Bearing residual life prediction method based on BN-RBM and DA-BiLSTM

Abstract

The invention discloses a bearing residual life prediction method based on BN-RBM and DA-BiLSTM, and the method comprises the steps: firstly collecting an original vibration signal in a bearing operation process, carrying out the data preprocessing, and then constructing a bearing residual life prediction model which comprises a BN-RBM feature extraction network, a DA-BiLSTM feature fusion network and a linear regression layer. The BN-RBM feature extraction network dynamically updates the weight prior distribution of a restricted Boltzmann machine RBM by using the posterior probability constructed by Bayesian BN, and performs feature extraction by using the trained RBM, the DA-BiLSTM feature fusion network performs bidirectional feature fusion, the linear regression layer performs linear activation, and the predicted value of the residual life of the bearing is output. The method effectively improves the prediction accuracy of the residual life of the bearing, and reduces the shutdown risk caused by the fault through timely and effective life prediction before the fault.

Description

Technical Field

[0001] This invention relates to the field of mechanical data processing technology, specifically a method for predicting the remaining life of bearings based on BN-RBM and DA-BiLSTM. Background Technology

[0002] Bearings are common mechanical components widely used in various industries. However, with increasing usage time, bearings are often subjected to extremely high loads during operation, and under such harsh conditions, bearing defects become increasingly apparent. Bearings frequently fail due to wear and fatigue, which can lead to machine downtime and production line interruptions. However, when a bearing first fails, the vibration signals are weak and difficult to detect. Therefore, predicting and assessing the remaining service life of bearings is crucial for scheduling planned maintenance before they approach failure, ensuring reliable equipment operation, and protecting personnel and equipment safety. Bearing remaining service life prediction is based on the current state and historical operating data of the bearing to predict its remaining life at a future point in time. Currently, significant progress has been made in the field of bearing life prediction, such as feature extraction methods based on multi-scale morphological decomposition spectral entropy, and pattern recognition methods using support vector machines and artificial neural networks. However, traditional RUL prediction methods have significant limitations in feature extraction capabilities and the ability to model dynamic changes in time-series data. RBMs excel at learning nonlinear features in high-dimensional data, but lack the ability to model time dependencies. BiLSTMs are good at capturing long-term dependency structures in time series, but they are sensitive to the quality of input features and easily affected by redundant information, thus impacting prediction stability and accuracy. Furthermore, traditional model parameter optimization relies on human experience or greedy search strategies, making them prone to getting trapped in local optima and lacking robustness. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a bearing remaining life prediction method based on BN-RBM and DA-BiLSTM, which effectively improves the accuracy of bearing remaining life prediction. By performing timely and effective life prediction before failure, the risk of downtime caused by failure is reduced, and the time, economic and energy losses are reduced.

[0004] The technical solution of this invention is as follows:

[0005] A bearing remaining life prediction method based on BN-RBM and DA-BiLSTM specifically includes the following steps:

[0006] (1) Collect the original vibration signals during the operation of the bearing and perform data preprocessing to construct the bearing dataset;

[0007] (2) Construct a bearing remaining life prediction model. The bearing remaining life prediction model includes a BN-RBM feature extraction network, a DA-BiLSTM feature fusion network, and a linear regression layer. The BN-RBM feature extraction network uses the posterior probability constructed by Bayesian BN to dynamically update the weight prior distribution of the Restricted Boltzmann Machine (RBM), obtaining a dynamic RBM weight prior distribution. Then, the RBM trained by the dynamic RBM weight prior distribution is used to extract the degradation features in the bearing data. The DA-BiLSTM feature fusion network uses the sliding window mechanism of the BiLSTM network to construct time-series samples and performs bidirectional feature fusion on the extracted features. The sliding window length is determined by the Dragonfly Optimization (DA) algorithm. The linear regression layer linearly activates the fused features of the bearing and outputs the predicted value of the bearing remaining life.

[0008] (3) Construct a loss function to train the bearing remaining life prediction model, and use the trained bearing remaining life prediction model to predict the bearing remaining life and obtain the bearing remaining life.

[0009] The data preprocessing involves performing a fast Fourier transform on the original vibration signal to convert the time-domain signal into a frequency-domain signal, obtaining the frequency-domain amplitude spectrum information, and constructing a bearing remaining life label, thereby constructing a bearing dataset.

