Study On Freeway Automatic Incident Detection (AID) Based On Mobile Phone Positioning

Along with the continuous development of economy and the ceaseless increase of traffic demand, the traffic safety problem becomes an urgent affair. It activates the exigent study and development of the advanced freeway Automatic Incident Detection (AID) system, which is considered as a breakthrough point to elevate the work efficiency of transportation management and to enhance the freeway management and serve level. Moreover, how to study and establish the freeway AID system is one of the core research fields in Intelligent Transportation Systems (ITS), it is urgently necessary to reduce the congestion and damage resulted from the freeway incident, ensure the vehicle running safety, and avoid the second accident. Based on the introduction of traditional freeway AID system, traffic data collection technologies and the deeply analysis of existing AID algorithms, mobile phone positioning technology in mobile communication network is utilized for freeway AID, and so the architecture of the freeway AID system based on mobile phone positioning is brought forward and analyzed in detail, including the modules'function, the characteristic of system and the key technologies of system's actualization. Mobile phone positioning and map matching is the basic and premise of freeway AID system, and the AID algorithm is the key factor to implement it. In term of the AID algorithm based on mobile phone positioning, a single mobile phone status based fuzzy comprehensive evaluation model is proposed firstly, which uses several measures, including the stand deviation of speed, coefficient of variation of speed etc., are calculated from each mobile phone'locations to identify the traffic incident. Aim at the characteristic of the algorithm, a mixed type fuzzy comprehensive evaluation model is presented. It combines all the mobile phones on the road section, according to the characteristic of basal fuzzy comprehensive evaluation models, to avoid the effect of single mobile phone's disturbance on the freeway AID. The simulation experiments are implemented to analyze the incident detection performance, and to study the influence of the mobile phone's positioning interval and parameters setting on the performance of algorithms mentioned above. The results show that two models have high detection rate, but with respect to the false alarm rate, the mixed type fuzzy comprehensive evaluation is better than the single mobile phone status based fuzzy comprehensive evaluation. At last, two proposed models are compared with the traditional California algorithm by the simulation experiments, which show that, using these projects, the freeway AID method based on mobile phone positioning is lower cost than that based on traditional traffic data collection technologies, and it can work synchronously to realize the all-weather and dynamic incident detection. So it is available to ensure the real time and veracity of freeway AID.