作　　者： ; ; ;

机构地区： 湖南大学电气与信息工程学院

出　　处： 《控制理论与应用》 2006年第1期1-6,共6页

摘　　要： 常规的模糊推理系统大多由专家经验建立模糊规则,自学习能力不强.提出了一种支持向量机-模糊推理系统,由支持向量机实现模糊推理系统的自学习,并设计了一种支持向量机-模糊推理自学习控制器.文章给出了自学习控制器的结构和学习算法,对比研究了变尺度梯度优化和混沌优化两种学习算法.针对非线性对象的仿真实验验证了该控制器的优良性能,控制效果比模糊逻辑控制器更好. As conventional fuzzy inference system （FIS） was derived from expert experience, it has poor ability in self-learning or adaptation. The self-learning capability of fuzzy inference system was realized in this paper using support vector machines （ SVM）, and a self-learning controller based on support vector machines-fuzzy inference system （ SVM- FIS） was proposed. Both the structure and learning algorithms of the proposed self-learning controller were analyzed. Two learning algorithms of Multi-scaled Davidon-Fletcher-Powell （MDFP） method and chaotic optimization were compared. Simulation results for a nonlinear system demonstrate that the proposed self-learning controller has better control performance over fuzzy logic controller.

关 键 词： 模糊逻辑 模糊推理系统 支持向量机 自学习

领　　域： [自动化与计算机技术] [自动化与计算机技术]