本論文提出一只以數位訊號處理器為基礎之非對稱歸屬函數之TSK機率模糊類神經網路控制非接觸式鋰錳電池充電器。此充電器將設計對一鋰錳電池組實現定電流-定電壓混合式充電策略，所提電路架構採用半橋串聯諧振電路。為改進U型鐵芯變壓器有限氣隙距離所造成電磁感應不良及其效率不佳問題，本文改以圓盤型線圈耦合器取代U型鐵芯變壓器。額定功率時，兩圓盤型線圈氣隙間距最大為20公厘，效率可達80%。為了要改善輸出電壓在負載調節及追蹤輸出電流命令變動時的暫態響應，而以非對稱歸屬函數之TSK機率模糊類神經網路控制器取代傳統的比例積分控制器。此外，使用所提出之非對稱歸屬函數之TSK機率模糊類神經網路控制器可改善電池定電流充電轉換為定電壓充電模式後的電流漣波。本文將詳細介紹非對稱歸屬函數之TSK機率模糊類神經網路的架構、線上學習法則以及收斂性分析，而所提之非對稱歸屬函數之TSK機率模糊類神經網路控制器實現對二次電池之定電流-定電壓混合式充電策略的控制性能將由實驗結果驗證。;A digital signal processor (DSP)-based TSK-type probabilistic fuzzy neural network with asymmetric membership function (TSKPFNN-AMF) is proposed in this study to control a contactless battery charger. The half-bridge series resonant converter (SRC) is employed in the power stage while the designed charger adopts constant-current and constant-voltage (CC-CV) charging strategy to charge a Li-Mn battery pack. In order to improve the inferior electromagnetic induction and efficiency of the U-shape ferrite core transformer, the U-shape ferrite core transformer is replaced by the circular pad couplers. As a result, the air gap distance of two circular pads can reach 20mm and the efficiency is 80% at the rated output power. Moreover, to improve the transient of voltage regulation during load variation and the tracking of current command change, a TSKPFNN-AMF controller is proposed to replace the traditional proportional-integral (PI) controller. The proposed TSKPFNN-AMF is incorporated into the CC-CV charging strategy in order to overcome the current ripple that comes after the transition from CC to CV charging. The network structure and the online learning algorithms of the TSKPFNN-AMF controller are introduced in detail. Furthermore, the control performances of the proposed TSKPFNN-AMF control system for CC-CV charging are evaluated by some experimental results.
