Enhancing Safety in Locomotive Maintenance through Deep Learning: A CNN-LSTM Approach for Forecasting

This study presents the utilization of a deep learning model, specifically a multi-classification CNN-LSTM approach, to enhance safety in locomotive maintenance. The model achieved an impressive accuracy of 91.97422%, showcasing its potential to predict and prioritize maintenance issues accurately. Notable predictions include a high likelihood of bearing-related complications and battery performance problems. Further insights highlighted moderate likelihoods for issues concerning cooling systems, electrical components, and other engine parts. Additionally, the model provides a percentage distribution of prevalent issues across specific facilities, such as those in Chicago and Indiana. This highlights the immense potential of the CNN-LSTM model in regionalizing preventive measures, emphasizing its transformative capacity in locomotive maintenance. With continued refinement and application, we anticipate a safer and more efficient trajectory for locomotive maintenance endeavors.