本論文的研究目是研製精密定位控制的內藏式永磁同步馬達驅動系統，並提出智慧型非奇異點終端滑動模態控制，以改變非線性時變系統固有的非線性和時變控制特性。該研究首先設計了一個非奇異點終端滑動模態控制系統，用於追蹤非線性和時變系統的狀態。實際應用中，因為受控系統的參數變化與外力干擾等不確定項的影響，不確定項的上界難以精確取得，因此本研究提出非對稱歸屬函數之遞迴式派翠機率模糊類神經網路創建一個智慧型控制系統，來近似理想的非奇異點終端滑動模態控制律，並加入自適應補償器，以主動調整非對稱歸屬函數之遞迴式派翠機率模糊類神經網路的計算偏差。此外，本文採用每安培最大轉矩控制方法，除了提高馬達效率之外，為了提升控制迴路的性能和頻寬，亦使用非對稱歸屬函數之遞迴式派翠機率模糊類神經網路速度控制器以增加速度迴路頻寬。最後通過實驗結果驗證了所提出的非對稱歸屬函數之遞迴式派翠機率模糊類神經網路智慧型控制器的有效性和強健性。實驗所使用之硬體為應用德州儀器公司生產之浮點數數位訊號處理器TMS320F28075之內藏式永磁同步馬達伺服驅動系統。;This study aims to create an intelligent control system to alter the inherent nonlinear characteristics of a nonlinear time-varying system by using an intelligent nonsingular terminal sliding mode control recurrent Petri probabilistic fuzzy neural network (INTSMCRPPFNN) that features an intelligent nonsingular terminal sliding mode control. This study first designs a nonsingular terminal sliding mode control (NTSMC) system to track the states of a nonlinear time-varying system. Creating a working NTSMC for practical applications is quite complex because the detailed system dynamics, which includes the unpredictability of the controlled plant, is not available beforehand. Thus, this study suggests that a recurrent Petri probabilistic fuzzy neural network with asymmetric membership function (RPPFNN-AMF) can act as a close substitute for the ideal NTSMC to resolve its existing complications. Furthermore, this study modifies an adaptive compensator to proactively adjust for the potential calculated deviance of the RPPFNN-AMF. Asymptotical stability is assured by using the Lyapunov stability method, which generates the RPPFNN-AMF’s online learning algorithms. Finally, in the case study, some experimental results of a maximum torque per ampere (MTPA) operated interior permanent magnet synchronous motor (IPMSM) position servo drive are provided to verify the effective and robust qualities of the suggested INTSMCRPPFNN.
