摘要
为了提高高教园区出行生成预测数据的可靠性,提出了一种基于模型优选及融合的出行生成预测方法。根据选取的模型评价指标,借助ELECTRE方法对区域出行产生模型进行优选,利用证据理论方法,对优选模型的权重进行标定,最后借助标定后的权重对区域交通生成量进行数据融合预测,并与实际结果进行了对比。分析结果表明:多模型融合预测相对误差在10%以下,预测结果与实际情况非常接近,该预测方法有效。
To improve the forecasting data reliability of trip generation for higher education zone, a forecasting method of trip generation was presented based on model selection and model integration. According to the model's evaluation indices, the trip generation models of zone were optimally selected by using ELECTRE method. The weights of optimal selected models were demarcated by using D-S evidence theory. Based on the weights, the trip generation of higher education zone was forecasted by using data fusion, and the forecasting result was contrasted with actual result. Analysis result indicates that the relative error of multi-model integration forecast is below 10%, and the forecast data is very close to actual situation, so the method is effective. 5 tabs, 2 figs, 11 refs.
出处
《交通运输工程学报》
EI
CSCD
北大核心
2010年第1期66-71,共6页
Journal of Traffic and Transportation Engineering
基金
广东省科技计划项目(2007B080701001)
广东省自然科学基金项目(9151022501000015)
关键词
交通预测
模型优选
D-S证据理论
模型融合
出行生成预测
traffic forecast
model selection
D-S evidence theory
model integration
trip generation forecast