基于GPS实时数据的在线过滤与补遗研究

Study on GPS real-time data on-line filtration and complementation

为提高GPS数据的有效性与可靠性,使之为实现实时交通流量预测与交通诱导服务,首先基于GPS历史数据,以变异系数极小化为优化目标进行GPS数据过滤模型的优选,并对实时数据缺损提出两种补遗算法;然后给出基于动态GPS实时数据的在线数据过滤与补遗一体化算法;最后结合2008年杭州市GPS历史数据对模型进行应用。结果表明,基于简单算术平均的滤波模型即为GPS数据过滤的最佳模型;基于时间序列缺失数据的快速补遗算法能够很好地满足实时预报要求,实现快速补遗。

Global positioning system（ GPS） technique effectively helps to forecast traffic flow accurately in intelligent traffic management system（ ITMS） development. But GPS crude data presented random or system error inevitably during traffic data acquisition by GPS. Thus,it was meaningful to research on how to effectively filter GPS crude data for accurate forecasting of traffic flow,which could further help to realize traffic guidance. Based on GPS historical data,set up GPS data filtering model according to the principle of minimization of coefficient of variance,and two complementation models when data missing. And then combined on-line filtration and complementation to be an integrative program based on dymic GPS real-time data. Finally, provided an application based on Hangzhou GPS data. Results show that the filtering model based on simple arithmetic mean is the best alternative,and the quick complementation model based on time series is well fit for the real-time forecasting.