Observer-Based Predictive Vector-Resonant Current Control of Permanent Magnet Synchronous Machines

In this paper, an observer-based predictive vector-resonant current control strategy for permanent magnet synchronous machines is proposed, which can effectively improve system performance in case of disturbances caused by converter dead-time effect and model mismatch. Taking the external periodic, nonperiodic disturbances, and system delay into consideration, the system model in discrete-time domain is built. This model is then embedded with vector-resonant scheme with the aim of suppressing periodic disturbances resulting from converter dead-time effect and rotor flux harmonics. An extended-state observer is constructed in order to predict system behavior and to estimate the nonperiodic disturbances caused by parameter variations. The overall control law is then derived on basis of cost function minimization. With the proposed method adopted, the undesired anomalous peak around the resonant frequency in the frequency responses can be avoided. Moreover, enhanced transient performances and wider stability margin can also be obtained. The effectiveness and advantages of the proposed strategy are verified by frequency-domain analysis and comparative experimental studies.