[0010] The BN-RBM feature extraction network performs feature extraction in the following steps:

[0011] S21. Construct a three-layer Bayesian BN network B = (F, D, R), F = {f1, f2, ..., f...} D} represents the set of nodes representing the original degradation characteristics of the bearing. These original degradation characteristics are selected from historical degradation data and reflect the degradation pattern of the bearing; D = {d1, d2, ..., d K} represents the set of nodes in the degradation stage; R represents the set of nodes with remaining bearing life;

[0012] The conditional dependencies between nodes in a BN network are described using a conditional probability table, as shown in equation (1) below:

[0013] P(d k |f i ) = Pr(degeneration stage d) k Appearance | Feature f i Observed values) (1);

[0014] In equation (1), P(d) k |f i The role of the bearing is to encode prior knowledge of degradation feature changes, degradation stages and remaining life, which is obtained by training based on the historical degradation data of the bearing.

[0015] S22. Posterior probability of the degenerate stage of inference: given the degenerate feature vector at the current time t. The posterior probability of the degradation stage is inferred using a Bayesian BN network, as shown in equation (2) below:

[0016]

[0017] In equation (2), F t d represents the degenerate feature vector extracted at the current time t; k P(d) represents the k-th degradation stage; k ) represents the degradation stage d k The prior probability is obtained from the statistical analysis of historical degradation data of the bearing; P(F t ∣d k ) represents the degradation stage d k Below, the degradation feature F was observed. t The likelihood probability is obtained by looking up the conditional probability table; P(d k |F t ) represents the given current degradation feature F t Afterwards, it enters the degenerative stage d k The posterior probability;

[0018] The output of the Bayesian BN network is the current degradation feature F. t Below, the posterior probability distribution {P(d1|F}, composed of the posterior probabilities of each degradation stage, is given. t ), P(d2∣F t ),…,P(d k |F t )};

[0019] S23. Construct the weight prior distribution of the Restricted Boltzmann Machine (RBM), as shown in the following equation (3):

[0020]

[0021] In equation (3), P(W∣d) k ) represents the degradation stage d k The prior distribution of the RBM weights W; Represents the degradation stage d k The corresponding weighted mean; Represents the degradation stage d k The corresponding weighted covariance;

[0022] S24. Using Bayesian BN to construct the posterior probability, dynamically update the weight prior distribution of the Restricted Boltzmann Machine (RBM), and obtain the dynamic RBM weight prior distribution P(W|F). t ), see the following formula (4) for details:

[0023]

[0024] In equation (4), P(d) k |F t ) represents the given current degradation feature F t Afterwards, it enters the degenerative stage d k The posterior probability; μ t Represents the updated dynamic weight mean at time t; ∑ t This represents the dynamic weighted covariance at the current time t after the update;

[0025] S25. A Restricted Boltzmann Machine (RBM) trained using a dynamic RBM weight prior distribution is used to extract degenerate features from bearing data. The extracted feature vector is denoted as... D ′ This represents the number of nodes in the hidden layer of the RBM;

[0026] S26. A Dropout layer is introduced to randomly zero out the features extracted by the Restricted Boltzmann Machine (RBM), as shown in the following equation (5):

[0027]

[0028] In equation (5), x dropout1,i,j x represents the value of the j-th degenerate feature of the i-th sample after processing by the Dropout layer. RBM,i,j p represents the j-th degenerate feature vector of the i-th sample input to the Dropout layer. dropout1 This represents the set retention threshold;

[0029] The features extracted by the Restricted Boltzmann Machine (RBM) are randomly zeroed out to output a higher-order feature representation. x RBM F represents a degenerate feature vector after being randomly zeroed out. RBM Represents all degenerate feature vectors x RBM A set of.

