Research on abnormal monitoring of vehicle traffic network data based on support vector machine

Abstract

In order to solve the problems of low accuracy and long delay in traditional data monitoring methods of vehicle-mounted traffic network, an anomaly monitoring method based on Support Vector Machine (SVM) is proposed. The data of acceleration sensor, gyroscope and magnetic field sensor are collected and filtered. The online analysis method of driving behaviour based on support vector machine is introduced to identify various driving behaviours. By simulating the normal behaviour and abnormal behaviour based on HTTP protocol, the obtained data are analysed to construct the HTTP protocol behaviour. The neural network based on Radial Basis Function (RBF) was trained to monitor the abnormal data in driving behaviours by simulating the behaviour records generated by experiments for many times. The experimental results show that the proposed method can accurately monitor the abnormal data in driving behaviour, and the delay is short, which provides a favourable basis for relevant studies.