Enhanced Predictive Maintenance Strategy for Industrial Robotics using Hybrid Deep Learning and Sensor Fusion

Abstract

This paper presents an enhanced predictive maintenance (PdM) strategy for industrial robots utilizing a hybrid deep learning approach integrated with sensor fusion. The proposed methodology combines data from multiple sensors (vibration, current, temperature, and position encoders) to provide a comprehensive understanding of robot health. A novel deep learning architecture, consisting of a Convolutional Neural Network (CNN) for feature extraction and a Long Short-Term Memory (LSTM) network for temporal dependency modeling, is employed to predict Remaining Useful Life (RUL) and detect anomalies. The performance of the hybrid model is compared against traditional machine learning algorithms and single deep learning models, demonstrating superior accuracy and robustness. The findings suggest that this integrated approach can significantly reduce downtime, optimize maintenance schedules, and improve the overall efficiency of industrial robotic systems. Furthermore, the practical implications of this research for smart manufacturing environments are discussed.