弹药自动装填机器人免疫遗传模糊神经滑模控制  

Fuzzy Neural Sliding Mode Control of Immune Genetic Algorithm for Ammunition Auto-loading Robot

作　　者：李郁峰[1] 李魁武[2] 潘玉田[1] 李科蕾[1] 

机构地区：[1]中北大学计算机与控制工程学院,太原030051 [2]西北机电工程研究所,陕西咸阳712099

出　　处：《火力与指挥控制》2014年第12期36-39,共4页Fire Control & Command Control

基　　金：国家自然科学基金资助项目(51275489)

摘　　要：提出了一种基于改进免疫遗传算法的弹药自动装填机器人自适应模糊神经滑模控制器(IIGAAFNSMC)。用径向基神经网络来近似等效滑模控制中的不确定参数,通过自适应免疫遗传算法在线调整径向基神经网络非线性隐含层的结构和参数。利用最小二乘法计算线性输出层的权值,自适应模糊系统调节滑模控制的增益,减小了网络逼近误差和外部干扰并消除了传统滑模控制中的抖振问题。仿真结果表明,该方法比传统的神经网络滑模控制器具有更高的逼近精度和速度。A adaptive fuzzy neural sliding mode controller based on improved immune genetic algorithm （IIGAAFNSMC）for ammunition auto-loading robot is proposed.In the scheme,RBFNN is used to approximate the internal uncertain parameter in equivalent sliding mode control,the structure and parameters of RBFNN nonlinear hidden layers are tuned online according to adaptive immune genetic algorithm,and weights of RBFNN linear output layers are computed with least square method. In order to guarantee the stability and the convergence of the system,the sliding mode control gain is adjusted by the adaptive fuzzy systems to compensate the network approximation error and the external disturbances. The simulation results prove this method is more significantly higher of approaching precision and speed than that of RBFNN sliding mode controller.

关 键 词：滑模控制 RBFNN 免疫遗传算法 自适应模糊控制 机器人 

分 类 号：TJ303.3[兵器科学与技术—火炮、自动武器与弹药工程] TP242[自动化与计算机技术—检测技术与自动化装置]