本論文提出使用適應性模糊控制器改善PDFF(Pseudo Derivative Feedback with Feedforward gain)控制器用於永磁同步馬達速度命令追蹤控制時的追蹤誤差及暫態響應，另一方面，本論文也分別比較使用補償式模糊控制器及參數調整式模糊控制器這兩種改善PDFF控制器的方法之速度追蹤性能。 傳統磁場導向控制法一般是藉由調節比例積分(Proportional Integral, PI)控制器來達到馬達速度及位置控制，PI控制器具有簡單的演算法及傑出的暫態響應的優點，但容易造成過調量和低直流剛性，因此偽微分反饋控制(Pseudo Derivative Feedback, PDF)控制器被提出來改善PI控制器的缺點，它無過調量且高直流剛性，但由於暫態響應慢，所以在很多應用場合被大幅的限制，而PDFF控制器的特色為可以透過調整前饋增益，等同於PI控制器或PDF控制器，所以可獲得彈性的暫態響應。 為了增強PDFF控制器的速度命令追蹤性能，本論文利用適應性模糊控制器控制輸入至PDFF控制器的速度誤差，並且藉由梯度下降法推導適應控制律，讓模糊控制器的規則可以線上即時更新，改善模糊控制器不具學習能力的缺點，最後藉由TMS320F28335 數位訊號處理器(DSP)實作結果及PSIM 模擬結果來實現及驗證所提出的方法。 ;A pseudo derivative feedback with feed forward gain (PDFF) controller combined with an adaptive fuzzy controller is proposed in this study in order to improve the speed command tracking error and transient response of a permanent magnet synchronous motor (PMSM). The speed tracking performance attained using two improved versions of the PDFF controller, which incorporate a compensation fuzzy controller and a parameter adjusting fuzzy controller respectively, are also compared in this study. Traditional field oriented control methods used to control the motor speed or position is generally achieved by adjusting a proportional integral (PI) controller. The PI controller has the advantages of a simple algorithm and excellent transient response, but causes overshoot and low DC stiffness. Therefore, the pseudo derivative feedback (PDF) controller is proposed to improve the aforementioned drawbacks of the PI controller. It has high DC stiffness and no overshoot, but is seriously constrained in many applications due to its slow transient response. The unique feature of the PDFF controller is that it can become a PI or PDF controller by adjusting the feed forward gain. Thus, a flexible transient response can be obtained. In order to increase the speed command tracking performance of the PDFF controller, an adaptive fuzzy controller is used to control the speed error, which is the input of the PDFF controller. In addition, the adaptive control law is derived by the gradient descent method, thereby enabling the rules of the fuzzy controller to update online in real time; the learning ability drawback of a fuzzy controller is thereby improved in this study. Finally, the proposed methods were implemented in the PSIM simulation and also validated in the digital signal processor (DSP) TMS320F28335 experimental results.
