A method, system, device and storage medium for predicting the remaining life of a rolling bearing

Abstract

The present invention discloses a method, system, device, and storage medium for predicting the remaining life of a rolling bearing, relating to the technical field of mechanical fault prediction. The method comprises: obtaining a vibration signal of a rolling bearing and the corresponding detection time, and constructing a dynamic state space and a continuous action space parameterized by a Gaussian distribution; designing a dynamic reward function with prediction error as feedback, and constructing a Markov decision process model based on the state space, the action space, and the reward function based on a ResNet network framework; iteratively training the Markov decision process model using a PPO algorithm to obtain a trained Markov decision process model; and predicting the remaining life of a rolling bearing using the trained Markov decision process model. The present invention can adaptively learn the degradation pattern of a rolling bearing, significantly improving the robustness and accuracy of the prediction model.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical failure prediction, and in particular to a method, system, device and storage medium for predicting the remaining life of a rolling bearing. Background Art

[0002] Rolling bearings are core components of rotating machinery, and their health status directly affects the reliability and safety of equipment operation. Remaining life prediction is a key step in predictive maintenance. Traditional methods rely primarily on two technologies: statistical models and traditional deep reinforcement learning algorithms. Statistical models (such as Weibull distribution and hidden Markov models) require manual feature extraction and struggle to capture nonlinear degradation processes under complex operating conditions. Traditional deep reinforcement learning (DRL) algorithms (such as DQN and DDPG) are prone to unstable policy updates, vanishing or exploding gradients in continuous action spaces, resulting in limited prediction accuracy.

[0003] The shortcomings of the existing technology are mainly reflected in:

[0004] Feature extraction relies on manual experience: Traditional methods rely on expert knowledge to extract signal features, resulting in poor generalization. Unstable training: Traditional DRL algorithms are susceptible to noise in complex environments, making policy updates difficult to control. Low gradient propagation efficiency: Deep networks are prone to vanishing gradients during training, impacting model convergence speed and prediction accuracy. Summary of the Invention

[0005] The purpose of the present invention is to provide a method, system, device and storage medium for predicting the remaining life of a rolling bearing, which can adaptively learn the degradation pattern of the rolling bearing and significantly improve the robustness and accuracy of the prediction model.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for predicting the remaining life of a rolling bearing, comprising:

[0008] Obtain the vibration signal of the rolling bearing and the corresponding detection time, and construct the dynamic state space and the continuous action space parameterized by Gaussian distribution;

[0009] Designing a dynamic reward function with prediction error as feedback, and constructing a Markov decision process model based on the state space, the action space, and the reward function based on the ResNet network framework, and iteratively training the Markov decision process model using the PPO algorithm to obtain a trained Markov decision process model;

[0010] The trained Markov decision process model is used to predict the remaining life of rolling bearings.

[0011] Optionally, the state space is represented as In each training cycle (pass) e, m state-time pairs are randomly sampled As the state set for trajectory prediction; where s i is the vibration signal, y i is the detection time, s t is the current state, y t is the actual remaining life, i is the initial monitoring time point, and T is the final monitoring time point.

[0012] Optionally, the action space is expressed as:

[0013] Λ=μ+σ

[0014] Where μ is the mean and σ is the standard deviation.

[0015] Optionally, the reward function is expressed as:

[0016] r t =-1×|a t -y t |

[0017] Among them, a t is the predicted value of the agent, y t The actual remaining life.

[0018] Optionally, the Markov decision process model includes a policy network and a value network; the policy network and the value network are both constructed based on the ResNet network framework, and the mathematical expression of the residual block structure of the ResNet network framework is: H(x)=F(x)+x, where F(x) represents the residual component and x is the input information.

[0019] Optionally, the ResNet network framework specifically includes:

[0020] Input layer: accepts one-dimensional vibration signals, and the output shape is (bs, 1, s_len(4096));

[0021] Wide convolutional layer: used to capture the short-term dependency features of the signal, with an output shape of (bs, 32, 512);

[0022] Batch normalization layer: used to accelerate model convergence, with an output shape of (bs, 32, 512);

[0023] Max pooling layer: used for dimensionality reduction, with an output shape of (bs, 32, 256);

[0024] 8 residual blocks: Each residual block contains a convolutional layer, a batch normalization layer and an activation function, and the output shapes are (bs,32,128), (bs,64,64), (bs,64,64), (bs,128,32), (bs,128,32), (bs,64,16), (bs,64,16), (bs,32,8);

[0025] Global average pooling layer: used to integrate spatial information, with an output shape of (bs, 32);

[0026] Fully connected layer: outputs policy value or value estimate, with output shape of (bs,2) or (bs,1).

