本論文目的為研製以數位訊號處理器為基礎之智慧型控制串聯諧振轉換器，其中串聯諧振電路採用LLC架構。在LLC架構中，是以固定50%責任週期之互補訊號驅動功率晶體，並藉由調變頻率使變壓器一次側之漏電感、激磁電感、諧振電容以及開關上的寄生元件產生諧振，達到零電壓切換來降低元件的切換損失，且在變壓器二次側整流電路使用同步整流功能，以功率開關取代二極體來減少導通損失進而提升系統效率。為了改善輸出電壓在負載調節時的暫態響應，本文提出兩種智慧型控制器取代比例積分控制器，一個是機率型模糊類神經網路，另一個是非對稱歸屬函數之TSK機率模糊類神經網路。本文將詳細介紹機率型模糊類神經網路以及非對稱歸屬函數之TSK機率模糊類神經網路控制器之網路架構、線上學習法則以及收斂性分析。最後透過實驗結果來驗證其可行性。 The purpose of this thesis is to develop a digital signal processor (DSP) based intelligent control of series resonant converter (SRC). The SRC is constructed using LLC resonant tank which is driven by a half-bridge circuit with fixed 50%-duty carrier frequency. The resonant phenomenon can be obtained by using of the leakage inductance of primary coil, the magnetizing inductance of the transformer, the resonant capacitor and parasitic component of power MOSFETs. Moreover, the zero-voltage-switching (ZVS) can be achieved using resonant phenomenon to reduce the switching loss. Furthermore, the synchronous rectifier is added in the secondary side of transformer by using the power MOSFETs in place of the diode to reduce the conduction loss. In addition, to improve the transient response of the voltage regulation during load variation, two intelligent controls are proposed. One is the probabilistic fuzzy neural network (PFNN), and the other is the TSK-type probabilistic fuzzy neural network with asymmetric membership function (TSKPFNN-AMF). The network structures, online learning algorithms and convergence analyses of the PFNN and the TSKPFNN-AMF controls are introduced in detail. Finally, the feasibility of the proposed control schemes are verified through experimentation.