[0030] The specific steps for bidirectional feature fusion using the DA-BiLSTM feature fusion network are as follows:

[0031] The degenerate feature set output by the S31 and BN-RBM feature extraction networks is Organized into time series samples according to time series, the generated time series samples are shown in the following formula (6):

[0032]

[0033] In equation (6), X (m)represents the m-th time series sample; w represents the sliding window length, which is determined by the Dragonfly Optimization (DA) algorithm. This represents the set of degradation features at time t in the m-th time series sample;

[0034] S32. The BiLSTM network consists of multiple layers of BiLSTM structure. Each layer of BiLSTM structure includes a forward LSTM and a backward LSTM. The processing procedure of each layer of BiLSTM structure is shown in the following equation (7):

[0035]

[0036] In equation (7), This represents the hidden state of the feedforward LSTM at time t. This represents the hidden state of the feedforward LSTM at time t-1. This represents the hidden state of the backward LSTM at time t. This represents the hidden state of the backward LSTM at time t+1. f Represents forward LSTM, LSTM b Represents a backward LSTM, h t This represents the bidirectional fusion characteristic at time t;

[0037] S33. A Dropout layer is introduced to randomly zero out the time-series samples processed by the BiLSTM network, and global average pooling is used to obtain a fixed vector representation, as shown in the following equation (8):

[0038]

[0039] In equation (8), Dropout represents the Dropout layer, and the processing procedure is the same as in equation (5); This represents the fused features output by the DA-BiLSTM feature fusion network.

[0040] The sliding window length w is determined by the Dragonfly Optimization Algorithm (DA), and the specific steps are as follows:

[0041] S311. Construct the configuration parameters of the BiLSTM network and combine them to form vector X. i =[l,H,η,p,w],X i In the diagram, l represents the number of layers in the BiLSTM network, H represents the size of the hidden layers in the BiLSTM network, η represents the learning rate, p represents the Dropout ratio, and w represents the sliding window length.

[0042] Each X i For an individual, defined within a preset range, n random X values ​​are generated. i At the same time, initialize its velocity vector ΔX i ;

[0043] S312. In each iteration, X i The parameters in the formula are applied to the BiLSTM network, trained and evaluated on the validation set, and the prediction mean square error (MSE) is calculated. The fitness value is set to the validation MSE. The group with the smallest and largest MSE in the current group is labeled, corresponding to food source and enemy, respectively.

[0044] S313, the Dragonfly Optimization Algorithm (DA) includes the following five types of behavior vectors:

[0045] Separate S i Position avoidance between individuals;

[0046] Align A i Speed ​​is averaged compared to neighbors;

[0047] Cohesion C i : Move closer to the center of the group;

[0048] Feeding F i : To move closer to the optimal individual;

[0049] Avoid the enemy E i Stay away from the worst individuals;

[0050] In conjunction with the inertia weight ω, the velocity and parameter configuration are dynamically updated, as shown in equations (9) and (10) below:

[0051]

[0052] In equations (9) and (10), The velocity vector at time t+1 The velocity vector at time t. This represents the configuration parameters of the BiLSTM network at time t. The parameters represent the configuration parameters of the BiLSTM network at time t+1; the weights s, a, c, f, e, and the inertial weight ω gradually transition from exploration to utilization during iteration, stopping when the maximum number of iterations is reached or the fitness improvement is lower than the threshold; the output result is the optimal individual, which is the optimal parameter configuration of the BiLSTM network.

[0053] The linear regression layer linearly activates the fusion characteristics of the bearing and outputs a predicted value of the remaining bearing life, as shown in the following formula (11):

[0054]

[0055] In equation (11), W represents the weight parameters of the fully connected layer; b represents the bias of the fully connected layer; and f represents the activation function. This represents the predicted value of the remaining life of the bearing.

[0056] The loss function is shown in equation (12):

[0057]

[0058] In equation (12), RUL i This represents the actual remaining bearing life label for the i-th sample. Let represent the predicted remaining life of the i-th sample bearing, λ represent the regularization coefficient, θ represent the parameter set of the bearing remaining life prediction model, and N represent the total number of samples. 2 This represents the square of the L1 norm.

