Performance Improvement of Nonlinear Flux Observer for Sensorless Control of PMSM

The effectiveness of nonlinear flux observer has been validated in practical application. However, it is not easy to tune the gradient search gain. The observer is sensitive to the parameter variation and also the stability can be greatly affected by the gradient search gain. This article proposes an improved nonlinear flux observer to enhance the performance. The proposed method utilizes the gradient search for both magnitude and phase of the rotor, which changes the error dynamic equations. Further, it brings the benefits of smaller estimation error and quicker dynamic responses, since the convergence path is more reasonable. In addition, it has the advantages of better steady-state performance, robustness against parameter variation, and stronger stability against larger observer gain value. By comparisons of simulations and experiments, the superiority is demonstrated.