ROTOR BAR DAMAGE DETECTION IN MULTI LEVEL INVERTER DRIVEN INDUCTION MOTOR USING HILBERT HUANG TRANSFORM AND NEURAL NETWORKS

This paper proposes a Hilbert Huang Transform –Back Propagation Neural Networks method for Rotor damage detection, in a multi-level inverter driven induction motor. For a motor with a damaged or broken rotor bar, the nature of transient starting stator current will be different compared to that of a motor without any rotor damage. In this work, the stator current of the normal and damaged motor is transformed using Hilbert Huang Transform (HHT). The HHT decomposes the signal into intrinsic mode functions(IMF). In this work seven IMFs are used and statistical parameters are estimated for each IMF ‘s, and given to a trained Back Propagation Neural Network (BPNN) The BPNN classifies the extracted transient stator current to either belonging to normal or a motor with a damaged rotor. The effectiveness of Hilbert Huang Transform -Neural Networks method for rotor damage detection is compared with wavelet transform- Back Propagation Neural Networks, using confusion matrix parameters. The results obtained from confusion matrix parameters shows that HHT based method performs better compared to conventional wavelet transform based method.