Fault Detection of Motor Using Power Spectrum and Machine Learning

Low, Jun Kiat (2018) Fault Detection of Motor Using Power Spectrum and Machine Learning. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Low Jun Kiat.pdf Restricted to Registered users only Download (4MB)

Abstract

In this project, the three-phase induction motors are the “workhorses” of industry and are the most commonly used electrical machines. In a manufacturing line, they can normally consume between 40 to 50% of all the generated capacity of that country. Motor current signature analysis (MCSA) is detecting an electrical signal containing current components that are direct by-product of unique rotating flux components. Irregularities in operation of the motor modify harmonic content of motor supply current. Therefore the main focus on this project is to implement the Motor Current Signal Analysis to design system to monitor the current spectrum and determine the condition of the induction motor by machine learning. The system uses MATLAB to design, current signal will process through MATLAB for spectral analysis and the machine learning by using Fuzzy Logic. The harmonics of the signal from the power spectral density are extracted as discriminative input features to detect the condition of motor with fuzzy logic system. A comprehensive list of induction motor conditions such as healthy, broken bar, eccentricity, stator winding and unbalance voltage has been successfully classified using the fuzzy logic system.