Research On Short-term Traffic Flow Forecast And Shared Ride Recommendation Of Urban Taxi

In the face of the current urban road congestion,the existing taxi operating efficiency is low and the cost is high.Traditional traffic control methods have been unable to meet people's travel needs.However,with the continuous development and innovation of science and technology,artificial intelligence technology has become mature and has been widely used in our daily life,especially in the intelligent transportation system.Therefore,in order to solve the problems of urban road congestion and low taxi driving efficiency,and then to effectively manage and plan roads,this paper uses intelligent transportation technology to study the short-term traffic flow prediction and ride-sharing route recommendation of urban taxis.The main work done in this article is as follows:(1)According to the time series characteristics of historical traffic trajectory data,traffic flow information is predicted and analyzed.In order to solve the problems of periodicity,stationarity and outliers of time series,improve the prediction effect of traffic flow,and realize efficient traffic induction and traffic control,this paper proposes a combined prediction method based on deep learning on the basis of short-term traffic flow prediction.According to the characteristics of time series data,a multi-step prediction model combining LSTM(Long short-term Memory)and XGBoost(Extreme Gradient Boosting)algorithm was constructed to predict the driving speed of taxis on this road in the near future.The combined model not only enhanced the generalization ability of the prediction model,but also improved the prediction accuracy of traffic flow.(2)Based on the location information of the starting point of historical taxi data,the passenger ride-sharing route is analyzed.In order to solve the problems of low taxi loading rate,unreasonable driving route planning and balanced driver and passenger costs,and improve the operating efficiency of taxis,this paper proposes an evolutionary algorithm based on machine learning on the basis of shared ride planning.In view of the characteristics of GPS data,a multi-objective and multi-constraint ride-sharing model based on improved genetic algorithm is designed to realize the planning of ride-sharing routes under different directions,different starting points and ending points of passengers.The operating efficiency of the taxi also protects the interests of both drivers and passengers.Finally,this paper conducted a simulation analysis of the above two methods,and the results showed that the short-term prediction model and ride-sharing route recommendation model established in this paper can not only improve the prediction accuracy,but also effectively solve the taxi driving efficiency and passenger's issues such as travel costs are of great significance to intelligent traffic guidance systems.