Research on Fault Prognosis Technology for Mechanical Equipment Based on CNN-LSTM Model

Mechanical equipment comprises numerous components, and once a fault occurs, the processes of detection, isolation, and maintenance become highly complex, significantly increasing the difficulty of logistical support. To address the requirements of real-time performance, stability, and accuracy in fault prognosis for mechanical systems, this study proposes a CNN-LSTM deep learning model by integrating the rapid training capability of Convolutional Neural Networks (CNN) and the temporal signal processing strength of Long Short-Term Memory (LSTM) networks. The model is validated using bearing vibration data from Case Western Reserve University (CWRU). Experimental results demonstrate that the CNN-LSTM model achieves an accuracy of 98.24% in time-series fault prediction, outperforming standalone CNN or LSTM models in both accuracy and training efficiency. Furthermore, the introduction of the OCSSA-VMD algorithm enhances fault classification accuracy, with the hybrid OCSSA-VMD-CNN-LSTM model achieving 95% classification accuracy. The findings confirm the feasibility and effectiveness of the proposed approach in improving equipment safety, reducing maintenance cycles, prolonging service life, and minimizing accident risks.