[0027] Optionally, the objective function of the PPO algorithm is:

[0028]

[0029] Where ε is the cropping factor, is the strategy ratio, s t is the current state, a t is the predicted value of the agent, θ is the network parameter, π θ is the old strategy, π′ θ For the new strategy, E t is the expectation at time step t, is a multi-step advantage function, and its calculation formula is:

[0030]

[0031] Among them, γ is the discount factor, L is the multi-step learning length, V φ is the output of the value network.

[0032] The present invention also provides a rolling bearing remaining life prediction system, comprising:

[0033] A space construction unit is used to obtain the vibration signal of the rolling bearing and the corresponding detection time, and to construct a dynamic state space and a continuous action space parameterized based on Gaussian distribution;

[0034] a model building and training unit, configured to design a dynamic reward function using prediction error as feedback, and construct a Markov decision process model based on the state space, the action space, and the reward function based on a ResNet network framework, and iteratively train the Markov decision process model using a PPO algorithm to obtain a trained Markov decision process model;

[0035] The model prediction unit is used to predict the remaining life of the rolling bearing using the trained Markov decision process model.

[0036] The present invention also provides an electronic device, comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the above-mentioned rolling bearing remaining life prediction method.

[0037] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the above-mentioned method for predicting the remaining life of a rolling bearing.

[0038] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0039] The present invention discloses a method, system, device, and storage medium for predicting the remaining life of a rolling bearing. The method comprises obtaining a vibration signal of the rolling bearing and the corresponding detection time, constructing a dynamic state space and a continuous action space parameterized by a Gaussian distribution, designing a dynamic reward function with prediction error as feedback, constructing a Markov decision process model based on the state space, the action space, and the reward function based on a ResNet network framework, iteratively training the Markov decision process model using a PPO algorithm to obtain a trained Markov decision process model, and predicting the remaining life of the rolling bearing using the trained Markov decision process model. The present invention can adaptively learn the degradation pattern of the rolling bearing, significantly improving the robustness and accuracy of the prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 This is a basic concept diagram of reinforcement learning in this embodiment;

[0042] Figure 2 This is the architecture diagram of the PPO-ResNet remaining life prediction model in this embodiment;

[0043] Figure 3 is a diagram of the strategic value network structure in this embodiment;

[0044] Figure 4 is a structural diagram of the residual block in this embodiment;

[0045] Figure 5The figure is a line graph showing the changes and comparison between the predicted and actual values ​​of the bearing RUL based on PPO-ResNet in this embodiment; part (a) is a schematic diagram of the outer ring; part (b) is a schematic diagram of the rolling element; and part (c) is a schematic diagram of the inner ring. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0047] The purpose of the present invention is to provide a method, system, device and storage medium for predicting the remaining life of a rolling bearing, which can adaptively learn the degradation pattern of the rolling bearing and significantly improve the robustness and accuracy of the prediction model.

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] like Figure 1 As shown, the present invention provides a method for predicting the remaining life of a rolling bearing, comprising:

[0050] Step 100: Obtain the vibration signal of the rolling bearing and the corresponding detection time, and construct a dynamic state space and a continuous action space parameterized based on Gaussian distribution.

[0051] Step 200: Design a dynamic reward function with prediction error as feedback, and based on the ResNet network framework, construct a Markov decision process model according to the state space, the action space and the reward function, and use the PPO algorithm to iteratively train the Markov decision process model to obtain a trained Markov decision process model.

[0052] Step 300: Use the trained Markov decision process model to predict the remaining life of the rolling bearing.

[0053] The PPO algorithm reduces the impact of the update amplitude on the agent learning process by constraining the ratio between the new and old policies, and introduces an objective function for mini-batch updates and multi-step training. The Markov decision process model includes a policy network and a value network. The output layers of the policy network and the value network are mapped to policy values ​​and value estimates, respectively, through fully connected layers, with output dimensions of 2 (policy) and 1 (value), respectively. The final output of the ResNet module is compressed in spatial dimensions through a global average pooling layer (Global Avgpool) and then connected to a fully connected layer to generate a prediction result.

