基于粗糙集模糊神经网络的自适应逆控制  

Adaptive Inverse Control Based on Rough Fuzzy Neural Networks

作　　者：江秀红[1] 董宁[2] 胡爱玲[1] 

机构地区：[1]沈阳航空工业学院自动控制系,辽宁沈阳110034 [2]北京理工大学自动控制系,北京100081

出　　处：《计算机仿真》2008年第3期186-188,198,共4页Computer Simulation

摘　　要：粗糙集和模糊神经网络在智能信息处理方面各有优缺点,若将二者结合起来可增强信息处理的能力。将粗糙集理论中的贪心算法和缺省规则获取算法进行结合改进,并将结合算法应用到补偿模糊神经网络的输入模糊化和规则提取中。最后针对某型火炮伺服系统输出端噪声对重定位的影响,用粗糙集模糊神经网络对受扰对象进行逆建模,并将学习得到的模型作为逆控制系统的控制器来消除扰动,仿真结果表明此种网络在精简决策规则,缩短训练时间,提高误差精度等方面都有显著改善。Rough set and Fuzzy Neural Networks have different advantages in intellectualized information processing, so it can enhance the processing ability by adopting them together. This paper combines the Greed algorithm with default - rules generation algorithm in Rough Set, then applies the combined method to the Compensation Fuzzy Neural Networks to discrete input values and generate rules. Finally, the paper builds the inverse model for artillery servo system affected by sensor disturbance via rough fuzzy networks, and constitutes inverse control system to eliminate the disturbance in order to position accurately again. The simulation indicates that the ameliorative networks have the advantages of simplifying rules, shortening training time and improving error precision etc.

关 键 词：粗糙集 补偿模糊神经网络 逆控制 火炮伺服系统 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]