Anti-bunching Public Bus Prediction-control Approach

Giving priority to the development of public transportation is an important measure to solve urban traffic congestion and advocate lowcarbon travel.However,nowadays the bus service suffers from the problems of low punctuality,poor comfort and reliability and etc.Moreover,bus bunching occurs frequently and the quality of bus service needs to be further improved.Under such background,this thesis proposes a prediction-control method to reduce bus bunching.The main work of the thesis consists of the following three parts.(1)Analysis of the mechanism and influencing factors of bus bunching.This thesis deeply analyzes the mechanism of bus bunching,studies the influences of departure frequency,bus speed and passenger’s demand on bus bunching,and gives the method and ideas to calculate the probability of bus bunching.(2)Bus prediction-control method.Firstly,this thesis establishes a segmented prediction-control model.The prediction model takes consideration of the intersection signal lights,bus speed and passenger’s demand and etc.which can predict the bus headway probability of the next station.The control model uses the idea of double-layer decision-making to determine the optimal control strategy and parameters to realize the realtime control of the bus.Secondly,based on the segmented prediction model,this thesis establishes a multi-stage prediction control model,uses genetic algorithm to realize the prediction and optimal control of multiple links and stations.Thirdly,the effectiveness of the two types of prediction-control models is verified by simulation experiments.The results show that both types of models can effectively improve the stability of bus headways,reduce bus bunching,and deal with the execution error of bus drivers,showing good stability.(3)Case analysis.This thesis selects the statistics of the No.314 bus route in Changsha city for analysis and evaluates the operation status of this bus route.The accuracy of the prediction model is verified with actual data,and the artificial neural network model is selected as the comparison model.The results show that the prediction error of the prediction model is slightly smaller than that of the BP neural network model,and the prediction accuracy of the bus states can be maintained at about 90%.The current theoretical research in this thesis is to prevent bus bunching from the perspective of bus operation.In future research,it is necessary to consider the optimization at the planning level,such as reasonable route and station layout and setting of departure frequency.