Self-sensing control of PM synchronous machines including online system identification based on a novel MRAS approach

Some of the remaining challenges of self-sensing control for permanent magnet synchronous machines (PMSM) are: combination of EMF-based and magnetic saliency-based methods, influence of the fundamental current control on saliency-based methods, position-transducerless system identification of key parameters or mitigation of distortion effects, respectively, and reduction of computational effort. The authors present a novel Model Reference Adaptive System (MRAS) approach which takes these demands into account. In this paper, the focus is set on identification of saturation-dependent inductances and mitigation of harmonic estimation errors. A linear approximation of saturation effects is achieved within an auto-commissioning process. Furthermore an adaptive filter is presented mitigating harmonic estimation errors which are caused by multiple saliencies e.g. Thus in contrast to passive filter structures a good bandwidth of the self-sensing control is achieved. Experimental results demonstrate a good performance of self-sensing control for the entire speed range including dynamic operation (with maximal loading), although no offline measurements/identification of the drive system was applied.