本論文研究目的為研製與發展高性能永磁輔助同步磁阻馬達驅動系統，提出利用遞迴式勒壤得模糊類神經網路之計算轉矩控制法，以應對其非線性和時變特性。本論文首先介紹了使用有限元素分析法分析每安培最大轉矩控制，以獲得最佳的電流角命令，並將結果藉由查表法做應用。接著介紹計算轉矩控制法來追隨速度命令，但因系統存在總集不確定項很難事先得知，實際應用中難以實現。有鑒於此，提出結合了遞迴式勒壤得模糊類神經網路來近似計算轉矩控制。此外，為了補償遞迴式勒壤得模糊類神經網路可能的近似誤差，增加了一個自適應補償器，並利用李亞普諾夫穩定性理論推導，以保證遞迴式勒壤得模糊類神經網路線上學習法為漸進穩定。最後通過實驗結果驗證了所提出的遞迴式勒壤得模糊類神經網路之智慧型計算轉矩控制的有效性和強健性。 最後，本研究以32位元浮點運算數位訊號處理器TMS320F28075將所提出的智慧型控制實現於永磁輔助同步磁阻馬達驅動系統。 ;An intelligent computed torque control using recurrent Legendre fuzzy neural network (ICTCRLFNN) is proposed in this study to construct a high-performance PMASynRM drive system to confront its nonlinear and time-varying control characteristics. First, the dynamic model of a maximum torque per ampere (MTPA) controlled PMASynRM drive using ANSYS Maxwell-2D is introduced. The results of the finite element analysis (FEA) are made into a lookup table (LUT) to generate the current angle command of the MTPA. Then, a computed torque control (CTC) system is designed for the tracking of the speed reference. Since the detailed system dynamics including the uncertainty of PMASynRM drive system is unavailable in advance, it is very difficult to design an effective CTC in practical applications. Therefore, to alleviate the existed difficulties of the CTC, a recurrent Legendre fuzzy neural network (RLFNN) is proposed in this study to approximate the CTC. In addition, to compensate the possible approximated error of the RLFNN, an adaptive compensator is augmented. The online learning algorithms of the RLFNN are derived by using the Lyapunov stability method to assure asymptotical stability. Finally, the effectiveness and robustness of the proposed ICTCRLFNN controlled PMASynRM drive are verified by some experimental results. Finally, the proposed intelligent control system and the vector mechanism for the PMASynRM drive are implemented using a 32-bit floating point digital signal processor (DSP) TMS320F28075.
