駱尚廉2006-07-262018-06-282006-07-262018-06-282002-10-31http://ntur.lib.ntu.edu.tw//handle/246246/22092本研究提出一種新方法論—類神經模糊規則 轉換法—此法是依據實際系統操作所取得之數據 藉由類神經網路學習後，從網路之中擷取符合該系 統之操作控制法則。類神經網路於此方法論中是以 其具備高度學習能力而扮演專家的角色，學習過後 的網路有如一個專家而其知識是隱含於之鏈結權 重之中，透過轉換步驟可將該知識自網路中擷取出 來並實現成模糊規則，再由取出之規則建立模糊控 制器，並應用於廢水處理單元以進行處理效率之預 測。為了驗證我們所提出的操作支援模式，本計畫 在第二年裡設計了一個活性污泥與接觸氧化的串 聯系統的生物處理系統以供第三年驗證模式使用。Artificial neural networks (ANN), which can learn the historical data of a plant, provide operational guidance for plant operators, and fuzzy systems (FS) establish a framework that sets operators' control experiences into fuzzy rules. In this research, an interpretation of ANN is provided so that ANN will no longer be seen as a black-box. This is shown by establishing the equality between a certain class of ANN and FS. In addition, an automated knowledge acquisition procedure is obtained, employing ANN as a learning machine and a mathematical model from which to extract fuzzy rules. For testifying our model purpose, a two-phase biological treatment system of activated sludge/contact aeration process was established in this year.application/pdf2182072 bytesapplication/pdfzh-TW國立臺灣大學環境工程學研究所模糊控制模糊推論類神經網路類神經模糊控制器Fuzzy SystemNeural Networkreporthttp://ntur.lib.ntu.edu.tw/bitstream/246246/22092/1/902211E002051.pdf