[0054] As a specific implementation method, each of the above steps is described in detail.

[0055] State space definition: The state space consists of the vibration signal of the rolling bearing and the monitoring time, which is expressed as where s i is the vibration signal, y i is the detection time. Each state s t Represents the vibration signal at time step t and the corresponding detection time.

[0056] Action space definition: The action space is a continuous space that represents the agent's predicted value of the remaining lifespan. t The agent is parameterized by Gaussian distribution and expressed as Λ=μ+σ, where μ is the mean and σ is the standard deviation. t Select an action t , which is the predicted value of remaining life.

[0057] Reward function design: reward function r t Used to evaluate the prediction accuracy of the agent at time step t, the calculation formula is r t =-1×|a t -y t |, where a t is the predicted value of the agent, y t is the actual remaining lifespan. The goal of the reward function is to guide the agent to optimize its prediction strategy so that its predicted value is as close to the actual value as possible.

[0058] Strategy network and value network construction: Strategy network is used to generate action a t The probability distribution of θ (a t |s t), where θ is a network parameter. The policy network is built on the ResNet architecture. Through residual connections and batch normalization (BN) layers, it effectively alleviates the vanishing gradient problem and improves model convergence speed and generalization performance. The ResNet module contains 8 residual blocks. Each residual block has the structure H(x) = F(x) + x, where F(x) consists of a convolutional layer, a batch normalization layer, and a ReLU activation function.

[0059] The value network is used to estimate the state s t The value of V φ (s t ), where φ is a network parameter. The value network is also built on the ResNet architecture and has a similar structure to the policy network. It is used to evaluate the long-term return of the current state.

[0060] The policy network and value network generate trajectory data τ = (s0, a0, r0; s1, a1, r1; ...) by interacting with the environment, and optimize the network parameters based on the PPO algorithm.

[0061] Application of PPO algorithm: PPO algorithm ensures the stability of the training process by constraining the update range of new and old strategies. Its objective function is:

[0062]

[0063] in: is the strategy ratio; is the advantage function, which is calculated as:

[0064]

[0065] Where ε is the clipping factor. The PPO algorithm optimizes the objective function and gradually updates the policy network parameters θ to maximize the cumulative reward.

[0066] Multi-step reinforcement learning: Through multi-step reinforcement learning, advantage function Estimated to be:

[0067]

[0068] Where γ is the discount factor and L is the step size of multi-step learning. Multi-step reinforcement learning can more accurately estimate long-term rewards and improve the effectiveness of policy optimization.

[0069] Model training and prediction: The PPO agent and ResNet network are initialized, trajectory data is generated through interaction with the environment, and the PPO algorithm is used to update the policy network and value network. During training, the agent continuously optimizes its strategy to maximize cumulative reward. During the testing phase, the vibration signal of a rolling bearing is input, and the trained model outputs a predicted remaining life. The model adaptively learns the degradation pattern of the rolling bearing, significantly improving the accuracy and robustness of the prediction.

[0070] The beneficial effects of the present invention are:

[0071] High-precision prediction: The PPO algorithm significantly improves RUL prediction accuracy through stable strategy updates combined with deep feature extraction of ResNet.

[0072] Strong robustness: Residual connections and BN layers effectively alleviate the vanishing gradient problem and enhance the model's ability to resist interference from noisy signals.

[0073] Efficient training: The multi-step reinforcement learning design reduces sample requirements, and the sample efficiency of PPO is significantly improved compared to the traditional DRL algorithm.

[0074] Based on the above technical solution, the following embodiments are provided.

[0075] Dataset Description: The IMS (Intelligent Maintenance System) bearing dataset from the University of Cincinnati covers the entire lifecycle of bearings, from healthy to failure. It includes multiple failure modes (including inner races, outer races, and rolling elements) and complete lifecycle data, making it suitable for validating the model's generalization capabilities for complex degradation patterns. Four ZA-2115 double-row rolling element bearings were mounted on a shaft and driven by an AC motor at a constant speed of 2000 rpm. Vibration signals were collected using an accelerometer with a sampling frequency of 20.48 kHz, capturing one second of data every ten minutes (a total of 20,480 data points). Dataset 1 contains 2,156 files. At the end of the run-to-failure experiment, bearing 3 experienced an inner race failure, and bearing 4 experienced a rolling element failure. Dataset 2 contains 984 files. At the end of the run-to-failure experiment, bearing 1 experienced an outer race failure.

