摘要
基于一类参数化的模糊量表示 ,将模糊量同精确量统一起来。在此基础上提出规则增加、规则删除、规则修正算法 ,进行了一类自学习模糊控制器的设计。对二阶系统的参数鲁棒性仿真实验表明 ,与常规模糊控制器相比 ,这类自学习模糊控制器对控制对象的某些参数变化有较强的适应性。
To improve the performance of fuzzy rules, a class of self-learning fuzzy controller is presented. Firstly, the fuzzy variable with Gauss membership function is represented by two parameters. Based on it, the analysis of the relation between fuzzy variable and precise variable is given. Then, with the nwe representation of fuzzy variable, the fuzzy rules are simply parameterized. According to the simple structure of fuzzy rules, a new algorithms of rules′ adjustment is given, including rule-adding, rule-deleting and rule-modification. The self-learning algorithm here is much simpler and more systematic than others that based on liguistic rules. Robustness experiments about two order plants are given, which show that the new self-learning fuzzy controller has better robustness than common fuzzy controller and PID controller.
出处
《控制与决策》
EI
CSCD
北大核心
2000年第3期286-289,共4页
Control and Decision
