Study on Automated Incident Detection Algorithms Based on PCA and SVM for Freeway

According to the freeway traffic flow characteristics under incidents situation, this paper puts forward a kind of AID algorithm based on PCA & SVM. Above all, PCA has been applied to extract principle components from preselected indexes, to realize dimension reductions, and construct the traffic eigenvectors. And then, validity of the new algorithm proposed in this paper has been tested using various SVM models, and also contrasts with the results of algorithms only based on SVM. The simulation results demonstrate that the new algorithm can detect incidents more effectively and more expeditiously, with the preferable detection performance. Especially, detection effects of the new algorithm based on linear SVM models are the best.