Algorithm Research for Freeway Incident Detection Based on SVM

摘要

Consider that floating cars and fixed coil detectors have complementary advantages and the merit of Support Vector Machines technology, this paper uses Support Vector Machines technology at first to fuse two kinds of data, and make freeway incidents detecting. Secondly, combining with the characteristics of support vector, the freeway incident detection algorithm based on reduced training set and support vector machines is designed to use fuzzy cut set theory. The method reduces not only training samples but also training time and improves the practical application ability of incident detection algorithm based on maintaining the detection rate and false alarming rate.