Hybrid GA based online support vector machine model for short-term traffic flow forecasting

Abstract

In this paper, a hybrid genetic algorithm (GA) based online support vector machine (OSVM) prediction model for short-term traffic flow forecasting is proposed, according to the data collected sequentially by the probe vehicle or the loop detectors, which can update the forecasting function in real time via online learning way, and the parameters used in the OSVM were optimized by GA. As a result, it is fitter for the real engineering application. The GA based OSVM model was tested by using the 1-880 database, the result shows that this model is superior to the back-propagation neural network (BPNN) model. © Springer-Verlag Berlin Heidelberg 2007.