Study Of High Performance Control For PMSM Drive System

Permanent-magnet synchronous motors (PMSMs) drive system has been widely used in modern industrial applications due to its easy control realization, high power density and high control precision. In practice, PID based cascade control is an classical control structure in PMSM drive systems. Since the bandwith of the inter control loop is higher than outer control loop in the cascade control, it can provide a robust control property and a high steady precision. It is also convenient for engineers to tune controller gains in the field. However, the traditional PID controller may not achieve a satisfactory performance, when the imposed load is unknown and variable. Therefore, it has become a significant subject to improve speed control performance of PMSM drive system with unknown load.In this dissertation, based on discrete model reference adaptive control, feed-forward stategy, error back propagation (EBP) and optimal control theorem, the contorl of PMSM with rigid couple and flexible couple mechanical link has been studied. The following topics are discussed such as adaptive observer, adaptive feed-forward control design and Linear Quadratic Regulator (LQR) design. The main contents of this dissertation are organized as follows:1. In rigid couple mechanical link applications, a discrete adaptive observer is introduced to estimate load inertia, friction cofficient and load torque. Based on these estimated results, a feed-forward control branch has been designed using the EBP algorithm to improve the dynamic performance of PMSM drive system. To slove the problems of mechnical noise introduced by differentiating and the delay caused be low-pass filter in calculation of acceleration, phase locking loop method is introduced. Experimental studies demonstrate that the proposed method is easy to be implemented and effective to improve system dynamic performance.2. In flexible couple link applications, an adaptive observer is designed to estimate load inertia, tortional torque and load speed. Using the estimated parameters and states, an LQR based optimal state feed back control for speed is designed and implemented to suppress the mechanical oscillation. To calculate the feed back gains on-line, Numerical solution for Riccati equation is implemented in DSP-based drives. Simulation and Experimental results demonstrate that the observer can estimate parameters and states effectively, and the LQR based optimal feed back control can improve speed performance significantly. 3. The features of mechanical models in both load conditions has been studied. Then, a criterion to differentiate the both load conditions has been proposed and verified via studies. After that, the proposed constrol strategies are analized and synthesized on the base of state feedback control theory. Finally, the structures of the hardware and the software are introduced briefly.At the end of this dissertation, a conclusion has been made, and the prospect of the further research cosidering the application has been given.