人工鱼群算法在FPN参数优化中的应用  

Application of Artificial Fish School Algorithm in Parameters Optimization of FPN

作　　者：杨劲松[1] 凌培亮[1] 

机构地区：[1]同济大学机械工程学院,上海201804

出　　处：《计算机工程》2010年第4期169-170,173,共3页Computer Engineering

基　　金：国家自然科学基金资助重点项目(70531020);国家自然科学基金资助项目(70871091;G0525002);上海市科委科研基金资助项目(09DZ1123300)

摘　　要：模糊产生式规则置信度的确定在很大程度上依赖专家的经验,难以获得精确的结果。针对该问题,将人工鱼群算法引入模糊Petri网(FPN)的置信度寻优过程中,提出一种基于改进人工鱼群算法的参数优化算法,不依赖于经验数据,对初始输入无严格要求。实验结果表明,该算法训练出的模糊Petri网参数正确率较高,能提高FPN的自学习能力,降低实际应用难度。Certainty Factor（CF） of fuzzy production rules depends on the experience of experts at a large extent, it is difficult to obtain accurate results. Aiming at this problem, Artificial Fish School Algorithm（AFSA） is introduced into the procedure of exploring the certainty factor parameters of Fuzzy Petri Net（FPN） and an parameters optimization algorithm based on improved AFSA is proposed. It does not depend on experiential data and the requirements for primary input are not critical. Experimental results show that the trained parameters gained from the algorithm are highly accurate and the strong self-learning capability of resultant FPN model can be improved, it reduces the difficulty of the practical application.

关 键 词：人工鱼群算法 模糊PETRI网 置信度 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]