Study On Freeway Traffic Safety Level Of Service Classification Method

In order to evaluate freeway segment traffic safety conditions objectively and accurately and to increase the quantitative evaluation level, this paper carry out a series of related research on freeway segment division and traffic safety level of service classification methods.At present, in the traffic safety level of service classification, the first work is dividing freeway into one by one segment and then do traffic safety evaluation to every single segment. On the foundation of analysis existing domestic and international segment division methods, from the perspective of traffic safety, the dynamic clustering segment division method based on accident is propose. Compare this segment division result and the fixed- lengths through case study to find that this new method can improve the identification work's impartiality and the level of automation greatly.On the basis of segment division, this paper chooses accident rate, injury rate and death rate as traffic safety level of service classification index. Firstly, by establish the relationship of accident rate, injure rate, death rate, the three weighted composite index and AADT. Secondly, calculate the mean and standard deviation of each flow range. Thirdly, based on probability distribution theory and use the mean and standard deviation to form a series of cut-off point. By fit these cut-off point to find out curve, which can divide the accident sample space, and each region of space that represents a traffic safety condition. Learn from the U.S. HCM of highway level of service, freeway traffic safety level of service is divided into six levels too. With Beijing-Zhuhai freeway accidents and related data gives the classification standards under different range of AADT.According to the segment accident rate, injury rate and death rate to determine the traffic safety level of service have some certain of pattern recognition features. This paper establishes another traffic safety level of service classification method based on self-organizing neural network. With the accident rate per one hundred billion vehicle kilometers, the injures rate per one hundred billion vehicle kilometers, the death rate per one hundred billion vehicle kilometers, the three integrated weighted value and the three weighted index as input variables to get segment traffic safety level of service. This method can classify traffic safety condition automatically. Finally, compare the freeway segment traffic safety level of service based on self-organization neural network and probability distribution to find that no matter what index is, the fifty-five segments'traffic safety level of service has a high conformity. These two methods have their own applicability on freeway segment traffic safety level of service classification.