基于BP神经网络的换档品质评价方法  

Methods for shift quality evaluation based on BP neural network

作　　者：张建国[1] 雷雨龙[1] 王健[1] 陆晓惠[1] 

机构地区：[1]吉林大学汽车动态模拟国家重点实验室

出　　处：《吉林大学学报（工学版）》2007年第5期1019-1022,共4页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金资助项目(50505014)

摘　　要：换档品质评价方法的研究是应现代车辆自动变速技术发展需要而提出的前沿研究课题,其评价过程可视为一个非线性动态系统。BP神经网络的非线性系统辨识,能够逼近任意一个非线性函数。通过确定换档品质评定指标,利用BP网络训练获得的数据样本,从而建立主观、客观评价标准之间的联系;利用Matlab/Simulink完成换档品质评价方法的客观描述并进行仿真。通过与实验结果对比,证明这种方法能够有效真实地评价换档品质并与传统主观评价方法具有很好的一致性。The study of the gear shift quality evaluation is required by the development of the vehicle automatic transmission. The evaluation process is unsteady and non-linear which may be identified and approximated by the back-propagation neural network（BPNN）. A BPNN was set and trained by a set of data sample obtained by the definition of the gear shift evaluation indexes. The relation between the subjective and the objective evaluations has been established. The objective description of the gear shift quality evaluation method was performed and simulated using the Matlab/Simulink. A comparison of the simulated results with the test ones shows that the proposed method is in good agreement with the traditional subjective evaluation method, can evaluate the vehicle gear shi/t quality properly.

关 键 词：车辆工程 换档品质 评价方法 BP神经网络 评价指标 

分 类 号：U270.1[机械工程—车辆工程]