Research On Information Acquisition Technologies Of Traffic Under Abnormal States

In recent years, it is frequently occurred of abnormal accidents in the world,such as nature disasters, manufacturing events, social public events, horribleincidents, war and so on, which had a serious impact on the road traffic system,causing huge casualties and property losses. Mudslides, landslides and othergeological disasters triggered by heavy rains will be direct damage to transportinfrastructure; a large area of ice after the snow will affect traffic safety and causetraffic congestion, even when avalanche, causing traffic completely disrupted;various large-scale events held in the venue as the center of the traffic network willmake the traffic to a severe overcrowding state. When the abnormal events occurred,whether the transportation system which known as the "life passage" is able to fullycarry out its function will directly affects the incident scope and the emergencyrescue efficiency.As the problems of road traffic information collection, processing and statediscrimination technology in deal with unexpected abnormal accidents nowadays athome and abroad, the aim of this dissertation is to enhance the emergency of roadtraffic information acquisition, take the information acquisition technologies oftraffic under abnormal states as a research object, use mathematical statistics,artificial intelligence, GIS technology and other advanced technology and methodsas means to study the key technologies include traffic information collection,processing and abnormal status discrimination. Designed to explore the rapid andcomprehensive monitoring information collection technology can be applied in thedynamic changes for a large scale of environments, to develop efficient online dataquality evaluation and the rapid data integration technologies for multi-source trafficdata collected, and to accurately discriminate the abnormal status of road traffic andquickly estimate its trend. Therefore, only be able to accurately predict the impactscope and duration of the abnormal status enhance strategy selection andimplementation of the traffic contingency plans under the abnormal states.This dissertation relies on the national "863Program" topic "Research on Information Acquisition Technologies of Traffic under Abnormal States", and themain content of this paper are as follows:(1) Introduction. Firstly,the paper introduces the research status of trafficinformation collection, processing and status discrimination under abnormal states athome and abroad, and this gives the main purpose and significance of trafficabnormal status information acquisition technology under abnormal states, and putsforward the thinking, research framework and chapters arrangement.(2)Road traffic information collection technology under abnormal states. Basedon summing up the normal traffic information acquisition technology, comparativeanalysis of various means of information collection technology, combined withabnormal traffic information demand characteristics, this part studies cyclingtravel-time estimation method based on GPS equipped floating car, smallest samplesize of vehicles determination method, traffic flow average travel time estimationmethod based on median in insufficient sample size, and bottleneck analysis methodof network connectivity based on GPS data. These can provide real-time andaccurate data support to traffic parameters forecasting and traffic status identificationin next step.(3) Large-scale road traffic information processing technology in abnormalstates. First of all, based on the summary and description of the commonly used indata quality evaluation and control methods and the combination of the event inabnormal states, a multi-source traffic information quality evaluation and controlmethod under abnormal states is put forward. Threshold method and traffic flowmechanism method are mainly taken in the discrimination of abnormal data and dataloss in the multi-source traffic information. According to the temporal correlationand spatial correlation, temporal and spatial correlation of the data recovery methodis taken to repair the loss of data and data error, so as to ensure the integrity, validity,accuracy and real-time of the input information of the traffic model. Secondly, themulti-modal dynamic traffic information and adaptive fusion model based on theintegration function and information hierarchies is proposed, which is as the basis forthe development of the traffic information adaptive fusion technology in abnormalstates based on the weighted average method. Finally, the new method aboutshort-time information prediction under the abnormal states of road trafficinformation is designed to get the intrinsic mode function (IMF) by means of theEMD decomposition of traffic parameter data. After the reorganization ofclassification, high-frequency IMF is predicted by gray theory model, intermediate-frequency IMF is predicted by Kalman filter method, andlow-frequency IMF is predicted by adaptive moving average method. Cumulate theresults of above to get the prediction about traffic parameter data in the next timeinterval. Take the multi-step prediction to the real-time traffic parameter ofpredictive data and the historical data under abnormal states, so as to get the finalpredictive data about traffic parameters.(4) The road traffic discrimination technology about abnormal status underabnormal states. It is mainly divided into two parts, which include road abnormaltraffic states discrimination technology and the conditions development estimationmethods of abnormal status in road traffic under abnormal states. In the aspect ofstate discrimination, based on the summary of existing traffic state discriminationtechnology under normal states and the full consideration to the characteristics ofeach abnormal state, use different methods to calculate the road capacity of eachabnormal state against several common abnormal states, in order to provide basicdata for the traffic states discrimination and traffic guidance in the next step. Takecongestion and travel speed as discrimination parameters against sections, takesaturation to discriminate the intersections, and take regional congestion todiscriminate the regional traffic states. Discrimination method of different conditionsunder different abnormal states is put forward, which can reflect the real traffic statesbetter and is more adaptable. In the aspect of conditions development estimationunder abnormal states, based on the existing diffusion estimation in normal trafficcongestion, analyze diffusion process about congested space and time underabnormal states by means of the definition of the meaning of traffic congestion andclassification under abnormal states, and design the diffusion estimation method inabnormal state space of road traffic under abnormal states and estimation method inabnormal state duration of road traffic under abnormal states.(5) Conclusion and prospect. This part makes a comprehensive summary, andpoints out the existing limitations of the study and the paper further research work ofprospected.