[0076] Data Preprocessing: This implementation uses Datasets 1 and 2. Dataset 1 corresponds to bearing inner race faults (bearing named Bearing_3_3) and rolling element faults (bearing named Bearing_2_3), while Dataset 2 corresponds to bearing outer race faults (bearing named Bearing_1_2). 70% of the total samples in each dataset are divided into a training set and 30% into a test set. A fast Fourier transform (FFT) is performed on the time-domain vibration signals to extract frequency-domain features, highlighting the fault frequency components and making it easier to identify bearing degradation characteristics.

[0077] Model training: The PPO-ResNet framework is used. The input is a one-dimensional vibration signal, and the output is the remaining life prediction value. Both the strategy network and the value network use the ResNet architecture. Specifically, it includes:

[0078] Input layer: accepts one-dimensional vibration signals, and the output shape is (bs, 1, s_len(4096));

[0079] Wide convolutional layer: used to capture the short-term dependency features of the signal, with an output shape of (bs, 32, 512);

[0080] Batch normalization layer: used to accelerate model convergence, with an output shape of (bs, 32, 512);

[0081] Max pooling layer: used for dimensionality reduction, with an output shape of (bs, 32, 256);

[0082] 8 residual blocks: Each residual block contains a convolutional layer, a batch normalization layer and an activation function, and the output shapes are (bs,32,128), (bs,64,64), (bs,64,64), (bs,128,32), (bs,128,32), (bs,64,16), (bs,64,16), (bs,32,8);

[0083] Global average pooling layer: used to integrate spatial information, with an output shape of (bs, 32);

[0084] Fully connected layer: outputs policy value or value estimate, with output shape of (bs,2) or (bs,1).

[0085] The PPO algorithm uses a clipping factor ε of 0.2, a discount factor γ of 0.9, and a multi-step learning step size L of 5. The training cycle (epoch) is 20, with 5000 interactions between the agent and the environment per cycle (num_episodes). The policy network learning rate (actor_lr) is 0.0001, and the value network learning rate (critic_lr) is 0.005. To alleviate the credit allocation problem of delayed rewards in the model, a GAE (Advantage Estimator) parameter λ is introduced in the advantage function with a value of 0.9.

[0086] By interacting with the environment to generate trajectory data, the PPO objective function is used to update the network parameters after each training cycle to optimize the strategy to maximize the cumulative reward.

[0087] Prediction verification: The trained model is used for the test set, and the prediction results are compared with the actual remaining life value. Figure 5The results show that the predicted curve is highly consistent with the actual remaining life value. The evaluation indicators are RMSE (root mean square error), MAE (mean absolute error), and MAPE (mean absolute percentage error). The calculation formula is as follows:

[0088]

[0089] Among them, a i To predict the remaining life value, y i is the actual remaining life value, and m is the total sample size.

[0090] After calculation, the prediction results are as follows:

[0091] Outer circle (Bearing_1_2): RMSE is 0.0391; MAE is 0.0312; MAPE is 13.6290%;

[0092] Rolling element (Bearing_2_3): RMSE is 0.0482; MAE is 0.0341; MAPE is 10.4301%;

[0093] Inner circle (Bearing_3_3): RMSE is 0.0402; MAE is 0.0211; MAPE is 4.3288%;

[0094] Compared with traditional deep neural network (DNN) and support vector machine (SVM) methods, the prediction error of the present invention is significantly reduced. For example, the MAPE of the inner circle prediction is reduced by about 51% and 64% compared with DNN (8.72%) and SVM (12.15%), respectively.

[0095] Among them, the inner ring has the best prediction performance (MAPE = 4.3288%), thanks to its relatively stable degradation mode, ResNet can effectively extract deep frequency domain features; the outer ring and rolling element have higher errors, which may be related to their structural complexity and working noise interference, such as Figure 5 As shown in part (a) of the figure, there are multi-band noise peaks in the outer ring vibration signal spectrum; the multi-step reinforcement learning design enables the model to capture long-term degradation trends, further improving the prediction stability of complex components such as rolling elements.

