Developing robust algorithms for speed and parameter estimation in induction machines

Presents the conditioning analysis of the speed and parameter estimation problem for an induction motor and techniques to mitigate the ill conditioning caused by low richness of input signals. We use relative sensitivity functions to quantify the conditioning of individual machine parameters and rotor speed to input data. The sensitivity functions allow us to identify ill conditioned parameters in the estimation problem. To deal will ill conditioning, we use a parameter subset selection method where only a subset of parameters is estimated from the available data while the other parameters are fixed to a priori values. The resulting speed and parameter estimation algorithms are more robust because we are including in the estimation approach information to stabilize the estimates and make them less sensitive to perturbations in the measured data. Experimental results are presented which illustrate the potential of the proposed methodology to deal with ill conditioning in parameter estimation problems in energy processing systems