Use of fuzzy logic for modeling, estimation, and prediction in switched reluctance motor drives

Abstract

Switched reluctance motor drives may be used in many commercial applications due to their simplicity and low cost. These drives require rotor position feedback to operate. However, in many systems, rotor position sensors have disadvantages. In this paper, a position sensorless scheme is described which uses fuzzy modeling, estimation, and prediction. An important feature is that saturation and real-time nonideal effects are not ignored, but that no mathematical model is required. Instead, a fuzzy-logic-based model is constructed from both static and real-time motor data, and from this model the rotor position is estimated. The system also incorporates fuzzy-logic-based methods to provide a high robustness against noise. This includes a fuzzy predictive filter which combines both fuzzy-logic-based time-series prediction, as well as a heuristic-knowledge-based algorithm to detect and discard feedback signal error. In addition, the method uses heuristic knowledge to choose the most desirable phase for angle estimation in order to minimize the effect of feedback error. It is also shown that, by using fuzzy logic, the estimation scheme offers a high robustness and reliability and is, thus, well suited to a wide range of systems.