粗糙集理论在补偿模糊神经网络中的应用

pplication of Rough Set in Compensation Fuzzy Neural Networks

该文首先引入一种具有快速算法的补偿模糊神经网络。通过对粗糙集理论中的贪心算法进行改进,提出一种新的模糊化方法,并将此方法运用到补偿模糊神经网络的输入模糊化和规则提取中。通过用MATLAB编制程序进行仿真研究,证明改进后的网络与原补偿模糊神经网络相比,在精简决策规则、缩短训练时间、提高误差精度等方面都有显著改善。最后将改进后的网络应用到某位置伺服系统的扰动消除控制中,仿真结果表明此方法的有效性。

This paper introduces a kind of compensation fuzzy neural networks having quick learning speed, and proposes a new kind of fuzzification method via improving the Greed algorithm in Rough Set. Then the method is applied to the compensation fuzzy neural networks to descerete input values and generate rules. The result of emulation programmed by MATLAB indicates that the ameliorative networks have the advantage of simplified rules, shorter training time and increased error precision etc compared to the originally networks. Finally, the approach is applied to a bearing artillery servo system to eliminate disturbance,the result of simulation shows its effectiveness.