[0059] Advantages of this invention:

[0060] (1) The BN-RBM feature extraction network of this invention uses the posterior probability constructed by Bayesian BN to dynamically update the weight prior distribution of the Restricted Boltzmann Machine (RBM). Based on historical fault data, it calculates the posterior probability of feature parameters, constructs a probabilistic graphical model to quantify the uncertainty of the feature extraction process, and outputs confidence intervals instead of single predicted values. This enables the RBM to automatically focus on more sensitive features in the current stage during feature extraction (such as focusing on lubrication features in the mild degradation stage and focusing on metal wear features in the severe degradation stage), thereby enhancing the feature expression ability and the pertinence and accuracy of remaining life prediction. It can more comprehensively reflect the uncertainty of bearing remaining life prediction and provide richer information for decision-making in industrial applications.

[0061] (2) The present invention inputs the high-order feature vector output by the BN-RBM feature extraction network into the BiLSTM network, and uses its bidirectional memory mechanism to capture the long-term dependency relationship between the bearing's previous and subsequent states during operation, thereby accurately depicting the fault development trend and outputting the life prediction value at the corresponding time step.

[0062] (3) The present invention can effectively avoid getting stuck in local optima by using the global search mechanism of the Dragonfly Optimization Algorithm (DA), which effectively improves the prediction accuracy, accelerates the convergence process of the model, and significantly reduces the training time. By simulating the five behaviors of dragonflies in nature: separation, alignment, alliance, prey hunting and evasion of predators, it can effectively search for the optimal feature combination in the high-dimensional feature space and optimize the parameter configuration of BiLSTM to adapt to different data features, so that the BiLSTM network can more flexibly cope with data changes when processing complex time series data. Attached Figure Description

[0063] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0064] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0065] See Figure 1 , a bearing remaining life prediction method based on BN-RBM and DA-BiLSTM, specifically including the following steps:

[0066] (1). Collect the original vibration signals during the operation of the bearing and perform data preprocessing. The data preprocessing is to perform fast Fourier transform processing on the original vibration signals, convert the time-domain signals into frequency-domain signals, obtain the frequency-domain amplitude spectrum information, and construct the bearing remaining life labels, thereby constructing a bearing data set;

[0067] (2). Construct a bearing remaining life prediction model. The bearing remaining life prediction model includes a BN-RBM feature extraction network, a DA-BiLSTM feature fusion network, and a linear regression layer;

[0068] S2. The BN-RBM feature extraction network performs feature extraction, specifically including the following steps:

[0069] S21. Construct a three-layer Bayesian BN network B=(F, D, R), where F={f1, f2,..., f D} represents the set of nodes of the original bearing degradation features (including but not limited to the 1x frequency amplitude in the frequency domain, kurtosis, energy entropy, impact factor, maximum value of the envelope spectrum, wavelet packet energy entropy, time-domain statistics (mean, standard deviation, skewness, kurtosis)). These degradation features are selected from historical degradation data and have high sensitivity related to the bearing remaining life, and can effectively reflect bearing degradation modes such as lubrication failure and metal wear; D={d1, d2,..., d K} represents the set of nodes of the degradation stage. K = 3, that is, the degradation stage nodes include the healthy stage d1, the mild degradation stage d2, and the severe degradation stage d3; R represents the set of nodes of the bearing remaining life (RUL) (continuous value or discrete interval, such as "RUL>100h", "50h<RUL≤100h", etc.);

[0070] The conditional dependencies between the nodes in the BN network are described by a conditional probability table, specifically as shown in the following formula (1):

[0071] P(d k |f i ) = Pr(The degradation stage d k appears | feature f iObserved values) (1);

[0072] In equation (1), P(d) k |f i The role of the bearing is to encode prior knowledge of degradation feature changes, degradation stages and remaining life, which is obtained by training based on the historical degradation data of the bearing.

[0073] S22. Posterior probability of the degenerate stage of inference: given the degenerate feature vector at the current time t. The posterior probability of the degradation stage is inferred using a Bayesian BN network, as shown in equation (2) below:

[0074]

[0075] In equation (2), F t d represents the degenerate feature vector extracted at the current time t; k P(d) represents the k-th degradation stage; k ) represents the degradation stage d k The prior probability is obtained from the statistical analysis of historical degradation data of the bearing; P(F t ∣d j ) represents the degradation stage d k Below, the degradation feature F was observed. t The likelihood probability is obtained by looking up the conditional probability table; P(d k |F t ) represents the given current degradation feature F t Afterwards, it enters the degenerative stage d k The posterior probability;

