摘要
针对常规遗传算法的缺陷,提出了一种依据新的种群"早熟"程度评价指标和自适应调整算法参数的遗传算法,并以单级倒立摆为被控对象,采用基于改进遗传算法的RBF模糊神经网络控制器进行了仿真研究。仿真结果表明,改进的遗传算法是有效的,新型模糊神经网络控制器的控制效果优于FLC和LQR控制。
In view of the shortcomings of the normal genetic algorithms, an improved self - adaptive GA based on a new evolution index of the premature degree of population is proposed. The fuzzy RBF neural network controller, which is used in controlling the single inverted pendulum, is base on the proved GA.The simulation shows the improved GA is feasible and the control effect of this controller is superior to that of FLC and LQR.
出处
《电机与控制学报》
EI
CSCD
北大核心
2005年第6期617-620,共4页
Electric Machines and Control
关键词
遗传算法
RBF
自适应
genetic algorithms
RBF
self - adaptive