基于模糊系统聚类学习的自适应控制算法

在自适应控制最小方差自校正控制器设计中，当被控对象的数学模型未知时，可采用模糊系统代替实际系统。提出了一种新的模糊系统的聚类学习算法，根据初始聚类中心的选取原则，可以使最终获得的聚类结果是全局近优解。该方法只需计算一遍样本间的广义距离，即可完成初步的聚类，通过迭代运算可以使聚类结果得到进一步优化。仿真结果证明了自适应控制器的控制效果。

Adaptive Control Algorithm Based on Fuzzy System Clustering

For minimal square error self-emendation controller design in adaptive control,the plant math model is unsure.This paper discusses the method of the replacing real system model with a fuzzy system.A new clustering algorithm for fuzzy system was proposed;we can obtain a clustering result that is near to the global optimal solution base on the selection rule of the original cluster center.A primary solution that is near to the global optimal solution can be completed only calculating the distant between the data for one time; we can further optimize the solution if we use the iterative algorithm. The emulated result of adaptive controller proves the control effect