[0076] The output of the Bayesian BN network is the current degradation feature F. t Below, the posterior probability distribution {P(d1|F}, composed of the posterior probabilities of each degradation stage, is given. t ), P(d2∣F t ),P(d3∣F t )};For example:

[0077]

[0078] S23. Construct the weight prior distribution of the Restricted Boltzmann Machine (RBM), as shown in the following equation (3):

[0079]

[0080] In equation (3), P(W∣d) k ) represents the degradation stage d k The prior distribution of the weights W of the RBM is a Gaussian distribution. Represents the degradation stage dk The corresponding weighted mean; Represents the degradation stage d k The corresponding weighted covariance;

[0081] S24. Using Bayesian BN to construct the posterior probability, dynamically update the weight prior distribution of the Restricted Boltzmann Machine (RBM), and obtain the dynamic RBM weight prior distribution P(W|F). t ), see the following formula (4) for details:

[0082]

[0083] In equation (4), P(d) k |F t ) represents the given current degradation feature F t Afterwards, it enters the degenerative stage d k The posterior probability; μ t Represents the updated dynamic weight mean at time t; ∑ t This represents the dynamic weighted covariance at the current time t after the update;

[0084] S25. A Restricted Boltzmann Machine (RBM) trained using a dynamic RBM weight prior distribution is used to extract degenerate features from bearing data. The extracted feature vector is denoted as... D ′ The number of nodes in the hidden layer of the RBM; the feature vector x RBM It is a high-order feature representation learned by an RBM network with the node set F of the original degradation features of the bearing as input. It reflects the deep potential feature structure of the current state of the bearing. Compared with the original degradation feature set F of the bearing, it has stronger discriminativeness and robustness, and is suitable for subsequent time series modeling and remaining life prediction tasks.

[0085] S26. A Dropout layer is introduced to randomly zero out the features extracted by the Restricted Boltzmann Machine (RBM), as shown in the following equation (5):

[0086]

[0087] In equation (5), x dropout1,i,j x represents the value of the j-th degenerate feature of the i-th sample after processing by the Dropout layer. RBM,i,j p represents the j-th degenerate feature vector of the i-th sample input to the Dropout layer. dropout1 This represents the set retention threshold, with a value between 0.5 and 0.8; scaling operation (divided by p) dropout This ensures consistency between the expected features during the training and inference phases, preventing the overall output from shrinking due to random neuron shutdowns.

[0088] The features extracted by the Restricted Boltzmann Machine (RBM) are randomly zeroed out to output a higher-order feature representation. x RBM F represents a degenerate feature vector after being randomly zeroed out. RBM Represents all degenerate feature vectors x RBM A set;

[0089] F RBM Before inputting the data into the BiLSTM network, a sliding window mechanism is used to construct time-series samples: w is the length of the sliding window, X (m) This represents the m-th time series sample; this process ensures a natural connection between feature extraction and temporal modeling, providing structured input for subsequent lifetime prediction;

[0090] The specific steps for bidirectional feature fusion using the S3 and DA-BiLSTM feature fusion networks are as follows:

[0091] The degenerate feature set output by the S31 and BN-RBM feature extraction networks is Organized into time series samples according to time series, the generated time series samples are shown in the following formula (6):

[0092]

[0093] In equation (6), X (m) Represents the m-th time series sample; w represents the sliding window length; This represents the set of degradation features at time t in the m-th time series sample;

[0094] The sliding window length w in equation (6) is determined by the Dragonfly Optimization Algorithm (DA), and the specific steps are as follows:

[0095] S311. Construct the configuration parameters of the BiLSTM network and combine them to form vector X. i =[l,H,η,p,w],X i In the diagram, l represents the number of layers in the BiLSTM network, H represents the size of the hidden layers in the BiLSTM network, η represents the learning rate, p represents the Dropout ratio, and w represents the sliding window length.

[0096] Each X i For an individual, defined within a preset range, n random X values ​​are generated. i At the same time, initialize its velocity vector ΔX i ;

[0097] S312. In each iteration, X iThe parameters in the formula are applied to the BiLSTM network, trained and evaluated on the validation set, and the prediction mean square error (MSE) is calculated. The fitness value is set to the validation MSE. The group with the smallest and largest MSE in the current group is labeled, corresponding to food source and enemy, respectively.

