粗糙集CMAC神经网络及其在非线性系统辩识中的应用

Rough Set CMAC Neural Network and Its Application in Nonlinear System Identification

提出了一种基于粗糙集规则提取的CMAC神经网络非线性系统辩识策略。该策略利用粗糙集理论对数据样本进行数据浓缩,提取初步的映射规则。对初步的规则通过神经网络进行粗映射,利用神经网络的分类逼近能力,建立输入状态空间到输出空间的精确映射,大大提高了神经网络的收敛速度和逼近精度。通过一个非线性系统对该神经网络进行了实验,结果表明,该神经网络具有分类逼近能力强、计算量小等优点。

In this paper, based on complementary strategies, a new AI method based on rough set CMAC neural network is put forward to nonlinear system identification. Using rough set theory to enrich the data, extract the rules from the sample data. Then exploiting rough set CMAC neural network to describe the extracted rules, the approach to construct the exact mapping between input state-space and output space is presented. The example of identification of nonlinear system is demonstrated to the convergence ability of the suggested network.