Research On Traffic Demand Forecast With Uncertain Trip Generation

Traffic demand forecasting process is the basis of traffic planning, and the resultdirectly as an analysis basis of decision maker. Due to a lot of uncertain factors exist intraffic demand forecasting process, the traditional demand forecasting process tend toignore the uncertainty, given the forecasting result is the single and certain "point"prediction, the author thinks that fails to consider the affection that uncertainty bring tothe prediction results will result in higher decision-making risk. The uncertainty of tripgeneration stage is the source of uncertainty in the four stage forecast model, theuncertainy will be enlarged in the subsequent stages, therefore, this paper mainly studythe uncertainty of trip generation stage as the first forecasting stage, investigatedemand forecasting process under uncertain trip generation.Firstly, the uncertainty of travel demand is qualitatively analyzed, including theuncertain factors and the transitivity of uncertainty in demand forecasting process.Then, through to the comparative analysis of measure method to uncertain problems,the advantage of interval number measuring the travel demand uncertainty isdemonstrated and relevant theory of interval numbers is also studied in this paper.Secondly, considering to measure the uncertainty of travel generation in intervalnumber, from the point of view of uncertainty in forecast model, this paper analysis theuncertainty of travel demand multiple regression forecasting model, based on fuzzymathematics theory, fuzzifies the uncertain information in uncertainty of model formand inaccuracy data, establishs the fuzzy multiple regression model, fit the intervalrange of travel generation under certain level of credibility. Then, this thesis shows theeffectiveness of the model through a case.Finally, to investigate the affection that uncertain travel demand bring to theprediction results, after the measurement of the uncertainty of trip generation in intervalnumber, based on the impact analysis of uncertainty to network state, a shortest pathalgorithm is designed based on min-max regret principle, then combine the the trafficdistribution and traffic assignment model and describe the combination forecast modelunder interval travel demand. According to the operation characteristics of intervalnumbers and the character of the model, on the condition that trip distribution satisfy thesingle constraints, combining interval analysis with traditional MSA algorithm to findthe solution of the model. Then, a simple example is used to test the above models andalgorithms, including shortest path algorithm under interval impedance and method ofinterval traffic combination model. The forecast result is an interval value include uncertain information of travel demand, which can reduce the risk of decision-makingin a degree.