Construction And Application Of Traffic Trip Prediction Model Based On GSM Data

In the era of big data,traffic travel data showed an exponential growth pattern.The collection of travel data is not limited to traditional fixed traffic flow collectors,and mobile smart devices such as mobile terminals,smart phones and vehicle streaming media can generate real-time massive data.With the increasing data size,real-time high-frequency trajectory data provides rich data resources for dynamic traffic travel prediction.Based on the GSM(Global System for Mobile Communications)traffic travel data,this paper studies the dynamic traffic guidance technology,and uses the large-scale travel trajectory data to improve the existing data processing algorithms,real-time estimation and dynamic prediction of traffic conditions,and then achieve accurate vehicle induction.The main research contents of this paper are as follows:(1)Vehicle guidance at the intersectionThe delay at the intersection of the city is an important factor affecting the traffic efficiency.In this paper,the two factors of the queue length estimation and the speed guidance are introduced in the vehicle guidance at the intersection.Before the popularization of mobile intelligent devices,the traffic volume estimation at intersections can only be detected by fixed traffic flow meters.This paper proposes a real-time queuing length estimation model based on traffic wave theory at the intersection of urban traffic,and introduces the concept of“dissipation delay”.The length of the queue line is used to estimate the "dissipation delay" value of the vehicle,thereby realizing the estimation of the queue length that may be caused by each signal cycle at the intersection.Speed guidance refers to the use of real-time vehicle GSM data and traffic signal timing information to make the vehicle reach the ideal state when it reaches the intersection.The vehicle is designed to take into account the driver’s driving habits and traffic conditions,and achieve the goal of stopping the vehicle through the intersection,thereby reducing the vehicle’s parking delay time.The test results show that the proposed algorithm can accurately predict the vehicle queue length at the intersection of the city and effectively improve the utilization of the green light at the intersection.(2)Short-term traffic flow forecasting Short-term traffic flow prediction is one of the commonly used techniques in urban traffic guidance,and has high requirements for real-time and accuracy.Exponential smoothing method is one of the commonly used algorithms in the dynamic induction of massive data,and the calculation efficiency is high.However,the parameters required in the traditional exponential smoothing calculation process are relatively fixed,and the prediction process is rigid.In view of the shortcomings,this paper proposes a variable time domain dynamic algorithm.By optimizing the coordinate rotation mechanism in the exponential smoothing method,the algorithm automatically updates the smoothing coefficient and the data search range,so that the trend of the data can be tracked during the prediction process,and the accuracy of the prediction can be improved.The experimental data show that the prediction accuracy and algorithm reliability of the proposed algorithm are better than the traditional exponential smoothing method,and the relationship between the smoothing coefficient and the data search range is significantly negatively correlated.Through a comprehensive study on the collected GSM data of real-time vehicle trajectory,the problem of dynamic induction of urban traffic travel is solved,which has certain reference and reference significance for solving the problem of congestion at urban intersections.