A comparison of phase space reconstruction and spectral coherence approaches for diagnostics of bar and end-ring connector breakage faults in polyphase induction motors using current waveforms

Two signal (waveform) analysis approaches are investigated in this paper for motor drive fault identification-one linear and the other nonlinear. Twenty-one different motor-drive operating conditions including healthy, 1 through 10 broken bars, and 1 through 10 broken end-ring connectors are investigated. Highly accurate numerical simulations of current waveforms for the various operating conditions are generated using the time stepping coupled finite element-state space method for a 208-V, 60-Hz, 2-pole, 1.2-hp, squirrel cage 3-phase induction motor. The linear signal analysis method is based on spectral coherence, whereas the nonlinear signal analysis method is based on stochastic models of reconstructed phase spaces. Conclusions resulting from the comparisons of these two methods are drawn.