Machine Learning-Based Approach to Predict Remaining Useful Life of Rolling Bearings: OCP as case study

Predictive maintenance is essential for minimizing downtime and optimizing the performance of critical industrial systems. In this study, we propose a machine learning-based approach to predict the Remaining Useful Life (RUL) of rolling bearings, a critical component in phosphate grinding mills. Unlike traditional methods that rely on vibration data, our approach leverages operational data, including shift of work, phosphate layer, phosphate quality, working time, tonnage of rock and cumulative metrics. Two predictive models were developed: Multiple Linear Regression (MLR) and Artificial Neural Networks (ANN). Experimental results demonstrate the potential of ANN-based model to enable precise RUL predictions, facilitating proactive maintenance scheduling and reducing unplanned downtime. The proposed approach is cost-effective, utilizes existing historical data, and can be extended to other industrial equipment and defect types.