Machine Learning Driven Predic-tive Maintenance for Industrial IoT of Electrica Systems

Abstract

This study introduces a predictive maintenance framework that combines Internet of Things (IoT) sensor data with advanced machine learning (ML) techniques to improve the reliability and efficiency of industrial electrical systems. The proposed system analyzes real-time measurements such as temperature, vi bration, current, and voltage to detect anomalies and predict equipment failures before they occur. It integrates essential processes including data cleaning, nor-malization, feature extraction, and model optimization to ensure accurate and timely predictions. Several ML models, including Random Forest, Support Vector Machines (SVM), and Artificial Neural Networks (ANNs), were evaluated for their performance. A real-world case study involving three industrial motors was con ducted using over 10,000 time-series sensor records, 500 maintenance logs, and 100 documented failure cases. The results demonstrate a 35% reduction in unplanned downtime, a 25% increase in resource efficiency, and a 20% extension of equipment lifespan. These findings validate the effectiveness of the proposed approach in sup porting proactive maintenance decisions in industrial environments.