Study On Forecasting Of Short-term Traffic Flow Based On Fractal Theory

The current methods of short-term forecasting for traffic flow are simple and easy to calculate. But these approaches are lack of the functions to describe the nonlinear characteristic of traffic flow, especially to overcome the interference by random factors.Time and space character of traffic flow was analyzed based on fractal theory. Through the technology of space reconstruction, the key parameter entropy value K was determined. Then the fractal dimension of traffic flow in data feature was calculated. Further, short term traffic flow was forecasted using these parameters. The accuracy of traffic flow prediction of 2 min was more than 92% in the instance. On the basis of above work, the forecasting scale based on fraction was discussed. It was indicated that the range of effective prediction scale can be expanded from 5 min to 2 min. The result of research is shown that the approach based on fraction will be ineffective within 2 min.A useful approach for short-term traffic flow forecasting was given. Some guides for prediction scale based on fraction were provided in this paper.