A Novel Abnormal Traffic Incident Detection Method Based on Improved Support Vector Machine

Yan Sun ； Zhixiang Hou

《淡江理工學刊》 21卷1期 (2018/03) Pp. 45-50

https://doi.org/10.6180/jase.201803_21(1).0006

摘要

In recent years, the traffic incident automatic detection technology has become a central issue in intelligent transportation field. In order to detect traffic incidents accurately, highway traffic incident detection model is set up based on the characteristics of highway traffic flow and the basic principles of traffic incident detection. This model includes data preprocessing module, construction of SVM and decision output module. Improved particle swarm optimization is adopted to optimize parameters of SVM model. By adjusting the penalty parameter and the size of the radial basis kernel parameter, we can make SVM has better classification performance. The simulation data is extracted from I-880 traffic data set. By comparing and analyzing the simulation results, the SVM model based on improved particle swarm optimization achieves better comprehensive detection performance with less modeling time and higher detection accuracy, which provides important reference for highway abnormal traffic incident detection.

關鍵字

Traffic Incident Detection ； Particle Swarm Optimization ； Support Vector Machine