Enhanced Predictive Maintenance for Industrial Machinery using Hybrid Machine Learning and IoT Sensor Fusion

Abstract

This paper proposes a upgraded predictive maintenance (PdM) framework for industrial equipment based on a hybrid machine learning strategy combined with IoT sensor fusion. The proposed framework combines information from diverse sensor modalities (acoustic emissions, temperature, pressure, vibration) to yield a holistic evaluation of equipment condition. A new hybrid approach, where a deep learning-based autoencoder is employed for feature learning and a Random Forest classifier is used to detect anomalies and predict Remaining Useful Life (RUL), is suggested. The proposed framework is tested using a real-world industrial dataset and found to achieve considerable improvements in prediction accuracy and lower false alarm rates than conventional methods. The findings emphasize the promise of this methodology to maximize maintenance schedules, reduce downtime, and enhance overall industrial operation efficiency.