Research On Repair Methods Of Urban Expressway Traffic Flow Fault Data

ABSTRACT:Traffic flow data is the foundation of traffic state identification, traffic management and control and other research and work carried out in transportation field. With the development of traffic monitoring systems, massive traffic flow data continue to emerge. However, due to the fault detectors, the transport network failure and environmental factors, there will inevitably be a variety of quality issues (incomplete, error, noise, etc.) in the traffic flow data collected. In order to reflect the traffic running truly, we should identify and repair traffic flow fault data (including missing data and abnormal data) efficiently, which is the data support and base guarantee for the research’ successfully carry out.In this paper, we research the repair methods of urban expressway traffic flow fault data on the basic of identifying and analyzing the fault data from the traffic flow data collected by detectors effectively. First, the data’s analysis of spatial and temporal characteristics is used to determine the characteristic parameters in time and space for fault data repair. The next is the fault data analysis. We propose fault data identify analysis methods based on smoothed estimate threshold and fault data repair analysis based on statistical correlation analysis. At then, we propose a two-stage adaptive weight fault data repair combination model under the idea of "decomposition-combination". The local repair model is data repair model based on least squares support vector machine model.And traffic state probability distribution estimation model based on maximum entropy is the adaptive weight model.At last, we test the model combining the actual traffic data collected by microwave detectors in Beijing. The experimental results show that this method can reduce the interference of random factors in traffic system and can repair the data accurately.