一种模糊支持向量机控制器的研究

将支持向量机与模糊逻辑相结合,设计了一种模糊支持向量机控制器,并分析了控制器的结构和学习算法.学习过程分为离线学习支持向量机和在线整定模糊比例因子两部分.与模糊神经网络控制器相比,模糊支持向量机控制器适应小样本学习,泛化能力强,解决了过学习、结构设计依赖经验等问题.仿真研究表明,所设计的控制器具有较优的控制性能.

On fuzzy support vector machine controller

A fuzzy support vector machine (FSVM) controller is proposed which integrates fuzzy logic and support vector machines (SVM). The structure and learning algorithms of proposed FSVM controller is analysed. And its learning procedure includes two parts: offline training of SVM and online training of fuzzy scale factors. Compared with fuzzy neural networks (FNN) controller, FSVM controller has smaller training-size, better generalization (ability,) and avoids problems of over-fitting, experiential selection of structure. Simulation results show that FSVM controller has good control capability.