[0098] S313, the Dragonfly Optimization Algorithm (DA) includes the following five types of behavior vectors:

[0099] Separate S i Position avoidance between individuals;

[0100] Align A i Speed ​​is averaged compared to neighbors;

[0101] Cohesion C i : Move closer to the center of the group;

[0102] Feeding F i : To move closer to the optimal individual;

[0103] Avoid the enemy E i Stay away from the worst individuals;

[0104] In conjunction with the inertia weight ω, the velocity and parameter configuration are dynamically updated, as shown in equations (9) and (10) below:

[0105]

[0106] In equations (9) and (10), The velocity vector at time t+1 The velocity vector at time t. This represents the configuration parameters of the BiLSTM network at time t. The parameters represent the configuration parameters of the BiLSTM network at time t+1; the weights s, a, c, f, e, and the inertial weight ω gradually transition from exploration (diversity) to utilization (convergence) during iteration, stopping when the maximum number of iterations is reached or the fitness improvement is lower than the threshold; the output result is the optimal individual, which is the optimal parameter configuration of the BiLSTM network.

[0107] S32. The BiLSTM network consists of multiple layers of BiLSTM structure. Each layer of BiLSTM structure includes a forward LSTM and a backward LSTM. The processing procedure of each layer of BiLSTM structure is shown in the following equation (7):

[0108]

[0109] In equation (7), This represents the hidden state of the feedforward LSTM at time t. This represents the hidden state of the feedforward LSTM at time t-1. This represents the hidden state of the backward LSTM at time t. This represents the hidden state of the backward LSTM at time t+1. f Represents forward LSTM, LSTM b Represents a backward LSTM, h t This represents the bidirectional fusion characteristic at time t;

[0110] S33. A Dropout layer is introduced to randomly zero out the time-series samples processed by the BiLSTM network, randomly discarding some neuron activation values ​​to prevent overfitting, and a fixed vector representation is obtained using global average pooling, as shown in the following equation (8):

[0111]

[0112] In equation (8), Dropout represents the Dropout layer, and the processing procedure is the same as in equation (5); The fused features represent the output of the DA-BiLSTM feature fusion network;

[0113] The linear regression layer performs linear activation on the fusion characteristics of the bearing and outputs the predicted value of the remaining bearing life, as shown in the following equation (11):

[0114]

[0115] In equation (11), W represents the weight parameters of the fully connected layer; b represents the bias of the fully connected layer; and f represents the ReLU activation function. The predicted value representing the remaining life of the bearing;

[0116] (3) Construct a loss function to train the bearing remaining life prediction model. The loss function is shown in the following formula (12):

[0117]

[0118] In equation (12), RUL i This represents the actual remaining bearing life label for the i-th sample. Let represent the predicted remaining life of the i-th sample bearing, λ represent the regularization coefficient, θ represent the parameter set of the bearing remaining life prediction model, and N represent the total number of samples. 2 Represents the square of the L1 norm;

[0119] Then, the trained bearing remaining life prediction model is used to predict the bearing remaining life and obtain the bearing remaining life.

[0120] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting the remaining life of bearings based on BN-RBM and DA-BiLSTM, characterized in that: Specifically, it includes the following steps: (1) Collect the original vibration signals during the operation of the bearing and perform data preprocessing to construct the bearing dataset; (2) Construct a bearing remaining life prediction model. The bearing remaining life prediction model includes a BN-RBM feature extraction network, a DA-BiLSTM feature fusion network, and a linear regression layer. The BN-RBM feature extraction network uses the posterior probability constructed by Bayesian BN to dynamically update the weight prior distribution of the Restricted Boltzmann Machine (RBM), obtaining a dynamic RBM weight prior distribution. Then, the RBM trained by the dynamic RBM weight prior distribution is used to extract the degradation features in the bearing data. The DA-BiLSTM feature fusion network uses the sliding window mechanism of the BiLSTM network to construct time-series samples and performs bidirectional feature fusion on the extracted features. The sliding window length is determined by the Dragonfly Optimization (DA) algorithm. The linear regression layer linearly activates the fused features of the bearing and outputs the predicted value of the bearing remaining life. (3) Construct a loss function to train the bearing remaining life prediction model, and use the trained bearing remaining life prediction model to predict the bearing remaining life and obtain the bearing remaining life.

