The Study Of Buses Road Running Time Estimation And Forecasting Methods Based On SVM And Dynamic Hybrid Algorithm

With the rapid development of China’s economy, the rapid increase in urban population,the ownership of motor vehicle,especially the ownership of private cars have increased dramatically, the issues of urban transport are also increasingly prominent consequently. Traffic problems have become one of the factors that seriously hinder urbans’development.Priority to the development of urban public transport and encourage people to take the bus is one of the most important ways to solve the urban traffic problems. With the advent of intelligent transportation systems,how to achieve the information and intelligence of the public transport system is particularly important. Intelligent transportation system is a key technology of bus arrival time prediction, prediction of bus arrival time,is important research direction to improve the operational efficiency of urban public transport systems and the development of intelligent public transport system.Firstly, existing travel time prediction research status is analysised and compared at home and abroad, pointed out the advantages and disadvantages of each model separately. However, with the increase of data acquisition technology matures and the prediction accuracy for time, the above model is still insufficient in time prediction method and accuracy. In this paper, the data collection method based positioning system GPS vehicle, The original data collected on the bus is pretreated and analysised, to get bus history travel time. Based on approach to combined with bus real-time operating data and historical data, fully consider the many random factors, proposed a hybrid bus running time prediction model based on the combination of support vector machine (SVM) and real-time prediction. In this model, SVM based on historical data,selected time period input characteristics, holidays, weather, road length and speed five variables, predicted reference time of each road vehicle uptime. Because of the support vector machine prediction model entirely based on bus running historical data, Due to the presence of the characteristics of the bus running which is real-time and by the influence of dynamic random status, In order to reflect the characteristics of the current moment buses running, presents a dynamic hybrid bus arrival time prediction model, This model combines support vector machine static output reference time to predict buses running real-time of travel time to the next site. The model includes semi station travel time prediction model based on the real-time speed and latitude and longitude buses and the entire bus station site travel time prediction model based on dynamic allocation weight.Finally, The model is verified by experiments using Jinan one bus route. he results show, The hybrid bus running time prediction model based on the combination of support vector machine (SVM) and real-time prediction has significantly improve the prediction accuracy compared to single SVM forecasting model, nd forecast significantly enhanced stability, The model has better robustness.