优化模糊Petri网参数的一种混合智能算法  

A HYBRID ALGORITHM FOR PARAMETERS OPTIMIZATION OF FUZZY PETRI NET

作　　者：靳文辉[1] 葛洪伟[1] 

机构地区：[1]江南大学信息工程学院,江苏无锡214122

出　　处：《计算机应用与软件》2008年第9期252-254,共3页Computer Applications and Software

摘　　要：确定模糊产生式规则的各项参数对模糊Petri网(FPN)的建立具有非常重要的意义,是目前研究热点之一。提出了一种充分结合量子粒子群优化算法QPSO(Quantum-behaved partide swarm optimization algorithm)和BP网络学习算法各自优点的混合智能算法HQBA,并将其引入到模糊Petri网的参数寻优过程。仿真实例表明,这种混合算法计算简单,收敛速度快,能够明显减少迭代次数,具有更好的全局收敛性能。由此训练出的参数正确率较高,所得的FPN具有很强的泛化能力和自适应性。The determination of the parameters of fuzzy production rules is significant for the construction a fuzzy Petri net, and it is one of the research hotspots. A hybrid algorithm HQBA is proposed, which takes full advantages of Quantum-behaved particle swarm optimization （QPSO） algorithm and BP neural networks learning algorithm. The hybrid algorithm is introduced into the procedure of exploring the parameters of FPN. Simulated experiment shows that this hybrid algorithm has the characteristics of easier computation, faster convergence speed, less interation, and better global convergence ability compared with other traditional learning algorithms. The parameters gained from this algorithm are highly accurate, and the resultant FPN model owns strong generalization eapability and self-adaptability.

关 键 词：模糊PETRI网 产生式规则 模糊推理 BP网络学习算法 量子粒子群优化算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP301.6[自动化与计算机技术—控制科学与工程]