[0096] This example demonstrates that the PPO-ResNet-based rolling bearing remaining life prediction method exhibits superior performance across different bearing components, maintaining high robustness, particularly under complex operating conditions. The experimental results validate the effectiveness and practicality of this method for remaining life prediction tasks.

[0097] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0098] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for predicting the remaining life of a rolling bearing, characterized in that: include: Obtain the vibration signal of the rolling bearing and the corresponding detection time, and construct the dynamic state space and the continuous action space parameterized by Gaussian distribution; Designing a dynamic reward function with prediction error as feedback, and constructing a Markov decision process model based on the state space, the action space, and the reward function based on the ResNet network framework, and iteratively training the Markov decision process model using the PPO algorithm to obtain a trained Markov decision process model; The remaining life of rolling bearings is predicted using the trained Markov decision process model; The state space is represented as In each training cycle e, m state-time pairs are randomly sampled As the state set for trajectory prediction; where s i is the vibration signal, y i is the detection time, s t is the current state, y t is the actual remaining life, i is the initial monitoring time point, and T is the final monitoring time point; The action space is defined as: The action space is a continuous space, which represents the agent's prediction value of the remaining lifespan. Action a t The parameterization is done by Gaussian distribution, which is expressed as Λ=μ+σ, where μ is the mean and σ is the standard deviation. The agent is assigned a value at each time step t based on the current state s. t Select an action t , which is the predicted value of remaining life; The reward function is expressed as: r t =-1×|a t -y t | Among them, y t is the actual remaining life; The Markov decision process model includes a policy network and a value network; both the policy network and the value network are constructed based on the ResNet network framework, and the residual block structure mathematical expression of the ResNet network framework is: H(x)=F(x)+x, where F(x) represents the residual component and x is the input information; the policy network is used to generate action a t The probability distribution of θ (a t |s t ), where θ is the network parameter; The objective function of the PPO algorithm is: Where ε is the cropping factor, is the strategy ratio, s t is the current state, π θ is the old strategy, π′ θ For the new strategy, E t is the expectation at time step t, is a multi-step advantage function, and its calculation formula is: Among them, γ is the discount factor, L is the multi-step learning length, V φ is the output of the value network, and φ is the network parameter.

2. The method for predicting the remaining life of a rolling bearing according to claim 1, wherein: The ResNet network framework specifically includes: Input layer: accepts one-dimensional vibration signals, and the output shape is (bs, 1, s_len(4096)); Wide convolutional layer: used to capture the short-term dependency features of the signal, with an output shape of (bs, 32, 512); Batch normalization layer: used to accelerate model convergence, with an output shape of (bs, 32, 512); Max pooling layer: used for dimensionality reduction, with an output shape of (bs, 32, 256); 8 residual blocks: Each residual block contains a convolutional layer, a batch normalization layer and an activation function, and the output shapes are (bs,32,128), (bs,64,64), (bs,64,64), (bs,128,32), (bs,128,32), (bs,64,16), (bs,64,16), (bs,32,8); Global average pooling layer: used to integrate spatial information, with an output shape of (bs, 32); Fully connected layer: outputs policy value or value estimate, with output shape of (bs,2) or (bs,1).

3. A rolling bearing remaining life prediction system, applying the method according to any one of claims 1-2, characterized in that: include: A space construction unit is used to obtain the vibration signal of the rolling bearing and the corresponding detection time, and to construct a dynamic state space and a continuous action space parameterized based on Gaussian distribution; a model building and training unit, configured to design a dynamic reward function using prediction error as feedback, and construct a Markov decision process model based on the state space, the action space, and the reward function based on a ResNet network framework, and iteratively train the Markov decision process model using a PPO algorithm to obtain a trained Markov decision process model; The model prediction unit is used to predict the remaining life of the rolling bearing using the trained Markov decision process model.

4. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the method for predicting the remaining service life of a rolling bearing according to any one of claims 1 to 2.

5. A computer-readable storage medium, characterized in that It stores a computer program, which, when executed by a processor, implements the method for predicting the remaining life of a rolling bearing as described in any one of claims 1 to 2.

Images