基于粒子群优化投影寻踪回归模型的短时交通流预测  

Short–term traffic flow prediction method based on particle swarm optimization projection pursuit regression model

作　　者：邴其春[1,2] 龚勃文[1] 林赐云[1] 杨兆升[1] 曲鑫[1] BING Qichun GONG Bowen LIN Ciyun YANG Zhaosheng QU Xin(College of Transportation, Jilin University, Changchun 130022, China College of Automobile and Transportation, Qingdao Technological University, Qingdao 266520, China)

机构地区：[1]吉林大学交通学院,吉林长春130022 [2]青岛理工大学汽车与交通学院,山东青岛266520

出　　处：《中南大学学报（自然科学版）》2016年第12期4277-4282,共6页Journal of Central South University:Science and Technology

基　　金：国家高技术研究发展计划项目(2012AA112307);国家科技支撑计划项目(2014BAG03B03);国家自然科学基金资助项目(51308248;51408257);吉林省科技发展计划青年科研基金资助项目(20140520134JH)~~

摘　　要：针对短时交通流数据的高度复杂性、随机性和非稳定性,为了进一步提高短时交通流预测的精度,提出一种基于粒子群优化投影寻踪回归模型的短时交通流预测方法。通过灰色关联度分析确定交通流预测影响因子,然后采用粒子群优化算法构建非参数投影寻踪回归模型,并利用上海市南北高架快速路的感应线圈实测数据进行实验验证和对比分析。实验结果表明:PSO-PPR模型的短时交通流预测效果明显提高,其平均预测精度分别比ARIMA模型和BPNN模型提高37.8%和27.2%。Considering the highly complexity, randomness and non-stability characteristics of short-time traffic flow data, a short–term traffic flow prediction method based on particle swarm optimization projection pursuit regression model was put forward. Traffic flow forecasting impact factors were determined by grey relational analysis. Then the projection pursuit nonparametric regression traffic flow forecasting model was constructed using particle swarm optimization algorithm. Finally, validation and comparative analyses were carried out using inductive loop data measured from the north-south viaduct in Shanghai. The results indicate that the proposed PSO-PPR model achieves better prediction performance than comparison methods. The average prediction accuracy of proposed method is 37.8% and 27.2% higher than ARIMA model and BPNN model, respectively.

关 键 词：智能交通系统 短时交通流预测 投影寻踪回归模型 粒子群优化 灰色关联度分析 

分 类 号：U491[交通运输工程—交通运输规划与管理]