一种基于模糊CMAC自学习模糊逻辑系统及其在控制中的应用

CMAC Based Self-Learning Fuzzy Logical System and Its Application to Automatic Control

把 HCMAC(Hyperball Cerebellar Model Articulation Controller)与模糊逻辑理论有机结合起来 ,形成 FHCMAC(Fuzzy HCMAC) ,它便于从输入输出数据中提取模糊规则 ,直接用作控制器 .可以将 FHCMAC看作用基函数网络实现的模糊逻辑系统 ,兼有 HCMAC神经网络和模糊逻辑两者的优点 ,既可以较容易表达定性或模糊的经验知识 ,又具有很好的学习性能 .应用仿真实例验证了其有效性 .

Combining hyperball cerebellar model articulation controller (HCMAC) with fuzzy logic theories, a fuzzy HCMAC (FHCMAC) was obtained, which can extract fuzzy rules for neurocontrol and be used as a controller directly. The controller is of powerful robustness. The FHCMAC can be regarded as a fuzzy logical system implemented by a basis neural network. It has the advantage of CMAC and fuzzy logic systems, namely, on one hand, FCMAC can express qualitative or fuzzy knowledge easily; on the other hand, it has perfect learning ability. The simulation example confirms the advantages. The presented method can be used to extract logic rules from input output data.