蚁群和BP算法相结合的模糊Petri网参数寻优  

The Parameters Optimization of the Fuzzy Petri Nets Based on Ant Colony And BP Algorithm

作　　者：周恺卿[1] 乐晓波[1] 唐铭[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：《系统仿真学报》2008年第S2期20-24,共5页Journal of System Simulation

基　　金：湖南省自然科学基金项目(08JJ3124)

摘　　要：在模糊Petri网(FPN)的建立过程中如何确定模糊产生式规则的各项参数是尚未解决的热点问题。在研究蚁群算法和反向传播算法的基础上首次将二者结合,形成ACA-BP算法,并将其运用于FPN的参数寻优过程中。该算法的实现不依赖于经验数据,对初始输入无要求。仿真实例表明,经ACA-BP算法寻优结果令人满意,且得到的FPN模型具有较强的泛化能力和自适应功能。It is a hot and unresolved problem that how to determine the parameters of the fuzzy production rules when we are building the model of the Fuzzy Petri Net(FPN).Based on the theory of ant colony algorithm and back propagation algorithm,we combine these algorithms and pose a new algorithm which is called ACA-BP algorithm at the first time.Apply the algorithm to the process parameters optimization in FPN.The realization of the algorithm does not rely on empirical data,as well as initial input.Simulated result shows that the results given by the ACA-BP algorithm are satisfactory.The final FPN model learned and trained from ACA-BP was provided with strong generalizing capability and self-adjusting purpose.

关 键 词：模糊PETRI网 蚁群算法 反向传播算法 参数优化 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]