摘要
在车辆跟驰现象中,驾驶员-车辆系统可视为一个非线性的动态系统,而人工神经网络(ANN)是开发非线性系统模型的有效工具,采用ANN技术建立了车辆跟驰模型,开发了基于粒子群优化(PSO)算法的ANN训练算法。测试结果表明,基于神经网络的跟驰模型比传统模型具有更强的鲁棒性,基于PSO算法的ANN训练方法能够避免陷入局部最优。
The driver-vehicle system in Car-following process is a non-linear dynamic system, and the artificial neural network（ANN） is an effective tool to develop a model for the non-linear system. A car-following model was established based on the ANN to simulate the traffic phenomena as real as possible. A training algorithm for the ANN in this model was also proposed on the basis of the particle swarm optimization（PSO） algorithm. The test results show that the ANN based model is more robust than traditional models and the training algorithm based on the PSO algorithm may avoid falling into the local optimization.
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2009年第4期896-899,共4页
Journal of Jilin University:Engineering and Technology Edition
基金
国家自然科学基金项目(50338030/E0809)
关键词
交通运输系统工程
跟驰模型
人工神经网络
粒子群优化算法
engineering of communications and transportation system
car-following model
artificial neural network（ANN）
particle swarm optimization（PSO）
