CNN-BASED MACHINE LEARNING FOR REMAINING USEFUL LIFE PREDICTION OF BEARING MOTORS USING MULTI-SENSOR MONITORING

Abstract

This paper presents a study on predicting the Remaining Useful Life (RUL) of bearing motors using Convolutional Neural Networks (CNN) with multi-sensor condition monitoring. The monitored parameters include vibration (RMS), acoustic emission, temperature, stator current, and RPM. A simulation dataset spanning 1000 days, with daily sampling, was created for three bearing motors under varying conditions. Thresholds for failure events were predefined for each parameter. The CNN model is designed to learn temporal and spatial patterns in the multivariate sensor signals to predict RUL accurately. The results demonstrate that CNN can effectively capture degradation features and provide actionable predictions for preventive maintenance. This work serves as a foundation for integrating real-time IoT-enabled predictive maintenance systems in industrial applications.