Generation of Artificial Datasets for Fault Detection in Induction Motors Using Extreme Learning Machine

Abstract

Induction motors represent a category of electric motors widely adopted in various industrial applications. This paper addresses the challenge of effective fault detection in induction motors through the development of an artificial dataset using simulation techniques. Control strategies, failure modes, and complete dataset generation in Simulink/Matlab are discussed. Simulated data include normal operation and fault conditions. The study implements vector control for motor control strategies and explores fault categories such as electrical faults. The simulation generates data in the form of time series for phase currents, speed, and electromagnetic torque under normal and faulty conditions. The generated dataset is then validated using a fault detection method based on the Extreme Learning Machine. Results of this work for both artificial dataset generation and fault detection offer an economical and promising method to analyse faults without resorting to costly real-world data.