Gain-Scheduling PI Controller using JITL and BPNN for PMSM Speed Drive System

摘要

For the sake of realizing the high-peformance permanent magnet synchronous motor (PMSM) speed drive system, an improved gain-scheduling proportional-integral (PI) control scheme is proposed in this paper. First, to obtain the accute Jacobian information, a low-cost just-in-time learning (JITL) method is used to estimate the controlled model parameters of PMSM speed drive system, considering the nonlinearity and parameter uncertainties. Then, based on the on-line identification result for Jacobian information, the corresponding back propagetion neural network (BPNN) method is employed to supply the PI controller with the suitable parameters. The whole system stability including the above JITL and BPNN methods is analyzed based on the Lyapunov method. Finally, experimental results demonstrate the effectiveness of the proposed control scheme.