Enhanced Fault Diagnosis in Rotating Machinery Using a Hybrid CWT-LeNet-5-LSTM Model: Performance Across Various Load Conditions

Abstract

The presented research paper proposes a novel integrated technique combining LeNet-5 with Continuous Wavelet Transform (CWT) along with Long Short-Term Memory (LSTM). The purpose of this integration is to improve the performance of mechanisms used for the detection of defects in rotatory machines across various operating conditions. The Convolutional Neural Networks (CNN) assists the presented CWT-LeNet-5-LSTM technique in finding the complex characteristics in the data, while LSTM learns the trends in the dataset and performs the necessary analysis of vibrations occurring in faulty machines. The developed model was examined for various loads and faults to extract results having accuracies of 99.6%, 96.9%, 92.5% and 96.6% for load conditions 3, 2, 1, and 0, respectively. These results demonstrate the ability of the proposed model to adapt according to varying load conditions while having the necessary levels of accuracy. This validates the model to perform precise fault detection and diagnosis, offering capabilities of predictive maintenance in industrial settings.