropout mechanisms can better identify the nonlinear characteristics and different degradation stages of bearings. The proposed adaptive density peak clustering model effectively measures bearing health. Compared to traditional DPC and SVDD, the proposed method can identify early bearing degradation earlier and exhibits better monotonicity, robustness, and trend analysis. Compared to vibration monitoring, electrostatic monitoring is superior in bearing performance degradation assessment, enabling earlier detection of degradation.

[0159] Second Embodiment

[0160] Please refer to the following: Figure 8 Based on the first embodiment of this application, which provides a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring, the second embodiment of this application proposes another method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring. The second embodiment is merely a preferred embodiment of the first embodiment, and the implementation of the second embodiment will not affect the separate implementation of the first embodiment.

[0161] Specifically, the second embodiment of this application provides a different method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring. The experimental device includes an experimental device body 100, a base 200 fixedly installed at the bottom of the experimental device body 100, hydraulic cylinders 300 fixedly installed around the top of the base 200, a support block 400 fixedly connected to the bottom end of the output shaft of the hydraulic cylinder 300, and casters 500 provided around the bottom of the base 200.

[0162] The working principle of the rolling bearing performance degradation assessment method based on electrostatic monitoring provided by this invention is as follows:

[0163] During operation, the output shaft of the hydraulic cylinder 300 is first controlled by an external power source to move up or down, adjusting the height between the support block 400 and the moving wheel 500.

[0164] Compared with related technologies, the rolling bearing performance degradation assessment method based on electrostatic monitoring provided by this invention has the following beneficial effects:

[0165] This invention provides a method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring. By controlling the output shaft of the hydraulic cylinder 300 to move up or down through an external power source, the height between the support block 400 and the moving wheel 500 is adjusted. When the support block 400 supports the ground and the moving wheel 500 is suspended in the air, the stability of the experimental device body 100 is maintained. When the moving wheel 500 supports the ground and the support block 400 is suspended in the air, it is convenient for the user to push the experimental device body 100 for transportation, thus improving the convenience of transporting this device.

[0166] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring, characterized in that, Includes the following steps: S1: The electrostatic charge signal characteristics are integrated with conventional time-domain, frequency-domain, and time-frequency-domain characteristics to form a set of characteristic parameters for electrostatic monitoring systems that characterize the bearing state; S2: The deep denoising autoencoder optimized by introducing L1 regularization and Dropout mechanism performs nonlinear mapping dimensionality reduction on high-dimensional feature sets. IDAE can not only improve the feature selection ability, but also prevent overfitting in deep networks. S3: Through comprehensive optimization and entropy analysis, the adaptive determination of DPC cluster centers is achieved. A model is established using the improved ADPC algorithm to measure the health status of the bearing. The Mahalanobis distance MD between the test set and the cluster centers is calculated to quantitatively characterize the performance degradation of the rolling bearing. S4: Combination The principle is to establish a multinomial regression model on a sliding window to correct spurious fluctuations in the MD index, obtain an improved index IMD that can accurately characterize the degradation process of rolling bearings, and achieve bearing performance degradation assessment by analyzing the relationship curve between the health threshold and IMD.

2. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 1, characterized in that, The algorithm is improved in three ways. The first is to build a 5-layer deep denoising autoencoder, introduce L1 regularization, and use its sparsity to improve the data feature extraction capability. The Dropout mechanism is introduced to prevent overfitting caused by the deepening of the network. The second is to propose an adaptive density peak clustering algorithm and improve the original index calculation method by Mahalanobis distance. The third is to build a multinomial regression model on a sliding window to repair spurious fluctuations.

3. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 1, characterized in that, The method for validating the performance degradation assessment of the rolling bearing involves conducting accelerated fatigue life tests on the bearing using an ABLT-1A testing machine. The monitoring platform includes a loading system, a transmission system, a lubrication system, an electrical control system, a computer monitoring system, a test head, and a test head base. The test head includes a test apparatus, an electrostatic sensor mounting position, and a signal acquisition system. Electrostatic sensors ES1 and ES2 are used to monitor the electrostatic signals of the bearing.

4. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 1, characterized in that, The L1 regularization is the sum of the absolute values ​​of each element in the weight vector. By guiding the model to pay more attention to the weights with larger absolute values ​​and continuously adjusting the feature weights, the weights of some features become 0, thus achieving the effect of sparsity.

5. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 1, characterized in that, The principle behind L1 generating sparse models is as follows: In the formula, The loss function of the original model. It is the regularization coefficient; L1 regularization is used in sparse models. Therefore, an L1 norm penalty term is added to the loss function of the deep denoising autoencoder. Feature selection is achieved by minimizing the sum of the L1 norms of the objective function and the features. The loss function of the deep denoising autoencoder after adding L1 regularization is: 。 6. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 1, characterized in that, During the training of a deep learning network, Dropout temporarily and randomly discards hidden neurons with a set probability. In the next training iteration, these neurons are recombined according to the same set probability. After multiple Dropout processes, the original network becomes a combination of several thinner, different neural networks. The calculation process of Dropout in a neural network is represented as follows: in, Indicates the number of the neuron layer. It is the total number of neuron layers. Indicates the dropout rate; By training the same set of data with different neural networks and averaging the outputs of each network, overfitting can be reduced. The activation outputs of some hidden neurons are set to 0, so that these neurons are not included in the forward propagation training process, thereby optimizing the deep denoising autoencoder through dropout.

7. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 1, characterized in that, Density peak clustering algorithm efficiently allocates sample points and removes outliers by finding density peak points. It is suitable for cluster analysis of large-scale bearing data. ADPC is used to perform unsupervised learning on bearing data to construct the health status of bearings.

8. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 2, characterized in that, The density peak fast search clustering algorithm DPC is a novel density-based clustering algorithm. To find cluster centers that simultaneously satisfy both conditions, the algorithm introduces the definitions of local density and relative distance, and uses Mahalanobis distance to support the distance calculation of DPC. The formula for calculating the Mahalanobis distance between data points is: In the formula, Given the covariance matrix, the local density calculation process is as follows: when hour, ,otherwise ; Data points and Mahalanobis distance between them To cut off the distance and avoid the situation where the local density of data points is the same, a Gaussian function is used instead of the formula, as shown in the formula; The relative distance calculation process is as follows: In the formula, A collection representing data, containing... One sample; Represents density ratio A set of large points; Represent the empty set; select one that comprehensively considers local density. and relative distance parameters : (1) In the DPC algorithm, the cluster centers have a larger local density compared to other points. and relative distance Therefore, by using formula (1) to increase the cluster center Other points The difference in values ​​was calculated. Afterwards, Sort the values ​​in descending order, then calculate. The formula for calculating the downward trend of the value is: (2) In the formula, It is the current value; and Representing the previous moment and the next moment respectively After obtaining the downward trend, the point with the largest downward trend and the preceding few points are taken as the cluster center. Using the entropy method to constrain value; The formula for calculating information entropy is: in, Represents the number of samples in dataset S; for Pointed The proportion of all points The possible values ​​are as follows: When taking the minimum value Value as the final clustering value The value is used to replace empirical methods or experimental verification, thereby achieving adaptive selection of parameters.

9. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 1, characterized in that, A sliding window is constructed based on health indicators, with a window step size of 1 and a window length of 50, to control the data within the sliding window. Principles and setting fluctuation thresholds for polynomial regression Its expression is: Among them, the least squares method is used to determine , and The value; For time series of health indicators, This represents the distance between the data to be tested and the cluster centers. Based on the sliding window fluctuation threshold Determine health indicators after correcting spurious fluctuations Its expression is: in, and Represent The upper and lower limits; In the formula, express The mean; express Standard deviation; Based on the above formula, determine whether there are false fluctuations in the bearing health indicators and make corrections accordingly. hour, Move to the average to get the new forecast value ;when hour, Replace with the value predicted by multinomial regression; when hour, Move to Then, new predicted values ​​are calculated. By building a regression model on a sliding window, spurious fluctuations in health indicators can be effectively corrected, thereby obtaining health indicators with better performance in representing degradation trends.

10. The method for evaluating the performance degradation of rolling bearings based on electrostatic monitoring according to claim 3, characterized in that, The experimental device includes a main body (100), a base (200) is fixedly installed at the bottom of the main body (100), hydraulic cylinders (300) are fixedly installed around the top of the base (200), a support block (400) is fixedly connected to the bottom of the output shaft of the hydraulic cylinder (300), and casters (500) are provided around the bottom of the base (200).

Images