Research On Prediction Method Of Bus Passengers Alighting Stop Based On Deep Learning

Using big data to solve traffic problems has become one of the most important research tasks in the field of intelligent transportation.As an important part of the intelligent Transportation System,the Advanced Public Transportation System(APTS)can not only provide the control and management services of the bus System,but also generate and record a large number of bus passenger travel information data,providing reliable data support for the research on the method of getting off the bus.In space motion model based on human’s boundedness,periodicity and regularity characteristics,with the rapid development of artificial intelligence technology,with the aid of deep learning algorithm,by machine learning to high-dimensional bus passenger travel behavior characteristic value capture,giving individual travel patterns,to put forward the new idea and method of the judgment of the off station.With the rapid development of artificial intelligence technology,based on the characteristics of boundedness,periodicity and regularity of human beings in spatial motion mode,combined with the development of most urban public transport systems in China and the limitations of existing research methods,This paper proposes to obtain the data of the departure station of some bus passengers per day through the travel chain method,and use the deep learning algorithm to capture the high-dimensional eigenvalues of the bus passengers’ travel behavior,and obtain the individual travel rules.On this basis,the getting-off site is predicted,and compared with the judgment result of the travel chain method,the consistency rate is taken as the main evaluation index of the model,and then the research idea of prediction accuracy is determined.In order to ensure the accuracy of the model,this paper starts from the two dimensions of time and space,and visually analyzes the behavior of bus passengers for one week,which lays a good foundation for the extraction of model feature values.On this basis,this paper proposes a comprehensive and rigorous model construction process,and compares the results with the calculation results of the travel chain method.The agreement rate is as high as 82.02%,and the model prediction accuracy is 91.13%.In addition,according to the bus travel frequency and travel time of the bus passengers,this paper uses the model to predict the getting off site in the specified situation-by visualizing the different situations of the travel frequency,the different grid density and corresponding The consistent rate results can fully reflect the regularity of individual passenger travel behavior.