Application of Neural Networks in Predictive Maintenance of AC and DC Motors

Abstract

This study examines how Artificial Neural Networks (ANNs) can be used in predictive maintenance procedures in AC and DC motors in terms of real-time fault detection rates, diagnosis accuracy and performance accuracy at variable loads. The purpose of conducting the study was to create a stable ANN-based system that would eliminate shortcomings of the traditional approaches. It was revealed on analysis that ANN models could efficiently process multi-sensor data and provide diagnosis with an accuracy of over 95 % in fault detection and provide a 60 % reduction in diagnosis time. ANN was found to be highly reliable in predictive terms whereby the error rates of the predictor were lower than 7%. Comparative findings indicated that ANN is much faster, precise and adaptable than conventional models. All these results prove that ANN systems provide powerful, intelligent tools to maintaining industrial motor with the minimal amount of unexplainable failures and operations stereo. The study is in line with all the goals and authenticates the validity of ANN within an intractable maintenance environment.