Abstract:To address the challenges faced in the prediction of rolling bearing life, where temporal signals are affected by noise, making fault feature extraction difficult and resulting in low prediction accuracy, a method based on optimal time–frequency spectra and the DenseNet-ALSTM network is proposed. Firstly, a signal reconstruction method is introduced to enhance vibration signals. This involves using the CEEMDAN deconvolution method combined with the Teager energy operator for signal reconstruction, aiming to denoise the signals and highlight fault impacts. Subsequently, a method based on the snake optimizer (SO) is proposed to optimize the generalized S-transform (GST) time–frequency spectra of the enhanced signals, obtaining the optimal time–frequency spectra. Finally, all sample data are transformed into the optimal time–frequency spectrum set and input into the DenseNet-ALSTM network for life prediction. The comparison experiment and ablation experiment show that the proposed method has high prediction accuracy and ideal prediction performance. The optimization terms used in different contexts in this paper are due to different optimization methods, specifically the CEEMDAN method.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the challenges faced in the life prediction of rolling bearings, especially the difficulty in extracting fault features from time signals under the influence of noise, which leads to low prediction accuracy. Specifically, the paper conducts research in the following aspects: 1. **Signal Processing under Noise Influence**: A large amount of background noise is generated during the operation of rolling bearings. This noise can mask the real fault features, making it very difficult to accurately extract fault features from vibration signals. Therefore, effective signal enhancement and denoising methods are required to improve the visibility of fault features. 2. **Time - Frequency Spectrum Optimization**: Traditional time - domain or frequency - domain analysis methods have limitations when dealing with non - stationary signals and cannot effectively extract the local time - frequency characteristics of the signal. For this reason, the paper proposes a method to optimize the generalized S - transform (GST) time - frequency spectrum based on the snake - shaped optimization algorithm (SO) to obtain the optimal time - frequency spectrum. 3. **Application of Deep Learning Model**: In order to improve the prediction accuracy, the paper introduces the DenseNet - ALSTM network, which is a hybrid model combining the densely connected network (DenseNet), the attention mechanism, and the long - short - term memory network (LSTM). Through this network structure, the fault features in the time - frequency spectrum can be effectively extracted and utilized, thereby improving the prediction accuracy. ### Specific Methods 1. **Signal Enhancement and Reconstruction**: - Use the CEEMDAN deconvolution method combined with the Teager energy operator for signal reconstruction, aiming to denoise and highlight the fault impact features. - The signal reconstruction method can effectively enhance the impact components in the signal while retaining the time - frequency characteristics of the signal. 2. **Time - Frequency Spectrum Optimization**: - Use the snake - shaped optimization algorithm (SO) to optimize the generalized S - transform (GST) time - frequency spectrum to obtain the optimal time - frequency spectrum. - The optimized generalized S - transform time - frequency spectrum can more clearly show the time - frequency characteristics of the fault signal and improve the anti - interference ability. 3. **DenseNet - ALSTM Network**: - Convert all sample data into the optimal time - frequency spectrum set and input it into the DenseNet - ALSTM network for life prediction. - The DenseNet - ALSTM network can effectively extract and utilize the fault features in the time - frequency spectrum through the densely connected layers and the self - attention module, avoid the influence of redundant features, and improve the prediction accuracy. ### Experimental Results The paper verifies the high prediction accuracy and ideal prediction performance of the proposed method through comparative experiments and ablation experiments. The experimental results show that this method can still maintain high prediction accuracy in a noisy environment and has good practical application prospects. ### Conclusion The rolling bearing life prediction method based on the optimal time - frequency spectrum and the DenseNet - ALSTM network proposed in this paper effectively solves the problem of fault feature extraction under the influence of noise, improves the prediction accuracy, and provides a new solution for the health monitoring and life prediction of rolling bearings.

Life Prediction of Rolling Bearing Based on Optimal Time–Frequency Spectrum and DenseNet-ALSTM

A life prediction method of rolling bearing based on signal reconstruction and fusion dual channel network

Remaining Useful Life Prediction and Fault Diagnosis of Rolling Bearings Based on Short-Time Fourier Transform and Convolutional Neural Network

A dual-stream temporal convolutional network for remaining useful life prediction of rolling bearings

Long-term Temporal Attention Neural Network with Adaptive Stage Division for Remaining Useful Life Prediction of Rolling Bearings

Direct Remaining Useful Life Prediction for Rolling Bearing Using Temporal Convolutional Networks.

Predicting remaining useful life of rolling bearings based on deep feature representation and long short-term memory neural network

Combined Remaining Life Prediction of Multiple Bearings Based on EEMD-BILSTM

RUL prediction method for rolling bearing using Convolutional Denoising Autoencoder and Bidirectional LSTM

A predictive analytics framework for rolling bearing vibration signal using deep learning and time series techniques

Rolling bearing remaining useful life prediction based on dilated causal convolutional DenseNet and an exponential model

A novel method for online prediction of the remaining useful life of rolling bearings based on wavelet power spectrogram and Transformer structure

Residual Life Prediction of Bearings Based on SENet-TCN and Transfer Learning.

A Method for Predicting the Remaining Life of Rolling Bearings Based on Multi-Scale Feature Extraction and Attention Mechanism

A Hybrid Model-based Prognostics Approach for Estimating Remaining Useful Life of Rolling Bearings

A Novel Continuous Delay Hidden Layer Deep Belief Network and Its Application in Life Prediction of Rolling Bearings

Multi-bearing remaining useful life collaborative prediction: A deep learning approach

Residual Life Prediction of Rolling Bearings Based on a CEEMDAN Algorithm Fused with CNN–Attention-Based Bidirectional LSTM Modeling

Dual-Attention-Based Multiscale Convolutional Neural Network With Stage Division for Remaining Useful Life Prediction of Rolling Bearings

Remaining Useful Life Prediction of Rolling Bearings Based on Multi-scale Permutation Entropy and ISSA-LSTM

Remaining Life Prediction Method for Rolling Bearing Based on the Long Short-Term Memory Network