The Research Of Regional Bus Scheduling Optimization Based On Short-term Passenger Flow Forecast

As the contradiction between supply and demand of traffic increasing, the deepening of bus priority strategy implementation is imminent. Advanced technology of bus dispatching has been the important way of enhancing core competitiveness of public transit. At present, bus scheduling in many bus companies is manual in our country, which severely limits the potential application of technologies based on the Advanced Public Transportation system(APTS). Moreover, this way of scheduling will lead to so many problems that the bus service level will be affected badly, for example, passengers’ waiting time is long in bus station, the load factor of bus is different and so on. Compared with single operation mode, bus regional scheduling can increase operation efficiency by 8% ~ 20%. From the perspective of regional scheduling, this paper studied the optimization method of bus scheduling. Its aim is to promote the transformation of scheduling from manual work to intelligence. Accordingly, the quality of public transport services can be improved.Based on intelligent public traffic system, firstly this paper conducted data fusion and mining of the IC data, GPS data, bus station information. And it proposed methods of judging the bus station which passengers using IC card get on, get off and transfer to another line at. Then combined with the passengers flow data of one bus line in the city of harbin, the passengers flow’s space-time distribution characteristics were analyzed.On the analysis of the characteristics of short-term passenger flow, this paper built a model based on RBF neural network model to forecast short-term passengers flow and applicated in an example to evaluate model’s prediction accuracy. The result shows that the error of prediction model is small. What’s more, comparing the influence of each input variable on the result of prediction, it shows that the dependence of the prediction model on input variable type is less.Then the paper designed a bus scheduling strategy based on the transfer time window, this strategy can take full advantage of intelligent transport system and make up for the deficiencies of the current scheduling mode in science and technology application. Finally, based on the Setting of transfer time window, a optimization model on the regional dynamic scheduling was built, the model’s optimization goal is to minimize the waiting time of all the passengers in the sdudy area with the departure time and transfer time window of each bus as decision variables and it was solved by genetic algorithm. At the end of the paper, a computational example was designed to evaluatie model’s Performance.