Wavelet Based Highway Incident Detection Model with Three Degree Alarm

Expressway incidents are non-recurrent and random events that disrupt the normal traffic and create bottleneck. As high as 60%∼70% of the highway traffic delay is attributes to traffic accidents. A substantial reduction in delay can be achieved by quick and reliable incident detection. Because of high false alarm rates of the developed Automatic Incident Detection (AD), they are almost abandoned. Because the traffic flow is break when an incident happen, and wavelet transform is suitable to exam break point from a signal, this paper present a method to detect incidents based on wavelet transform with three degree alarming. The simulation results showed that this method is better than California algorithm in DR, FAR and MTTD.