Research on de-noise in prediction of road traffic flow

Noisy data in road traffic flow will ultimately affect the accuracy of their forecasts. For some forecasting model denoising will have some good results, but on the contrary it will reduce forecasting accuracy for some other models. It is very important to introduce effective denoising methods to improve the precision of forecasting in this circumstance. In this paper, the limited denoising based on wavelet transform is proposed. Through the improvements to the determination of the threshold value of non-linear wavelet transform denoising method, limited denoising is achieved. Application of road traffic flow forecasting show that the method is effective.