电动液压助力转向系统的控制算法研究_夏长高.pdfVIP

36 3 V o.l 36 N o. 3 2009 6 T acto F a m T anspo te June, 20 09 夏长高, 陈 松, 贺建军, 金永利 ( , 212013) : 首先针对电动液压助力转向( EH PS )系统助力特性非线性特点, 将 BP 神经网络引入 EH PS 系统中因 BP 神经网络存在 易 入局部最小缺点, 提出了利用遗传算法来优化 BP 神经网络的连接权值的改进型 BP 神经网络(简称 EN N) 控制算法, 对助力特 性曲线进行全车速拟合通过对拟合结果的分析得出ENN 算法具有收敛速度快泛化能力强的特点, 准确实现了非线性助力特性 的全车速拟合, 克服了 EHPS系统的助力盲区最后, 通过台架试验验证该 ENN控制算法的可行性 : EHPS; ; ; : U 463. 4 : A : 1006- 0006( 2009) 03- 0073- 03 Resea rch of Contro lA lgo rithm on E lectro-Hyd raulic Pow e r Steering System XIA Chang-gao, CHEN Song, HE J ian-jun, JIN Yong-li ( Jiangsu U n ive sity, Zhenj iang 212013, Ch ina) Abstract: F i stly, F o influenc ing facto s of assisted cha acte s of EH PS wh ich had a non- linea cha acte istic, the a tific ia l neu a l netwo ks w e e inco po ated in to EHPS sy stem. B ecause the back-p opaga tion ( BP ) a lgo ithm w as easy to fall in to local m inmi um po int, an evo lutiona y neu al netwo ks ( ENN) , wh ich t ained a BP neu a l netw o k by genetic algo ithm w as deve loped and fitting in full speed is m ade. F om the fitting esu lt, it is found that the ENN has fast conve gence ate, h igh g ene alization and can m ake an accu a te fitting in fu ll speed, so the blind zone is e lmi inated. L astly, it is p oved tha t the ENN is feasib le by test of test bed. Key wo rds: EH PS; A ssist cha ac te istics; G enetic a lgo ithm; N eu a l netwo ks ( EHPS) , [ 1] , , , [ 2] , , , ,