Study on Automated Incident Detection Algorithms for Freeways Based on SVM

Aimed at the problem that many AID algorithms have lower detection rate and higher false alarming rate, this paper proposed a kind of AID algorithms for freeways based on SVM. The eigenvector reflecting traffic state was designed according to selected traffic measures that can be provided by many kinds of traffic sensors. AID algorithms were designed based on different sorts of SVM models and tested and compared with simulated data. The results showed that the performances of proposed methods are better than selected classic AID algorithms