2. The bearing remaining life prediction method based on BN-RBM and DA-BiLSTM according to claim 1, characterized in that: The data preprocessing involves performing a fast Fourier transform on the original vibration signal to convert the time-domain signal into a frequency-domain signal, obtaining the frequency-domain amplitude spectrum information, and constructing a bearing remaining life label, thereby constructing a bearing dataset.

3. The bearing remaining life prediction method based on BN-RBM and DA-BiLSTM according to claim 2, characterized in that: The BN-RBM feature extraction network performs feature extraction in the following steps: S21. Construct a three-layer Bayesian BN network B = (F, D, R), F = {f1, f2, ..., f...} D } represents the set of nodes representing the original degradation characteristics of the bearing. These original degradation characteristics are selected from historical degradation data and reflect the degradation pattern of the bearing; D = {d1, d2, ..., d K } represents the set of nodes in the degradation stage; R represents the set of nodes with remaining bearing life; The conditional dependencies between nodes in a BN network are described using a conditional probability table, as shown in equation (1) below: P(d k |f i ) = Pr(degeneration stage d) k Appearance | Feature f i Observed values) (1); In equation (1), P(d) k |f i The role of the bearing is to encode prior knowledge of degradation feature changes, degradation stages and remaining life, which is obtained by training based on the historical degradation data of the bearing. S22. Posterior probability of the degenerate stage of inference: given the degenerate feature vector at the current time t. The posterior probability of the degradation stage is inferred using a Bayesian BN network, as shown in equation (2) below: In equation (2), F t This represents the degenerate feature vector extracted at the current time t; d k P(d) represents the k-th degradation stage; k ) represents the degradation stage d k The prior probability is obtained from the statistical analysis of historical degradation data of the bearing; P(F t ∣d k ) represents the degradation stage d k Below, the degradation feature F was observed. t The likelihood probability is obtained by looking up the conditional probability table; P(d k |F t ) represents the given current degradation feature F t Afterwards, it enters the degenerative stage d k The posterior probability; The output of the Bayesian BN network is the current degradation feature F. t Below, the posterior probability distribution {P(d1|F}, composed of the posterior probabilities of each degradation stage, is given. t ), P(d2∣F t ),…,P(d k |F t )}; S23. Construct the weight prior distribution of the Restricted Boltzmann Machine (RBM), as shown in the following equation (3): In equation (3), P(W∣d) k ) represents the degradation stage d k The prior distribution of the RBM weights W; Represents the degradation stage d k The corresponding weighted average; Represents the degradation stage d k The corresponding weighted covariance; S24. Using Bayesian BN to construct the posterior probability, dynamically update the weight prior distribution of the Restricted Boltzmann Machine (RBM), and obtain the dynamic RBM weight prior distribution P(W|F). t ), see the following formula (4) for details: In equation (4), P(d) k |F t ) represents the given current degradation feature F t Afterwards, it enters the degenerative stage d k The posterior probability; μ t Represents the updated dynamic weight mean at time t; ∑ t This represents the dynamic weighted covariance at the current time t after the update; S25. A Restricted Boltzmann Machine (RBM) trained using a dynamic RBM weight prior distribution is used to extract degenerate features from bearing data. The extracted feature vector is denoted as... D′ represents the number of nodes in the hidden layer of the RBM; S26. A Dropout layer is introduced to randomly zero out the features extracted by the Restricted Boltzmann Machine (RBM), as shown in equation (5) below: In equation (5), x dropout1,i,j x represents the value of the j-th degenerate feature of the i-th sample after processing by the Dropout layer. RBM,i,j p represents the j-th degenerate feature vector of the i-th sample input to the Dropout layer. dropout1 This represents the set retention threshold; The features extracted by the Restricted Boltzmann Machine (RBM) are randomly zeroed out to output a higher-order feature representation. x RBM F represents a degenerate feature vector after being randomly zeroed out. RBM Represents all degenerate feature vectors x RBM A set of.

