摘要
为了能够提升短时交通流量预测的性能及精度,在支持向量回归机算法基础上引入了经改进的粒子群算法做参数寻优。同时,针对短时交通流量具有不确定性和弱规律性等特点,本文通过对交通数据进行挖掘,得出相邻路口交通流量的时空关联性,构建出通过历史数据预测及结合时空关联性的预测模型。利用BP神经网络在线修正两者权值,让两者优势互补、实时更新。通过仿真实验表明:提出的预测模型可以达到预期结果,对短时流量的预测精度和性能都有所提高。
In order to improve the accuracy of short-term traffic flow prediction, this paper introduces an im-proved Particle Swarm Optimization (PSO) algorithm to optimize the parameters of traditional SVR prediction model. In view of the uncertainty and weak regularity of short-time traffic, moreover, this paper creates a prediction model based on historical data prediction and spatio-temporal correlation by digging traffic data in adjacent inter-section. The BP neural network is used to modify the weights of the two in-line, so as to renew the two advantages. The simulation results show that the model can effectively improve the prediction performance and accuracy of short time traffic flow.
出处
《软件》
2018年第1期126-131,共6页
Software
关键词
短时交通流量预测
支持向量回归机
粒子群
时空关联性
Short-term traffic flow prediction
Support vector regression
Particle swarm optimization
Time-space correlation