面向浮动车取样偏差修正的数据融合方法

Data Fusion Method on Modifying Sampling Bias of Floating Cars

摘要:

考虑到浮动车取样偏差会给路段平均行程时间估计值带来很大误差这一事实，提出了一个融合模型，该模型是在真实路段平均行程时间计算模型的基础上推导出来的，其融合了线圈数据和浮动车数据，减少了浮动车取样偏差对路段平均行程时间估计值的影响.最后利用仿真实验对一条具有连续5个交叉口的主干道进行仿真实验，实验证明，相对于浮动车法，融合方法无论在平峰期和高峰期都可以得到精度更高的平均行程时间估计值.

Abstract:

On the account of that sampling bias of floating cars may bring a lot of errors to the estimated average link travel time, the paper presents a data fusion model derived from average link travel time model based on the population vehicles. By fusing loop detector data and floating car data(FCD), the effect of biased sampling of floating cars on average link travel time decrease with the model. At last, a simulation of an urban arterial road with five intersections and four links shows that data fusion model can yield a better result than the traditional floating car method.