考虑先验知识的ICA在轨道交通客流预测中应用  

Application of ICA in Prediction of Rail Traffic Passenger Flow Volume Considering Prior Knowledge

作　　者：仇建华[1] 杨兴园 张亚岐[2] 李创 QIU Jian-hua;YANG Xing-yuan;ZHANG Ya-qi;LI Chuang(Xi'an Aeronautical University,Xi'an,Shaanxi 710077,China;Dongfeng Motor Corporation Technical Center,Wuhan,Hubei 430058,China;Dongfeng Honda Automobile Limited Liability Company,Wuhan,Hubei 430056,China)

机构地区：[1]西安航空学院,陕西西安710077 [2]东风汽车公司技术中心,湖北武汉430058 [3]东风本田汽车有限公司,湖北武汉430056

出　　处：《计算技术与自动化》2019年第2期15-18,共4页Computing Technology and Automation

摘　　要：针对轨道交通客流量的不确定性,采用考虑先验知识的独立成分分析法对轨道交通客流量进行预测。首先将采集部分时段同一线路上各个站点客流量数据作为先验样本,获得相应的概率密度;其次通过独立成分分析得各个站点客流量独立源数据,利用独立源数据构建基于相关向量机预测模型;最后通过增加扰动量来对模型进行评价。结果表明:未考虑先验知识的预测模型对相对较小的扰动量不敏感,且对较大扰动量响应的稳定性差;而考虑先验知识的预测模型有很好的敏感度和稳定性。In view of the uncertainty of rail traffic flow ,this paper adopted the independent component analysis considering prior knowledge to predict passenger flow volume of rail traffic. Firstly ,the passenger flow datas of all stations in the same line were collected and the datas were as the prior samples. Then the corresponding probability density was obtained. Secondly,the independent source datas of all stations can be achieved through independent component analysis.And the independent source datas were used to construct prediction model based on correlation vector machine. Finally,the model was evaluated by increasing the disturbance quantity and results show that the prediction model considering the prior knowledge is not sensitive to relatively small disturbance quantity,and its stability of the response of large disturbance quantity is poor. The prediction model considering the prior knowledge has good sensitivity and stability.

关 键 词：轨道交通 先验知识 预测 独立成分分析 扰动量 

分 类 号：U121[交通运输工程]