基于遗传-BP算法的FPN参数优化的研究  

Research of Parameters Optimization of Fuzzy Petri Nets Based on Genetic-BP Algorithm

作　　者：李洋[1] 乐晓波[1] 

机构地区：[1]长沙理工大学计算机与通信工程学院,长沙410076

出　　处：《计算机工程》2006年第24期189-191,231,共4页Computer Engineering

基　　金：湖南省教育厅自然科学基金资助项目(01JJY2061);湖南省教育厅科研基金资助项目(01C306)

摘　　要：如何确定模糊产生式规则的各项参数对模糊Petri网(FPN)的建立意义重要,一直是尚未解决的难题。该文把遗传算法与BP算法相结合,引入到模糊Petri网的参数寻优过程,提出了一种基于二阶段的FPN模型的参数优化策略,该策略实现不依赖于经验数据,对初始输入无严格要求。仿真实例表明,经二阶段优化后训练出的参数正确率很高,且所得的FPN模型具有较强的泛化能力和自适应功能。It is significant and being unsolved yet for building a fuzzy Petri net to determine all parameters of fuzzy production rules. Genetic algorithm combined with BP algorithm is originally introduced into the procedure of exploring parameters of FPN. An exploring strategy based on double-stage optimization is proposed. Realization of this strategy don＇t depend on experiential data and requirements for primary input arc not critical. Simulated ~xl^riment shows that the trained parameters gained from above strategy are highly accurate and the resultant FPN model owns strong generalmzing capability and self-adjustmeion purpose.

关 键 词：模糊PETRI网 产生式规则 模糊推理 改进的遗传算法 BP算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]