Performance Evaluation of Encoderless Control of Permanent Magnet Synchronous Machines using Predictive Current Observer Method

Starting-up a position or speed sensorless permanent magnet synchronous machine (PMSM) under full load can be challenging. Because of the non-saliency characteristics in most surface mounted PMSMs, the high frequency injection (HFI) method which is effective in staring-up interior mounted PMSMs is invalid, unless significant saliency exists. Other starting methods include Volts/Hz control which is sensitive to load and machine parameters, and I-F control method. Like HFI, I-F method needs to be transitioned to other modes such as back- electromotive force (EMF) method when the motor speed reaches a certain threshold. Even though the predictive current method has been used in steady state operation, its PMSM starting characteristics have not been quantified in published literatures. This paper describes an encoderless control method based on an enhanced predictive current observer model for not only being able to startup a PMSM under full load smoothly, but also provide a stable operation in the wide speed range. Theoretical analysis including the predictive current observer structure, and simulation studies are presented to demonstrate the effectiveness from standstill to full speed under full load. Experimental results on a 15kW, 380V, 50Hz, 100Nm PMSM verify the feasibility of the analysis and design. The PMSM can accelerate from zero to rated speed of 1500r/min (RPM) smoothly at full load. The performance is repeatable at very low speed of 2Hz as well. In addition, the position tracking error is quantified when the PMSM parameters vary ±20% with stable operation, demonstrating the robustness of the solution in practical industry applications.