摘要
针对TSK模糊模型的学习是多约束和多目标优化问题,提出将TSK模糊模型分解为两类不同的种群,协作共同进化的模型学习方法.论述了所涉及的相关问题,包括各种群的编码及其不同的进化计算,各种群个体的合作及其适应值评估策略,模型的后件参数估计方法.该方法要求先验知识少,收敛速度快,能形成简洁的模糊模型,最后以函数近似为例说明了该方法的有效性.
To the problem of multi-constraint and multi-target optimization in learning fuzzy model, TSK fuzzy model is decomposed of twain different species. The method of learning the model by cooperative coevolution is proposed. Some problems related to each species coding and different evolution computing means, individual cooperation and fitness evaluation strategy, consequent parameters estimation, are discussed. The characteristic of the method requests a little of previous information about objects, and is able to obtain compact fuzzy model. An example of function approximation shows the validity of the method.
出处
《控制与决策》
EI
CSCD
北大核心
2005年第4期439-443,共5页
Control and Decision
基金
湖南省自然科学基金项目(04JJY6036)
湖南省科技攻关项目(02JZY2006).
关键词
TSK模糊模型
模糊规则
协同进化
进化计算
Function evaluation
Genetic engineering
Matrix algebra
Optimization
Parameter estimation