The Research On How To Make Optimization For Public Transport Allocation

Bus bunching and long headway problems have a serious effect on the servicereliability of the transit system and they are the most concerned problems for travelersand transit agencies. Therefore, there is great significance to study these problems.This thesis firstly gives a definition for this problem, analyzes the specific hazardsand proposes a method based on arriving time to alleviate this problem. After that,this thesis builds the analysis model for the bus bunching and long headway problemfrom both macro and micro levels. Further more, this thesis proposes a dynamicfeedback control model based on the GPS data to solve the bus bunching and longheadway problems.Based on this ideal, this thesis designs the algorithm of data collection andinterpolation and method to calculate the actual distance of two points according totheir latitude and longitude coordinates in order to get the data needed for theprediction. Then, BP-Artificial Neural Network (BP-ANN) model is built and sampledata are used to train the non-linear relationships between the input and outputvariances so that it can be used to predict the distance between buses based on thereal-time GPS data.Finally, this thesis gives the case study of the680group data from line16ofBeijing Bus Company. The result of the cluster analysis shows that the distancebetween two buses is mostly related to their speeds. The prediction results of theBP-ANN show that the prediction model has good performances. According to thepredicted results，speed feedback control becomes available and the bus bunching andlong headway are effectively prevented.