Full metadata record
DC FieldValueLanguage
dc.contributor.author梁世璋en_US
dc.contributor.authorShih-Chang Liangen_US
dc.contributor.author鄧清政en_US
dc.contributor.authorChing-Cheng Tengen_US
dc.date.accessioned2014-12-12T02:21:44Z-
dc.date.available2014-12-12T02:21:44Z-
dc.date.issued1998en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#NT870591004en_US
dc.identifier.urihttp://hdl.handle.net/11536/64931-
dc.description.abstract在本論文中我們提出了一個新的模糊類神經網路的訓練方法,它是結合倒傳遞學習法則以及基因遺傳演算法則來訓練模糊類神經網路,我們突破了以往將歸屬函數以固定的函數形式(如三角形、梯形、鐘形)表示的限制,而是以許多的線段來組成歸屬函數。模糊類神經網路在經由倒傳遞學習法則訓練後,將其歸屬函數分割取樣以線段的形式表示,也使得我們能以字串的形式表達一組歸屬函數,接著再以基因遺傳演算法來搜尋每個分割點上歸屬函數值的最佳值,以使得模糊類神經網路的誤差函數值能再進一步的降低。經由模擬的結果顯示此訓練方法確實使得模糊類神經網路有更好的近似能力。zh_TW
dc.description.abstractIn this thesis, we propose a new method, which combining the backpropagation and the genetic algorithms, to train a fuzzy neural network (FNN). Here, we make a breakthrough of the restriction of membership function to be unified form (ex: triangular form, trapezoid form and shape of bell). Here, membership functions of the FNN are constructed by a group of line segment. The proposed training algorithm can be described as: (a) First, we train the FNN using the bakepropagation algorithm to obtain membership functions. (b) Membership functions with a group of line segment by partitioning and sampling themselves are constructed. Thus we can represent membership functions in a string form. (c) Finally, for every partition point, we use the genetic algorithm to search the optimal value and obtain the better membership functions. Therefore, the approximated error of the FNN can be reduced. That is, the approximated accuracy is improved by our approach. Simulation results show that the mapping capability of the FNN be train by proposed method is much better.en_US
dc.language.isozh_TWen_US
dc.subject模糊類神經網路zh_TW
dc.subject基因遺傳演算法zh_TW
dc.subject倒傳遞學習法則zh_TW
dc.subjectFuzzy Neural Networken_US
dc.subjectGAsen_US
dc.subjectbakepropagation algorithmen_US
dc.title利用基因遺傳演算法訓練模糊類神經網路zh_TW
dc.titleUse of Genetic Algorithms in Training of Fuzzy Neural Networken_US
dc.typeThesisen_US
dc.contributor.department電控工程研究所zh_TW
Appears in Collections:Thesis