HEALTH MONITORING AND IDENTIFICATION OF INDUCTION MOTOR FAULTS BASED ON CURRENT AND VIBRATION SIGNALS

Abstract

Induction motors, especially three-phase induction motors, play an important role in various industries owing to their advantages over other electrical machines due to their reliability and safe operation. However, if any faults or failures occur in the motor, it can lead to increased breakdown time and substantially generate huge losses in terms of revenue and maintenance. Therefore, an early fault detection scheme is needed for prior maintenance of these motors. In the current scenario, the demand for health monitoring of induction motors is growing with the objective of reducing their operational costs, enhancing their operational reliability and providing better service to customers. In recent trends, various monitoring methods are being used to investigate fault conditions in induction motors. On the basis of these methods, the health monitoring schemes of induction motors are categorised into two major areas such as feature extraction based analysis and knowledge based analysis. For real time health monitoring, signal processing based feature extraction methods are being used. The incipient faults are diagnosed by extracting signatures from measured signals such as vibration signal, noise signal, current and voltage signal etc. The knowledge based approaches are quite popular among researchers, which are being developed by using artificial intelligence, deep learning and machine learning algorithms to detect the faults most accurately at early stages. The research presented in this thesis is primarily focused on signal processing analysis, using machine learning algorithms to identify stator winding inter-turn faults and ball bearing faults. To identify these faults, experimental setups are developed for each fault separately that are capable of measuring both current and vibration signals by conducting experiments in the laboratory. The subsequent analysis is carried out using MATLAB and Python programming. These early stage incipient faults are analysed utilising through current and vibration signals with various signal processing techniques such as Fast Fourier Transform (FFT), Park’s Vector Magnitude Analysis (PVM), Signal Envelope Identification Analysis (SEI) and Zero Crossing Time Detection (ZCT).