关键词: 交通工程, 短时交通预测, 相空间方法, 极端学习机

Abstract: In order to increase the accuracy of short-time traffic flow prediction,a flow prediction model based on the phase space reconstruction and the regularized extreme learning machine is put forward.In this method，the CC method is used to calculate the best time delay and embedding dimension of traffic flow time series for phase space reconstruction,and the G-P algorithm is used to calculate the correlative dimension of the seriesthat is an important judgment index ofthe chaotic characteristics of traffic flow series,Then,the reconstructed phase point data are taken as the inputs and outputsto trainthe regularized extreme learning machine model,and the main parameters of the model are determined by means of grid searching.Finally,a comparative analysis is carried out based on the actual measured traffic flow data.The results show that the proposed model possesses high performance and is effective in improving the accuracy of short-time traffic flow prediction.

Key words: trafficengineering, short-term traffic prediction, phase space method, extreme learning machine