基于K-Means和GA-WNN的交通流量预测  

Traffic Flow Prediction Based on K-Means and GA-WNN

作　　者：慕伟[1] 陈国定[1] 钟引帆 

机构地区：[1]浙江工业大学信息工程学院,浙江杭州310023

出　　处：《现代交通技术》2015年第5期70-74,共5页Modern Transportation Technology

摘　　要：如何对交通流进行准确和实时的预测是实现交通管理的关键所在。文章根据交通流数据的时间序列特性,提出基于K-Means算法与遗传算法(GA)优化的小波神经网络(WNN)预测方法:首先对交通流流量序列按照流量采用K-Means算法进行分割,分割后的结果较符合流量的分布情况;然后使用GA-WNN对分割后的每一个时间段的交通流数据分别进行建模和预测。仿真结果表明,该方法对交通流量预测的精度较好。Accurate and real-time traffic flow prediction is the key to the implementation of traffic management. According to the characteristic of time series of traffic flow, a prediction method based on K-Means and wavelet neural network（WNN） optimized by genetic algorithm（GA） is proposed. Firstly, traffic flow sequence is segmented according to the flow using the K-Means algorithm, the segmented results is in agreement with the distribution of flow; and then the GA-WNN method is used for modeling and forecasting of the segmented traffic flow data respectively. The simulation results show that the traffic flow prediction accuracy of the method is better.

关 键 词：交通流量预测 K-MEANS算法 遗传算法 小波神经网络 

分 类 号：U491.112[交通运输工程—交通运输规划与管理]