一种粗糙神经网络的Simulink仿真研究  

Simulink of A Rough Neural Network

作　　者：常志玲[1] 赵旭鸽 CHANG Zhi-ling;ZHAO Xu-ge(Academy of Information Technology,Luoyang Normal University,Luoyang 471934,China)

机构地区：[1]洛阳师范学院信息技术学院,河南洛阳471934

出　　处：《电脑知识与技术》2020年第17期7-9,共3页Computer Knowledge and Technology

基　　金：河南省高校科技创新团队项目(编号:18IRTSTHN014)。

摘　　要：本粗糙神经网络由三层构成,其输入层由上下两个经典神经元构成的粗糙神经元组成,并与隐层之间为全连接,隐层与输出层也为全连接,用离散并约简后的一对输入向量的边界值作为输入向量。对于这种粗糙神经网络,利用Simulink工具进行了仿真,并详细叙述了整个建模过程,用经典实例进行了验证,实例表明本建模工具能够快速准确地处理复杂的粗糙神经网络问题,并且参数设置也比较方便。A rough neural network is made up of three layers,input layers are made up rough neurons,and each rough neuron is made up of a pair of typical neurons,and each input vector is a pair of boundary values of discrete interval of each condition attri⁃bute.The rough neurons of input layers and the typical neurons of hidden layers are linked completely,and the typical neurons of hidden layers and the typical neurons of output layers are linked completely too.The inconsistent decision table is reduced by the reduction algorithm based on variable precision rough set,and the input vectors of neural network is reduced on a large scale.It is simulinked by matlab/simulink.The experimental results show that the Simulink tool can deal with the complex rough neural net⁃work problem quickly and accurately,and the parameter setting is convenient.

关 键 词：变精度粗糙集 神经网络 SIMULINK 分类算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]