Publication

Abstract : Industrial actuators are the core components in automation systems. So, any actuator fault can lead to complete machine failure. Bearing faults are the primary causes of other unexpected failures due to it’s continuous running under varying loads and different environmental conditions. The traditional monitoring systems primarily depend on centralized processing or periodic inspections, making delayed fault detection and increased latency. This work presents an edge-AI-based real-time bearing health monitoring system for industrial actuators. The proposed system carries out a hybrid two-stage architecture on a low-cost ESP32 microcontroller. The system integrates multiple sensors for continuous actuator health monitoring. Pertinent statistical parameters are extracted and processed using lightweight machine learning models such as decision tree and CatBoost, deployed on an edge computing platform, that precisely identify the bearing’s health under four categories (Normal, Ball, Inner, and Outer Race faults). The system regularly checks the health of bearings and will display the health status on an LCD screen at the output side. The experimental findings showed enhanced response time, hardly dependent on cloud computing and safe detection of bearing health under varying conditions. The proposed system provides a scalable and efficient way for intelligent industrial monitoring and for preventative care.