论文部分内容阅读
提出输入层具有一定隶属度的模糊小脑模型神经网络（ＦｕｚｚｙＣＭＡＣ），它比小脑模型ＣＭＡＣ（ＣｅｒｅｂｅｌａｒＭｏｄｅｌＡｒｔｉｃｕｌａｔｉｏｎＣｏｎｔｒｏｌｅｒ）能更真实地描述客观世界．给出ｎ维ＦｕｚｚｙＣＭＡＣ算法，仿真结果表明ＦｕｚｙＣＭＡＣ比小脑模型ＣＭＡＣ具有如下优点：学习收敛速度快得多，可以学习模糊规则．ＦｕｚｙＣＭＡＣ比ＣＭＡＣ优越，使ＣＭＡＣ成为ＦｕｚｚｙＣＭＡＣ的特例．
The fuzzy CMAC neural network (FuzzyCMAC) with a certain degree of membership is proposed, which can describe the objective world more faithfully than the cerebellar model CMAC (Cerebelar Model Modeling Controller). The n-dimensional FuzzyCMAC algorithm is given. The simulation results show that FuzyCMAC has the following advantages over the cerebellar model CMAC: learning convergence is much faster and fuzzy rules can be learned. FuzyCMAC is superior to CMAC, making CMAC a special case of FuzzyCMAC.