Research On Bus Arrival Time Prediction Based On Wavelet Neural Network

Against the background of continuous urbanization,the rates of traffic congestion,traffic pollution,and traffic accidents have increased significantly.To this end,it is important to improve the quality of urban public transport services,so that more people can choose a cleaner and greener public transport travel becomes an inevitable trend.Advocating public transport travel and changing residents' travel habits are effective ways to solve traffic problems.In the process of developing public transportation,making accurate predictions on the arrival time of bus vehicles can enable citizens to have a higher degree of trust in public transportation,which has outstanding value and significance for promoting the development of urban transportation.In the process of viewpoint analysis,this paper discusses the running characteristics and status of public transport vehicles,builds a bus arrival time prediction model,and reasonably predicts bus arrival time.In thesis research,the core content includes:Firstly,based on the analysis of the operating characteristics of the bus,the factors affecting the bus arrival time are determined.First analyzed the characteristics of bus operation in terms of time and space,and then started from the actual operation process of the bus,divided the vehicle operation process into the inter-station driving process and the stop process,and analyzed in detail the factors that may affect the station time in the two processes.Select the four major factors of "average road operating time of w trips on the same day at the same time period","operating time of the last shift and work on the same day" and "driving time of the previous segment of the target prediction station" as the bus arrival time in this article Predicted influencing factors.In addition,the original bus operation data is obtained through the bus dispatching system,and a large amount of sample data is obtained through formatting and standardization.Secondly,starting from the actual life scene,considering the characteristics of data samples and the requirements of real-time,the overall design of the prediction model was carried out.The BAS-WNN prediction model of the inter-station travel time and the frequency weighted prediction model of the stop time of the station are established respectively.For each prediction model,detailed design ideas and implementation steps are given.Finally,the simulation model of the inter-station travel time prediction model and the station stop time prediction model were simulated.The K1 bus line in Ganzhou City is selected to simulate and analyze the travel time between stations.The MAE value is used to longitudinally measure the accuracy of the BAS-WNN model in different time periods.The absolute value of the prediction error is basically within 60 seconds.In order to further verify the accuracy and stability of the BAS-WNN prediction model,the prediction results are compared horizontally with the prediction results of WNN and BPNN,which finally proves that the BAS-WNN prediction model has the best effect.When verifying the stop time prediction model of the station,the historical data of three routes with different characteristics of K1 Road,K2 Road and 117 Road in Ganzhou City were selected.The frequency weighting method was used to predict the stop time.Fast calculation speed.Finally,the BAS-WNN model is further simulated and verified through examples,and the error can be controlled within a small range without unexpected situations,and the expected effect can be achieved.In addition,a brief introduction was made to the method and flow of the forecasted arrival information release.