基于CS算法优化的SVM短时交通流预测模型

Short-term traffic flow prediction model based on the optimization of SVM by CS algorithm

为了提高短时交通流预测模型的准确度,提出一种基于布谷鸟搜索算法(Cuckoo Search,CS)优化的支持向量机(Support Vector Machine,SVM)短时交通流预测模型(CS-SVM)。选取青岛市内的多组典型城市路段作为研究对象,将观测收集的车流量数据作为学习样本。利用CS算法对SVM模型的主要参数进行优化,建立以SVM为基础的短时交通流预测模型。最后将CS-SVM模型与多种现有模型进行仿真分析。结果表明,CS-SVM模型相比其他传统模型具有更低的预测误差和更好的稳定性,CS-SVM模型相比SVM模型的MAE值下降了6.56%,RMSE值下降了7.36%。因此该模型能够为城市交通出行和交通流理论研究提供有效帮助。

In order to improve the accuracy of short-term traffic flow prediction model,a prediction model based on CS-SVM is proposed in this study.This model uses cuckoo search(CS)algorithm to optimize support vector machine(SVM).Several groups of typical urban road sections in Qingdao are selected as the research objects.The observed and collected traffic flow data are taken as samples for learning.CS algorithm is used to optimize the main parameters of SVM model.And a short-term traffic flow prediction model based on SVM is established.Finally,CS-SVM model is simulated with several existing models.The results show that CS-SVM model has lower prediction error and better stability than other traditional models.Compared with SVM model,the MAE value of CS-SVM model decreased by 6.56%and the RMSE value decreased by 7.36%.Therefore,CS-SVM model can provide effective help to improve urban traffic and enhance the theoretical research on traffic flow.