Induction motor stator faults diagnosis by using parameter estimation algorithms

Parameter estimation is a cost-effective method for fault detection of induction motors. This method is based on detecting change of the characteristic parameters at presence of fault. However, the challenge of parameter estimation is nonlinearity of a machine model which results in multiple local minima involved during the computation process. This paper investigates the suitability of local and global search methods to be used in the estimation of characteristic parameters that are indicating stator short circuit faults. Results of practical case studies are presented where two search methods (local and global) are evaluated and compared. A further study in noisy environment proves the feasibility of diagnosing the fault based on stator currents with low signal to noise ratio.