一种模糊神经网络的改进学习算法  

Improved learning algorithm for fuzzy neural network

作　　者：武彬 沈幼庭[1] 

机构地区：[1]清华大学热能工程系,北京100084

出　　处：《清华大学学报（自然科学版）》1999年第10期31-34,共4页Journal of Tsinghua University(Science and Technology)

摘　　要：针对前人（Ｃ．Ｔ．Ｌｉｎ， ｅｔ ａｌ． Ｉｎ ＩＥＥＥ Ｔｒａｎｓ ＯｎＣｏｍ ｐｕｔｅｒ， １９９１， Ｖｏｌ．４０， Ｎｏ．１２） 提出的模糊神经网络结构，改进了已有的学习算法，新算法更加简洁有效。利用新算法，可以自动获取模糊规则。通过函数模拟实验，验证了新算法的有效性。提出了网络的分解与综合方法，避免了模糊神经网络用于实际复杂问题时，模糊规则的组合爆炸问题。该模糊神经网络可应用于换热器受热面的结垢过程模拟。Based on the network structure presented by C.T.Lin, et.al. In IEEE Trans on Computer, 1991, Vol.40 No.12, an improved learning algorithm was developed. The improved algorithm is simpler and easier to understand. With the improved learning algorithm, fuzzy rules can be derived more efficiently. The improved algorithm was validated by function simulating experiment. A method, namely disassembly and integration of network, was presented to avoid the combination explosion of fuzzy rules in real application. As an example, FNN was used to simulate the fouling process in heat exchanger. The results show the method is practical.

关 键 词：模糊神经网络 学习算法 模糊规则 换热器 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TK172[自动化与计算机技术—控制科学与工程]