Periodical speed ripple minimization in the PMSM -DTC system using Iterative Learning Control

The permanent magnet synchronous motor with direct torque control (PMSM-DTC) is finding expanded in high performance servo applications where the speed smoothness is essential. To minimize the speed ripple of PMSM-DTC system, the traditional speed ripple suppression methods were discussed, the main components of speed ripple in PMSM-DTC system were analyzed and its periodic characteristic was described. Based on these, a new algorithm using iterative learning control (ILC) was proposed to minimize the periodic speed ripple. The algorithm minimizes the speed ripple by on-line learning and compensating the torque signal. The ILC+PI controller was designed and its stability and convergence condition were derived. The simulation results of the proposed controller show that the speed ripple can be effectively minimized from 2.2% to 0.2% as compared to the traditional PI controller. This algorithm needs a small amount of changes to the original system, the structure and algorithm are both simple and do not depend on the prior system knowledge and parameters, thus can obtain strong robustness.