摘要
T-S模糊模型与自适应神经模糊推理系统(ANFIS)的结合便于描述多输入系统模糊控制规则.为解决规则前件过多时传统ANFIS结构导致的维数灾难问题,同时进一步提高ANFIS对于复杂系统T-S规则在线获取速度,首先采用多维输入向量对ANFIS网络进行修正,在此基础上提出了T-S模糊控制规则聚类获取方法;其次,利用所提出的方法分别对倒立摆和二阶滞后系统进行了控制仿真,该方法同Mamdani规则自组织模糊控制的控制效果比较表明两者的最大超调量、振荡次数、过渡时间基本一致,上升时间要较Mamdani模糊控制器缩短3个采样周期,控制规则较Mamdani控制器减少了45条.
The combination of T-S fuzzy model and adaptive-network-based fuzzy inference system（ANFIS） is convenient for description of multi-input fuzzy control rules. In order to solve dimension explosion problem of conventional ANFIS structure under large amount of rule antecedent and improve online extraction speed for complex systems,ANFIS network is firstly modified adopting multi-input vector,then a clustering obtainment method of T-S control rules is presented. Simulation work towards inverted pendulum and the second-order delay system is conducted,and comparative results of the presented method and Mamdani-type controller show similar control performances such as maximum overshoot,fluctuation times and transition time. Meanwhile,the presented method has 3 sampling periods shorter rise time and 45 decreased control rules than Mamdani-type controller.
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2010年第4期580-585,共6页
Journal of Dalian University of Technology
基金
国家自然科学基金资助项目(50578049)
