| The rapid development of freeway scale in China has facilitated people’s daily travel,but at the same time,it has also caused frequent traffic accidents,which poses a serious threat to people’s travel safety.Therefore,freeway traffic safety has been the focus and difficulty of transportation management over the years.Freeway proactive safety management has become the trend of future research and application benefits from its advantages in effectively preventing accidents and reducing the loss of life and property caused by accidents.It is one of the basic work of freeway proactive safety management and also an important basis for formulating emergency plans to analyze the spatiotemporal propagation impact of freeway traffic accidents.Therefore,this paper is of great significance to the safety management and evaluation of freeways.The core of freeway proactive safety management is to determine spatiotemporal propagation impact of accidents timely and effectively.A fundamental issue in determining the spatiotemporal propagation impact of an accident is to ensure that the shape of the affected area in the speed map is consistent with the propagation of shockwaves.However,the existing studies based on clustering method cannot meet the requirement of consistency.At the same time,many studies have overlooked the recurrent congestion that may be contained in the area affected by the accident,leading to an overestimation of the severity of the accident.Based on this,the main contents in this paper are as follows:(1)Determination of the spatiotemporal propagation impact range of freeway traffic accidentsConsidering the propagation of shockwaves when an accident occurs,a data-driven approach based on fuzzy clustering algorithm is proposed to determine the spatiotemporal propagation impact range of an accident in this paper,which can satisty consistency with the propagation of shockwaves.The accident traffic flow data and historical data are used to establish the speed map.On this basis,the clustering algorithm is used,and the shockwave propagation law is incorporated into the clustering algorithm by considering the neighborhood constraints,so that the shape of the affected area obtained by clustering can comply with the shockwave propagation rules.For this purpose,a filtering algorithm with filtering strategy in consistency with the shockwave propagation law is designed.The analysis proves that the proposed filter can eliminate the noise data that violate the shockwave propagation rules and suppress the damage to the boundary non-noise data.Therefore,this algorithm is superior to other clustering method in the identification of accident affected area.(2)Separation method of freeway recurrent congestion on the traffic accident propagationConsidering that the impact of recurrent traffic congestion can be included in the impact area of the accident,a K-Nearest Neighbor(KNN)pattern matching algorithm is proposed in this paper to match the traffic conditions most similar to the traffic pattern of the accident from the historical database to separate recurrent congestion and nonrecurrent congestion caused by the accident.Firstly,the historical congestion areas are identified based on clustering in the same space-time dimension,and the congestion frequency in the research period is counted to determine the recurrent congestion area in the spatiotemporal map and the possible bottleneck location.Then,the prior feature state vector is selected to design a KNN algorithm to search the nearest K states from the historical database,and the prediction accuracy is tested by the feature real value.The results show that compared with the historical average method,the KNN algorithm proposed in this paper can better predict the recurrent congestion,so it can be used to separate the impact of recurrent congestion on traffic event propagation. |
