A combined forecasting method for traffic volume

摘要

In this paper, a combined forecasting method is proposed and applied to traffic volume prediction of 24 hours in advance, called long-term prediction. The combined forecasting model includes two important modules, Kalman filtering module and Markov chains prediction module. Kalman filtering is an optimal estimator which is widely used to eliminate the random errors. This paper mainly uses Kalman filtering to filter the noisy data of traffic volume and reduce the impact of noisy data for a prediction model. Markov chains prediction module can give the forecasting results based on the filtered data. And the forecasting result of traffic volume is a region enclosed by an upper curve and a lower curve. According to the error analysis, the effectiveness of the combined forecasting model is verified. Therefore, the forecasting region can be taken as an important foundation for urban road planning, design and management.