多智能神经网络轮胎模型仿真研究  

Modeling and Simulation of Vehicle Tire Based on Multi-Agent Neural Network

作　　者：黄晨[1] 汪若尘[1] 

机构地区：[1]江苏大学汽车与交通工程学院,江苏镇江212013

出　　处：《计算机仿真》2009年第2期289-292,共4页Computer Simulation

基　　金：江苏省高新技术项目(BG2004025)

摘　　要：现在采用神经网络等新技术的高速、高频、大幅度和瞬态变工况的非线性模型是轮胎建模的主要研究问题。以前运用BP算法存在训练速度慢,易陷入局域极值而得不到适当的权值分布等缺点,以至于在实际应用中需要多次进行训练才能得到较为理想的结果。而采用进化算法对神经网络系统进行优化能在很大程度上解决上述问题。再将基于知识的多智能体思维进化算法(KMMEA)引入到神经网络权值优化中,建立了侧向力、纵向力联合工况下的非线性轮胎模型,绘制出有界非线性函数族并整理为非线性不确定系统。在Matlab中的仿真实验结果表明,算法对轮胎模型的精度和建模的效率有显著的提高。The modeling of vehicle high - speed high - frequency transient nonlinear tire based on the neural network is the principal research consideration. But BP algorithm is inefficient and easy to plunge into local solution so as to lack appropriate weighting value, so it must be trained in many times to produce the best results in practice. The problem is solved largely by adopting the Artificial Neural Network （ANN） approach based on Knowledge based Multi -Agent Mind Evolutionary Algorithms. A combined lateral and longitudinal force tire model was built and the bounded nonlinear functions were plotted and arranged to be a nonlinear and uncertain system. The performance of the proposed scheme was verified by simulation experiment using MATLAB.

关 键 词：轮胎模型 多智能体 思维进化 神经网络 

分 类 号：U463.341[机械工程—车辆工程]