Zhanshan Wang , Ryongho Jon, Chaomin Luo, and Myongguk Jong
Permanent magnet synchronous motor (PMSM), speed estimation, neural networks, adaptive observer, reconstruction error
In this paper, to eliminate the influence of system uncertainty to the speed estimation in a sensorless permanent magnet synchronous motor (PMSM) control system, a robust speed estimation strategy is proposed based on neural networks (NNs) adaptive observer. The proposed entire estimation strategy is composed of a NN adaptive observer and a NN speed estimator. First, a NN adaptive observer is designed to estimate the complete dynamics of PMSM control system based on an adaptive learning law. Next, as NN used in the adaptive observer has a NN reconstruction error to a certain degree and the vibration and non-convergence may happen in the control result, a robust compensation scheme is proposed to compensate the NN reconstruction error and to guarantee the asymptotic convergence of the speed estimation. Then, a NN speed estimator is implemented to estimate accurately the rotor speed by using the observed system state variables. The simulation results demonstrate the validity, feasibility, and high dynamic response of the proposed speed estimation strategy.
Important Links:
Go Back
