Research On Analysis Method Of Traffic State Reliability Of Road Network Based On Real-time Data

In the recent years, due to the short supply of roads, the traffic running statechanged from smooth to crowed and finally to congestion step by step. It badlydamaged the traffic running stability and reliability, lead to serious waste of traveltime, the decreasing of environment quality and the increasing of vehicle loss.Meanwhile, because of the continuously improvement of social life level, there areincrease both in the ways and frequency of residents travelling activities. In order tocomplete the travel activities as more as possible in the limit time, residents graduallyimprove the requirements of traffic running efficiency and reliability. Consequently,research on the traffic state reliability based on real-time traffic data is not onlyindispensable parts of intelligent traffic system, but also an important topic in today’stransportation reliability research.The evaluation, prediction, and allocation of traffic state reliability for roadnetwork are the three steps of transportation reliability research. Purpose of the firststep is to evaluate the actual situation of traffic state reliability in the current roadnetwork, OD pairs of sub-road network, path and road/intersection. Purpose of thesecond step is to predict future traffic state reliability of the road network and maketraffic management strategies and travel plans in advance. The final step is todetermine how to allocate the reliability index for road/intersection composed afterdetermining the target value of traffic state reliability for road network can be themost economic and efficient to achieve the target. Nowadays, most achievements ofexisting research are concerned evaluation of traffic state reliability, while few resultsin number of prediction and allocation research, little of special research on it. Largespace is left to continuously improve.This paper deepens and expands reliability evaluating methods in long-term andcurrent time for road network based on real-time dynamic traffic data acquired.Meanwhile the paper makes further research on prediction method and allocationmethod of traffic state reliability for road network, provides technical support forfurther improvement on running stability of the whole traffic flow. The innovativepoints and research progress mainly include the following four aspects.(1) In the field of traffic state long-term reliability evaluation, based on the current research progress, this paper has extended and supplement evaluation index oftraffic state reliability in the long-term. Combined with dividing layers for roadnetwork and traffic state discrimination, the paper has discussed failure mode oftraffic state and constructed online analysis algorithm based on fault tree analysismethod for traffic state long-term reliability. The algorithm applied the results oftraffic state fault tree analysis by means of setting minimum cut and path, realized thequalitative analysis and quantitative evaluation of traffic state long-term reliability ofroad network. At the same time, after comprehensive consideration of three indicators:structure importance, probability importance and key importance for road/intersection,the paper put forward ways to find out key road/intersection which greatly influentstraffic state long-term reliability. By defining diagnostic importance of minimum cutsets and sections/intersection, the paper designed a fault diagnosis decision system fortraffic state of road network. The research results further expanded evaluationmethods of traffic state long-term reliability, and successfully designed effective faultdiagnosis decision system for traffic state of road network.(2) In terms of traffic state current reliability evaluation, in order to overcomeshortcomings of the duality of traffic volume, the paper has redefined the traffic statecurrent reliability index and designed reliability index calculation method of thecurrent traffic state for road/intersection, path, OD pairs of sub-road network based onreal-time traffic data. Considering the physical topology characteristics, traffic volumebearing characteristic, traffic state current reliability characteristic and influencecharacteristic four aspects, this paper has constructed evaluation index system of keyroad/intersection which greatly influenced traffic state current reliability, and putforward a method of determining key road/intersection online based on the multipleattribute decision theory. The empirical analysis shows that reliability index of currenttraffic state by this paper is able to accurately reflect the dynamic characteristic of thetraffic state reliability. Meanwhile, it has proved the rationality of this method indetermining the critical sections/intersection based on multiple attribute decisiontheory. The research achievements further improve traffic state current reliabilityonline analysis method and provide decision-making evidence for dynamic trafficstate reliability management and control.(3) In terms of traffic state reliability prediction, in order to predict the futureroad network traffic state reliability of multiple times, this paper has discussedmulti-step prediction principle of traffic state reliability after referring to results ofreliability prediction in other areas. Against the phenomenon of increasing error caused by multi-step prediction, the paper put forward predictive estimation of trafficstate reliability and constructed double layer predicting model of traffic statereliability on basis of wavelet neural network optimized by particle swarm for trafficstate long-term reliability and current reliability index. The upper layer of the model isway to estimate predictability of traffic state reliability, lower is multi-step predictionmethod of traffic state reliability. The research results perform well in the predictionof traffic state reliability.(4) In the aspect of traffic state reliability allocation, In view of current trafficstate reliability management and improvement scheme for road network lack ofquantitative basis, according to particularity of traffic state reliability, the paper putforward thought of traffic state reliability allocation, construct the factors which affectthe system and the quantitative method, and come up with the method to determinethe weighing value of traffic state reliability allocation based on the analytic hierarchyprocess. By applying this, respectively according to the traffic state long-termreliability and current reliability two kinds of traffic state reliability index, the paperdesigned allocation method of traffic state reliability in three levels: network, the ODpairs of sub-road network and path. To the problems of allocation result overflow andoptimization, the paper respectively designed the reallocate method for traffic statereliability allocation and optimization method for the distribution results ofroad/intersection. The research results contribute to allocate traffic managementresources effectively, and provide decision-making basis for the traffic planning, theroad design and the traffic management and control.The research contents, research methods and conclusion involved in this paperare the supplement and beneficial exploration of the analytic method study of trafficstate reliability under real-time data. The research results obtained is of importantacademic significance for deepening the analysis method of the road network trafficstate reliability. Meanwhile, it has obviously practical value in improving roadnetwork traffic state reliability.