Forecast of Traffic Flow State During Large Competition Events

Abstract

The forecast of traffic flow state (TFS) is the key to the effective control of roads, especially during large competition events. To enhance the accuracy of TFS forecast, this paper puts forward a way to predict the TFS on the peripheral roads of large competition events, which is based on support vector regression (SVR) and parameter optimization. Firstly, the tensor recovery algorithm was adopted to fill up the missing data. Then, the simulated annealing (SA) algorithm was applied to optimize the SVR parameters like the penalty factor, and the insensitive loss. Next, a TFS forecast model for the road section near large venues was established based on the SVR and the optimized parameters. The example analysis shows that the parameter optimization improved the accuracy of the SVR forecast model, making the predicted results closer to the actual data. The proposed model greatly facilitates the management and control of road traffic during large competition events.