4. The bearing remaining life prediction method based on BN-RBM and DA-BiLSTM according to claim 3, characterized in that: The specific steps for bidirectional feature fusion using the DA-BiLSTM feature fusion network are as follows: The degenerate feature set output by the S31 and BN-RBM feature extraction networks is Organized into time series samples according to time series, the generated time series samples are shown in the following formula (6): In equation (6), X (m) represents the m-th time series sample; w represents the sliding window length, which is determined by the Dragonfly Optimization (DA) algorithm. This represents the set of degradation features at time t in the m-th time series sample; S32. The BiLSTM network consists of multiple layers of BiLSTM structure. Each layer of BiLSTM structure includes a forward LSTM and a backward LSTM. The processing procedure of each layer of BiLSTM structure is shown in the following equation (7): In equation (7), This represents the hidden state of the feedforward LSTM at time t. This represents the hidden state of the feedforward LSTM at time t-1. This represents the hidden state of the backward LSTM at time t. This represents the hidden state of the backward LSTM at time t+1. f Represents forward LSTM, LSTM b Represents a backward LSTM, h t This represents the bidirectional fusion characteristic at time t; S33. A Dropout layer is introduced to randomly zero out the time-series samples processed by the BiLSTM network, and global average pooling is used to obtain a fixed vector representation, as shown in the following equation (8): In equation (8), Dropout represents the Dropout layer, and the processing procedure is the same as in equation (5); This represents the fused features output by the DA-BiLSTM feature fusion network.

5. The bearing remaining life prediction method based on BN-RBM and DA-BiLSTM according to claim 4, characterized in that: The sliding window length w is determined by the Dragonfly Optimization Algorithm (DA), and the specific steps are as follows: S311. Construct the configuration parameters of the BiLSTM network and combine them to form vector X. i =[l,H,η,p,w],X i In the diagram, l represents the number of layers in the BiLSTM network, H represents the size of the hidden layers in the BiLSTM network, η represents the learning rate, p represents the Dropout ratio, and w represents the sliding window length. Each X i For an individual, defined within a preset range, n random X values ​​are generated. i At the same time, initialize its velocity vector ΔX i ; S312. In each iteration, X i The parameters in the formula are applied to the BiLSTM network, trained and evaluated on the validation set, and the prediction mean square error (MSE) is calculated. The fitness value is set to the validation MSE. The group with the smallest and largest MSE in the current group is labeled, corresponding to food source and enemy, respectively. S313, the Dragonfly Optimization Algorithm (DA) includes the following five types of behavior vectors: Separate S i Position avoidance between individuals; Align A i Speed ​​is averaged compared to neighbors; Cohesion C i : Move closer to the center of the group; Feeding F i : To move closer to the optimal individual; Avoid the enemy E i Stay away from the worst individuals; In conjunction with the inertia weight ω, the velocity and parameter configuration are dynamically updated, as shown in equations (9) and (10) below: In equations (9) and (10), The velocity vector at time t+1 The velocity vector at time t. This represents the configuration parameters of the BiLSTM network at time t. The parameters represent the configuration parameters of the BiLSTM network at time t+1; the weights s, a, c, f, e, and the inertial weight ω gradually transition from exploration to utilization during iteration, stopping when the maximum number of iterations is reached or the fitness improvement is lower than the threshold; the output result is the optimal individual, which is the optimal parameter configuration of the BiLSTM network.

6. The bearing remaining life prediction method based on BN-RBM and DA-BiLSTM according to claim 4, characterized in that: The linear regression layer linearly activates the fusion characteristics of the bearing and outputs a predicted value of the remaining bearing life, as shown in the following formula (11): In equation (11), W represents the weight parameters of the fully connected layer; b represents the bias of the fully connected layer; and f represents the activation function. This represents the predicted value of the remaining life of the bearing.

7. The bearing remaining life prediction method based on BN-RBM and DA-BiLSTM according to claim 6, characterized in that: The loss function is shown in equation (12): In equation (12), RUL i This represents the actual remaining bearing life label for the i-th sample. Let represent the predicted remaining life of the i-th sample bearing, λ represent the regularization coefficient, θ represent the parameter set of the bearing remaining life prediction model, and N represent the total number of samples. 2 This represents the square of the L1 norm.

Images