標題: 切換式磁阻電動機驅動系統之效率提昇策略An Efficiency Improvement Strategy for Switched Reluctance Motor Drives 作者: 黃文楠Wen-Nan Huang鄧清政Ching-Cheng Teng電控工程研究所 關鍵字: 切換式磁阻電動機驅動系統;模糊類神經網路;類神經網路;效率;Switched Reluctance Motor Drives;Fuzzy neural network;Artificial neural network;Efficiency 公開日期: 2005 摘要: 論文中提出一新式之基於分析模式推導結論的切換式磁阻電動機驅動系統之效率提昇策略，而此效率考量和控制方法乃由該電動機之模式而產生應用想法，其乃於獲知等效磁化電感對時間之變化之下，藉著調置相電流命令與電壓間之比例而實現其作動之功能；此效率提昇策略在本論文中經由探討後，構成實用上可行之效率提升方法，其是運用此電動機參數間之連結關係而能於驅動之操作時，由設定之調動規則，在不超過輸出性能需求所設定之範圍下執行命令，而經步階數值之更動，調置電流命令值來搜尋驅動系統運作下可能存在的效率提昇數值。 在完整之效率提昇之控制架構中，考量等效磁化電感為電流與該馬達轉子轉動位置之函數下，先採用模糊類神經網路來求算出電流、位置和該電感之映射關係，再運用模糊類神經網路，近似地計算出與輸出轉矩相關之電感對位置偏微分量；此外，數種需求之參數的新式估測和測量之架構與方法，包含有電阻值、電感值、互感值、轉矩值和轉速值，亦於本論文中提出，而提供其與效率提昇架構之整合和合併之可能考量；本論文也提出一類神經網路，在驅動系統未飽和運作下，乃為輸出性能判斷之核心單元，而於高比例飽和運作之系統中，則與前述模糊類神經網路共組性能判斷之雙核心。此外，於高性能切換式磁阻電動機驅動之考量下，二類就納入電阻值變動或互感影響下之性能提升架構，亦運用電流補償之做法於此論文中進行探討。 驗證所提出之相關論點之研究平台，乃針對可能運用於電動機車與洗衣機的切換式磁阻電動機而組構二系統，其實驗與模擬結果皆部份驗證了所提出架構之效能；此二式切換式磁阻電動機驅動系統實驗平台顯示了於五分之一，二分之一和全載之額定功率下，在效率提升上分別展現了百分之三點五、百分之五和百分之七點一，以及百分之三、百分之二點二和百分之五點一的效率提昇程度。A new control concept, the strategy of efficiency improvement for switched reluctance motor (SRM) drives applying derivation results based on analysis model, is proposed in this dissertation. The presented efficiency consideration and its control approach are inspired and originated from an SRM model, whereas can be realized by regulation of the ratio of the phase current command to voltage within derivatives of equivalent magnetic inductance with respect to time. Moreover, the efficiency improvement strategy is further discussed for constructing the applicable driving scheme in practical usage, operating based on the assigned regulation rule for searching the upgraded efficiency that may exist for the SRM drives by step-type variation of current command. The linking relation of parameters of SRM’s model is utilized to execute commands under operation of SRM drives while no exceeding to the setting ranges according to the outputted performance requirement. For the overall control scheme of the efficiency improving mechanism, a fuzzy neural network (FNN) system is applied to approximately compute the partial derivative of the equivalent magnetic inductance profile for the SRM with respect to the rotor position and current, while the inductance is obtained firstly by the mapping scheme of the FNN for relations among the position, current, and the inductance as well. In addition, several new estimation schemes and measurement approaches for getting the needed parameters, including the parameters of resistance, inductance, mutual inductance, torque, and speed, are also presented for considerations of the integration and combination to the efficiency improving schemes for extending its feasibility. Furthermore, an artificial neural network (ANN) is presented to establish the core unit with outputted performance judgment capability for under-saturation operation, as well as one of the dual-core operation with the FNN’s scheme for high-portion saturation working. Besides, two performance enhancement schemes that can deal with the variation of the resistance or take the mutual inductance into account by current compensation are discussed for the high-performance SRM dives. The research platforms for verification related to these issues are implemented applying two SRMs for possible applications of electrical bikes and washing machines, respectively. Simulation and experimental results partly demonstrated the validity of the capability of the proposed strategy with efficiency improvement up to 3.5 %, 5 %, and 7.1 % for one application target, and 3 %, 2.2 %, and 5.1 % to the other practice, both under the testing of ratio of 0.2, 0.5, and 1, rated power of the applied SRM drives. URI: http://140.113.39.130/cdrfb3/record/nctu/#GT008812821http://hdl.handle.net/11536/57112 Appears in Collections: Thesis

Files in This Item: