摘要
在大规模路网的路径诱导中,研究了基于神经网络的交通信息实时预测方法,构造了具有时变性的路阻矩阵,解决了传统静态路阻存在的局限性问题;探讨了基于并行遗传算法的最优路径求解问题,给出了相应的遗传、变异算子和群体更新方式,提出了矩形限制搜索区域方法,降低了并行遗传算法的搜索范围,解决了遗传算法在大规模路网中求解最优路径时存在的实时性差、收敛速度慢等问题;仿真实验表明该方法满足大规模路网路径诱导的准确性、实时性和快速性要求。
To work out route guidance in gigantic traffic network, the traffic information forecasting method based on Artificial Neural Network is studied in-depth and the time-varied road weight matrixes are constructed, which solve the problem of limitation in traditional and static road weight. The Parallel Genetic Algorithm （PGA） for optimal route choice is discussed in this paper and the corresponding genetic operator, mutation operator and the refresh way of the populations are also proposed. A method of Rectangle Restricted Searching Area （RRSA） which can reduce the searching area of PGA is presented. The problem of bad real-time and astringency of PGA existed in computing the optimal route in gigantic traffic network has also been solved by using RRSA. To probe into the technology of the Route Guidance, a large number of experiments combined with the required analysis of the results have been carried on. It is indicated by simulation that the presented method of optimal route choice has achieve the required accuracy, real-time and quick guidance in gigantic traffic network.
出处
《公路交通科技》
CAS
CSCD
北大核心
2006年第8期126-129,142,共5页
Journal of Highway and Transportation Research and Development
基金
科技部国际重点合作资助项目(2003DF020009)
关键词
神经网络
路阻矩阵
矩形限制搜索区域
并行遗传算法
最优路径选择
neural network
road weight matrix
rectangle restricted searching area
parallel genetic algorithm
optimal route choice
