A study of the detection of fault frequencies for condition monitoring of induction machines

Abstract

This paper describes a study aimed to improve the detection of faults in induction motors. Two hidden faults (misalignment and soft footing) and the broken rotor bar fault are investigated by analysing a number of noninvasive sensor quantities: voltage, current, vibration and axial flux. The measurements were performed under different loading conditions to determine the effects of loading on the characteristic fault frequencies. The actual motor speed has also been measured in real time to be able to determine the fault frequencies accurately, since the fault frequencies depend on the slip frequency and hence operating speed. The measured sensor parameters were analysed under steady-state operation using the frequency spectrum technique, and a number of results were provided to investigate the relationships between the healthy motor and the motor with multiple faults considering the characteristic fault frequencies.