Sensorless Synchronous Reluctance Motor Drives: A General Adaptive Projection Vector Approach for Position Estimation

The paper presents a mathematical framework for the design and analysis of position observers for sensorless control of synchronous reluctance machines. The approach expresses the observer error via a generalized projection vector. Moreover, the proposed framework includes an improved inductance model to account for the position error induced inductance variations. Firstly, the instability regions of active flux based position observer are analytically identified and validated. Then, a novel technique, Adaptive Projection vector for Position error estimation (APP), is introduced, immune to instability of active flux scheme. Furthermore, the proposed technique can be augmented with a second projection vector for independent estimation of speed error, referred to as Adaptive Projection vector matrix for Position and Speed error estimation (APPS). Sensitivity analysis to parameter errors is developed. Stability and performance of proposed technique is validated on a 1 kW synchronous reluctance motor test bench.