Short-term traffic flow prediction based on wavelet function and extreme learning machine

Abstract

As the traffic flow has the characteristics of non-linear and strong interference, it has different features in different time-frequency domain. The traditional short-term traffic flow forecasting methods have the disadvantages of lower prediction accuracy, harder parameter determination and poorer adaptability. Aiming at above problems, we proposed a short — term traffic flow forecasting algorithm based on the wavelet function and the Extreme Learning Machine (ELM) to optimize the short — term traffic flow forecasting method. Firstly, the activation function of hidden layer neurons in the prediction model of the ELM is optimized according to the denoising principle of the wavelet function. Secondly, the short-term traffic volume prediction model of the ELM is established, and the traffic volume during the evening peak hours of the Canadian Whitemud Drive highway is forecasted. Finally, the results of this paper are compared with ones that predicted by BP neural network model Compared the R2 value of 0.7 in this method with the one of 0.5331 in BP neural network, the results show that the proposed method in this paper has better generalization ability and more proper stability than BP neural network has. The prediction results are in good agreement with the desired short — term traffic volume, and the short-term traffic flow can be predicted more efficiently.