Improved Feature-Position-Based Sensorless Control Scheme for SRM Drives Based on Nonlinear State Observer at Medium and High Speeds

The article proposes a nonlinear state observer (NSO) for robust position-sensorless control of switched reluctance motor (SRM) drives over medium- and high-speed range. A classical reference flux-linkage method is adopted to capture a feature position of the SRM, which avoids the use of 3-D magnetic characteristics and has better universality. However, the estimation accuracy of this method would be deteriorated due to flux-linkage errors. To ease the problem, the NSO is developed to enhance the robustness against flux-linkage distortions for more accurate position and speed estimation. This observer can first reconstruct complete position information from a low-resolution feature position. The adverse impact of flux-linkage errors on position estimation is then investigated through a novel small-signal approximation and suppressed by an augmented state estimator. Afterward, a parameter design scheme is given to ensure the observer's stability and improve the capability in distortion suppression. To baseline the performance, comparative experimental validation between the proposed NSO and a widely used linear prediction method is conducted on a 128 SRM setup. The results show that the proposed strategy can improve the overall position-sensorless control performance in both the steady and transient states.