Research On Travel Time Prediction Of Multi-route Bus Based On MDARNN

Providing a good travel experience is the best way to guide the public to choose public transportation,and accurate and effective bus travel time prediction is the key to improve the travel experience.Bus travel time prediction needs to consider not only the driving status of the riding vehicle,but also the interchange waiting time.In this thesis,a hybrid prediction model PCF-MDARNN-LSTM is proposed.The main work of this thesis is as follows.First of all,two impact factors,which are macro impact factors and local impact factors,are defined.The statistical characteristics of various influencing factors are analyzed.Then,the raw data set is preprocessed to filter invalid and redundant information,and build travel chain,which make up the flaw of raw data set.Then,according to the travel time characteristics,the multi-route bus travel time is divided into 4 parts: travel time of the current driving segment(CDS),bus stop dwelling time module,stop-stop travel time and transfer point waiting time.The prediction networks are selected according to the characteristics of each time module,Classic LSTM model is used to predict the CDS traveling time and transferring point waiting time.An improved DA-RNN model,the MDARNN model,is proposed to predict the bus stop dwelling time module,station-station traveling time.At last,a hybrid prediction model PCF-MDARNN-LSTM is constructed by using a parallel processing structure.Moreover,4 real-time traffic flow models are used to calibrate the prediction time of the 4 sub-models.Next,the prediction time of each sub-models is combined to realize the prediction of total bus travel time.Simulation and comparative experiments are done based on the Tensor Flow1.14 platform.Experiment results prove the effectiveness of the proposed approach and 4 sub-models.Finally,the bus travel time prediction system is developed based on the theory and experiment of this paper.