检测器误差对路网流量推算影响的灵敏度分析

Effect of Sensor Errors on Network Traffic Flow Inference Based on Sensitivity Analysis

摘要:

以路网检测器布设可测流问题中检测器误差对路网流量推算的影响为研究对象，采用基于交叉口转弯比的流量守恒方程和网络检测器布设模型，给出灵敏度分析方法并将其作为检测器误差分析的通用方法.将路段流量推算结果受误差源路段检测器误差的影响定义为关键系数，通过理论推导得出，多个误差源路段流量检测器误差对路网流量推算结果的影响为单个误差源路段流量检测器误差影响的线性叠加.最后以方格式路网为实例，采用在路网的只进和只出路段布设检测器两种方案，利用关键系数给出了检测路段流量误差对未检测路段流量推算的影响.

关键词:检测器布设可测流问题;检测器误差;路网流量推算;灵敏度分析;关键系数

Abstract:

The effects of sensor errors on link flow inference in sensor location flowobservability problem are discussed in this paper. By using the equations of traffic flow conservation system and the model of network sensor location problem based on turning ratios at intersections, a general sensitivity analysis method is proposed to analyze the impact of sensor errors on the link flow reckoning. Firstly, the key factor is defined to represent the errors of link inferring caused by the sensor errors. Then, the total effect of multiple sensor errors on the link flow inferring is proved theoretically to be the linear superposition of the effects of single sensor error. Finally, a square road network is used to test the proposed method. Two sensor location schemes of installing sensors on the only entrance links or exit links of the road network respectively are adopted, which can both support the unique link flow inferring results. The effect of sensor errors on link flow inferring is given by using the key factor method.

Key words:sensor location flowobservability problem; sensor errors; link flow inference; sensitivity analysis; key factor