Short Term Traffic Volume Prediction Of Arterial Road Intersection Using Time Varying Filtering Based Empirical Mode Decomposition

With the rapid development of Chinese economy and the continuous improvement of overall national strength,the number of cars have continued to increase.Thus,the demand for transportation cannot meet the supply,which has led to serious problems of traffic congestion,traffic environmental pollution,and traffic safety.To alleviate traffic problems,developing intelligent transportation systems(ITS),which can explore the potential function of transportation infrastructure,is seen as an effective measure.In practice,the scientific,accurate and reliable short-time traffic volume prediction is one of the key technologies of ITS which can improve the traffic efficiency at intersections and provide the best route and traffic mode for travelers.Therefore,devoloping a short-term traffic volume prediction model with excellent performance by exploring the factors that affect the accuracy and reliability of short-term traffic volume prediction will help the construction of ITS.Simultaneously,these measures can improve the level of urban traffic management and the problem of urban traffic congestion.Firstly,the basic theory,the characteristics,and the advantages and disadvantages of several forecasting methods of short-term traffic volume are expounded respectively in this paper.Secondly,the preprocessing and decomposition methods of traffic volume original data are introduced.In addition,the problems of these methods are stated.According to the analysis,the time varying filtering based empirical mode decomposition(TVF-EMD)is adopted to solve the problems of decomposition.Thirdly,the TVF-EMD-LSSVM hybrid model that combines the TVF-EMD and the least square support vector machine(LSSVM)model is proposed.At the same time,the generalized autoregressive heteroscedasticity(GARCH)model is established to estimate the confidence interval and perform reliability analysis of the proposed model.Furthermore,the indexes of average absolute error,average relative percentage error,root mean square error,root mean square relative error,equality coefficient,average confidence interval width and kickoff percentage are selected to evaluate the reliability and accuracy of the model.Finally,case studies based on two groups of data measured from an arterial road entrances A and B are employed to evaluate the performance of the proposed method.To illustrate the performance of the proposed method,three additional forecasting models including ARIMA model,LSSVM model and EMD-LSSVM model are used to perform the performance comparison.With the help of the two sets of data,the experimental results indicate the proposed model outperforms the other involved models.Based on the TVF-EMD data decomposition method and the LSSVM forecasting model,a hybrid forecasting model of short-time traffic volume is constructed to study the short-time traffic volume prediction at arterial roads’ entrances.To some extent,the proposed method improves the prediction accuracy and reliability.Meanwhile,it complements the prediction model theory.