Parameter identification for vector controlled induction machines

Three algorithms have been presented for parameter identification of the rotor resistance of vector controlled induction machines. The improved Westphal method has the least demanding computational requirements and converges fastest, but does not track parameter changes and is too sensitive to noisy measurements for practical purposes. The extended Kalman filter has been shown to reliably converge and track parameter changes, even in the presence of sensor noise. Using covariance management allows faster convergence. The penalty is the computational requirements which are very high, but realistically attainable with modern digital signal processors. The algorithms presented will be tested on a vector controlled induction machine. A Motorola DSP96002 digital signal processor will be used to process the identification algorithms as well as the vector control.