基于自适应混合变异进化策略的神经模糊系统及应用研究  

The Research on Evolutionary Strategy with Self-adapting Hybrid Mutation based Neuro-Fuzzy System

作　　者：贾立[1] 俞金寿[1] 

机构地区：[1]华东理工大学自动化研究所,上海200237

出　　处：《系统仿真学报》2001年第z1期122-125,共4页Journal of System Simulation

摘　　要：针对现存神经模糊系统中存在的问题,提出了基于自适应混合变异进化策略的神经模糊系统:采用改进的最近邻域聚类算法对输入空间进行模糊聚类,确定模糊规则数以及模糊规则前件,这样做精简了模糊规则,不会因输入变量的增加而造成“维数灾难”;采用自适应混合进化策略确定模糊规则的后件,明显提高了算法的收敛速度和精度。将本文提出的基于自适应进化策略的神经模糊系统用于某炼油厂航煤干点的软测量建模,结果表明,该系统具有结构简单、建模精度高、泛化能力强等优点。In order to avoid the disadvantages in neuro-fuzzy system, a novel evolutionary strategy with self-adapting hybrid mutation based neuro-fuzzy system is proposed in this paper. The nearest-neighborhood clustering algorithm is used to classify the input space and find the number of fuzzy rules and antecedence simplifying the structure of fuzzy rules and avoiding dimension disaster. And evolutionary strategy with self-adapting hybrid mutation is adopted to determine the consequence of fuzzy rules improving the convergence and precision. On the basis of theory research, the algorithm presented in this paper is applied to practical application of modeling jet point. Experiments results show that it possesses better generalization ability and simple model structure.

关 键 词：神经模糊系统 进化策略 自适应变异 聚类 

分 类 号：TP[自动化与计算机技术] 18