High robustness of an SR motor angle estimation algorithm using fuzzy predictive filters and heuristic knowledge-based rules

In this paper, the operation of a fuzzy predictive filter used to provide high robustness against feedback signal noise in a fuzzy logic (FL)-based angle estimation algorithm for the switched reluctance motor is described. The fuzzy predictive filtering method combines both FL-based time-series prediction, as well as a heuristic knowledge-based algorithm to detect and discard feedback signal error. As it is predictive in nature, the scheme does not introduce any delay or phase shift in the feedback signals. In addition, the fuzzy predictive filter does not require any mathematical modeling of the noise and, therefore, can be used effectively to control non-Gaussian impulsive-type noise. An analysis of the noise and error commonly found in practical motor drives is given, and how this can effect position estimation. It is shown using experimental results that the FLbased scheme can cope well with erroneous and noisy feedback signals.