基于GRNN网络和遗传算法旋翼动平衡调整.PDFVIP

基于GRNN网络和遗传算法的旋翼动平衡调整： 2008 5 M ay 200 8 34 5 Journa l o f Be ijing U nivers ity of A eronautics and A stronautics V o.l 34 N o5 GRNN 刘红梅 王少萍 欧阳平超 ( , 100083 ) 针对传统旋翼调整方法没有考虑调整参数与振动信号 间的非线性关系, 提出一种结合广义回归神经网络GRNN ( G eneral R eg ression N eural N etwo rk) 和遗传算法的旋翼 调整方法, 采用GRNN 建立旋翼动平衡调整模型, 以桨叶调整参数作为GRNN 输入, 以旋翼转 轴 3 个方向的加速度测量值和机身 3 个方向加速度测量值作为网络输出, 建立调整参数与直 升机振动信号 间的模型. 以直升机振动作为目标函数, 采用改进的遗传算法对桨叶调整参数 进行寻优, 获得直升机振动最小时的桨叶的调整量. 飞行实验表明, 通过 1到 2 次飞行调整, 可 使 3 个方向机身振动( 旋翼的一阶振动)为最小, 完成旋翼的动平衡调整. 旋翼; 动平衡; 广义回归神经网络( GRNN) ; 遗传算法; 优化 V 267 A ( 2008) He licopte r roto r smooth ing based on GRNN neura l netw o rk and genetic a lgo rithm L iu H ongm e i W ang Sh aop ing Ouyang P ingch ao ( S chool of A utom ation S cien ce and E lectrical E ngineering, B eijing U n iv ers ity of A eronau tics and A stron aut ics, Beijing 100083, Ch ina) Abstract: Considering traditiona l adjustm ent m ethod w ithout calcu lat ing possible nonlinear be wt een ro tor ad ju stm ents and fu se lage v ibrat ion signa ls, a new rotor ad ju stm ent m ethod b ased on genera l regression neura l netw ork ( GRNN) and genetic algo rithm w as presented. GRNN network w as em p loyed to m ode l the relat ion sh ip betw een the rotor adjustm ents and the fu se lage v ibration s, w hose inputs are ro tor ad ju stm ent p aram eters and whose outputs are acce leration m easurem ents along the three axes of rotor sh aft and the fuse lage. W ith he licopter v ibrat ion as ob jective function, genetic