Implementing Machine Learn-ing for Industrial Electrical Systems Maintenance and Re-mote Monitoring

Abstract

Predictive maintenance and remote monitoring have become essential components in modern industrial environments, particularly for electrical systems, which are critical to ensuring operational continuity, safety, and productivity. Tra-ditional maintenance strategies, based on reactive or time-based interventions, of ten lead to unexpected failures, increased costs, and inefficient use of resources. This research addresses these limitations by proposing an intelligent maintenance system that integrates Machine Learning (ML) and Internet of Things (IoT) tech- nologies to predict failures and enable real-time monitoring of industrial electrical equipment. The methodology involves collecting real-time data from electrical systems using smart sensors, transmitting it through an loT infrastructure, and applying ML algorithms to analyze equipment behavior, detect anomalies, and forecast potential failures. The system is tested under simulated industrial con-ditions to assess its accuracy, responsiveness, and usability. A case study on a DC motor equipped with sensors (temperature, humidity, current, and vibration) and connected to an ESP8266 microcontroller demonstrates the system's effective ness, with data transmitted via MQTT to a cloud platform and processed using the Random Forest algorithm, achieving 90% accuracy in fault classification. The results show significant improvements in fault detection, maintenance scheduling. and system reliability, contributing to the development of intelligent maintenance frameworks and supporting the digital transformation of industrial practices in alignment with Industry 4.0 objectives.