The Prediction Of Link Travel Time At Urban Roads In The Recognition Of The Generalized Cost

Lately, with the development of economics, the demand of traffic in the city is increasing rapidly. At the same time, all kinds of serious traffic problems are coming. Intelligent Transportation Systems (ITS) brings the advanced information technique, data traffic transmission technique, automation technique and computer technique to the whole transportation management system, and it could give feasible solution to so many problems which the urban traffic faces to, and so the governments attach more importance to the related research.Traffic Route Guidance System, as the basis of ITS, can effectively prevent the occurrence of traffic congestion, reduce the time vehicles stay on the road, and eventually achieve optimization of traffic assignment in the network. The minimization of the generalized cost could be the basis of traffic route guidance. The scholars in this field agreed that travel time cost is the most important one in the costs contained in the generalized cost, so after the establishment of the generalized cost model, this thesis studied long-term estimation and short-term forecast of link travel time in the urban city. At last, the models are respectively validated by the floating car data from Hang Zhou. This thesis is including the following contents:(1) The establishment of the generalized cost model. After the comparison of the previous studies, the generalized cost model in the thesis contained four costs: the travel time cost, the transportation cost, the emissions cost and the traffic noises cost, and also presented the methods for the conversion of the costs in the model.(2) The establishment of models for travel time prediction. By lucubrating the travel time forecast method, the thesis established models for travel time prediction of some links in Hang Zhou, which includes long-term estimation and short-term forecast. In long-term estimation, the thesis chose the historical trend model, and in the short-term forecast, it chose General Regression Neural Network (GRNN) model. At last, the models are respectively validated by the floating car data.