Motor Misalignment Detection Based on Hidden Markov Model

This paper purpose a fault detection technique in three-phase induction motors based on hidden Markov model (HMM). This technique detects misalignment of motor's shaft by using HMM recognition of stator's current using cepstral coefficients as feature vectors. We divided the amount misalignment into six group levels. In each level, we trained the HMM with a set of data that represents each level of misalignment, the experiments indicated 83.34% recognition rate