摘要
在分析区域公路交通事故致因因素和预测特点的基础上,引入了基于改进PSO算法的RBF神经网络的混合优化(MPSO-RBF)算法,即将PSO算法的全局搜索能力和RBF神经网络局部优化相结合,并建立了区域公路交通事故的预测模型.最后,利用某城市1990-2003年交通事故资料和相关数据对MPSO-RBF神经网络预测模型进行了训练、拟和,同时用2004-2006年的外推样本数据对模型进行了检验,计算结果表明,MPSO-RBF预测模型较传统方法具有更高的预测精度,与此同时也证明了本文所选取区域公路交通事故致因因素的有效性.
Based on analyzing the influential factors and the forecast characteristic of regional road traffic accident,a hybrid optimized algorithm (MPSO-RBF)for radial basis function (RBF)neural networks based on modified particle swarm optimization (MPSO)was introduced in the paper, namely, combining global searching performance of MPSO with the local optimization of RBF neural networks, and founding a model to forecast regional road traffic accident. The historical data of traffic accident of a city during the period of 1990-2003 was used to train the MPSO-RBF neural networks forecasting model, then the testing data of 2004-2006 was tested. The result shows the forecasting model of MPSO-RBF has higher forecasting precision than traditional methods,and the influential factors of regional road traffic accident proposed in paper are feasible.
出处
《兰州交通大学学报》
CAS
2008年第4期75-79,共5页
Journal of Lanzhou Jiaotong University
基金
公安部应用创新计划项目(2006YYCXGSSS7021)
甘肃省科技计划资助项目(0804GKCA038)
关键词
改进PSO算法
RBF神经网络
混合优化算法
区域公路交通事故
预测模型
modified particle swarm optimization (MPSO)
RBF neural networks
hybrid optimized algorithm
regional road traffic accident
forecast model