Online parameter estimation and loss calculation using duplex neural — Lumped parameter thermal network for faulty induction motor

A stator winding fault in one phase of induction motor (IM) gives rise to higher harmonics distortion, increased torque ripple, temperature rise in the magnetic material, mechanical vibrations due to varying magnetic forces and magnetic noise. The fault leads to a change in the electromagnetic field generated in the motor as compared to the normal operation of motor. The copper losses generated in stator increases, thus leading to overall increase in the temperature of the motor. Looking from the aspect of electrical equivalent circuit model, the parameters of the motor changes due the occurrence of fault, which makes it difficult for designing a drive for the motor. In this paper a novel computational model has been presented which uses both artificial neural network model (ANN) and lumped parameter thermal network (LPTN) for parameter estimation and calculation of losses which can be used for designing a fault-tolerant, loss minimizing drive. This dual network model has been and experimented on a 7.5 hp aluminum-rotor induction motor.