The traffic congestion cost is continuously increasing and traffic jam is occurring due to increasing traffic demand. To slove this problem, The government introduced ITS(Intelligent Transport System) and has been struggling to mitigate traffic congestion through doing management and operation of traffic information. Traffic congestion could be divided into Recurrent congestion and Non-recurrent congestion, and especially Non-recurrent congestion is called incident such as accident and vegicle breakdown. Traffic congestion caused by incident is difficult to manage because it occurs irregularly. Moreover, the study of incident is limited to Uninterrupted Traffic Flow Facilities such as freeway, but study of incident on the interrupted Traffic Flow Facilities is still inadequate. Therefore, in this study , incident detection model was constructed to aim at urban arterial road that is interrupted traffic flow facility. For this, traffic data and incident data were collected by GPS probe vehicle at Gangnam-Gu in June 2013. The result of collected traffic information analysis after pre-processing, 62.32% of the total link was collected and the collection rate of Gangnam-Daero, Teheran-Ro, Youngdong-Daero and Dosan-Daero was higher than other road. In addition, change of traffic flow of interrupted traffic flow facility was analyzed according to incident occurrence by synthesizing the collected traffic information and the incident data. Analyses result, it was grasped that decrease of speed and the difference from historical pattern data, speed of down-stream, speed of up-stream and number of probe vehicle appeared due to incident occurrence. On the basis of theses characteristics, incident detection model was constructed using by Artificial Neural Network in this study. In the result of the reliability assessment, the detection rate were 46.15% and 38.46%, and false alarm rate were 25.00% and 16.67%. Considering the interrupted traffic flow facility and the use of real data, it is considered that the results were relatively high accuracy.