摘要
介绍了一种基于最小二乘算法的模糊支持向量机控制器,它将模糊控制与支持向量机结合起来,融合了两者的优点,既有不依赖被控对象模型又有泛化能力强等特点。同时采用混合学习算法来优化控制器参数,即先采用最小二乘算法离线优化支持向量机(SVM)性能参数,建立SVM控制系统,再根据对象的变化,采用遗传(GA)算法在线学习优化SVM性能参数和模糊比例因子,以使控制器的性能能适应对象的变化而达到最优。火电厂主汽温控制的仿真结果表明这种控制器具有良好的控制性能。
The fuzzy support vector controller based on least square algorithms was discussed. The controller integrated fuzzy control and support vector machine together, and digested the advantages of both fuzzy control and support vector networks. It was independent of the controlled object model, and had also good general change ability. The controller parameters were optimized by the hybrid learning algorithm, i.e. in a first step, least square algorithm was used for off-line optimization to form support vector machines(SVM) control system, then genetic algorithm was used for on-line optimization get the optimal control performance on the controlled object. The simulations on electric plant main steam temperature control system show that the controller is of good performance.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2007年第8期76-80,共5页
Proceedings of the CSEE
基金
上海市教育重点科研项目(06ZZ69)
上海市重点学科建设项目(P130
P1301)
关键词
支持向量机
模糊支持向量网络
最小二乘算法
遗传算法
主汽温
support vector machine
fuzzy support vector network
least square algorithm
genetic algorithm
main steam temperature