Data Fusion Model Based on Support Vector Machine for Traffic Flow Prediction

摘要

It is significant for intelligent traffic system management and control how to get accurate prediction of dynamic traffic flow. In this paper, a traffic flow prediction model of spatio-temporal 2D (2-dimension) data fusing based on SVM (Support Vector Machines) is put forward to. Temporal SVM and spatial SVM are running on two paratactic computer systems,and this will considerably save process time cost. The section flow results predicted by temporal SVM, spatial SVM and spatio -temporal 2D data fusing are all satisfied the precision requirement. However, the prediction precise is significantly improved by spatio-temporal 2D data fusing. Especially,when an abrupt incident happens (e.g., jam,traffic accident), system error of temporal prediction be avoided to a great extent with the spatio-temporal 2D data fusing model.