| With the number of automobile increasing, mechanization of society promotes and traffic accidents become more and more severity. In the course of Reformation and opening to the world , traffic accident ratio and severity degree of accident always increase and thing of traffic safety is very grimness in china. In order to cut down traffic accident and severity degree, traffic engineers have been striving and exploring for that. On the base of researching traffic safety and collecting the traffic accident data in YunNan province as well as Kun Ming, ShangHai , DaQing and Ha ErBin city so on, this paper studies the regularities of distribution of traffic accident , accident causes principles, safety evaluation and prediction method and some countermeasures for safety improvement based on the following important project" the project of traffic synthesis management in KunMing". In addition, traffic conflict technique that is a new analysis method is studied for evaluating traffic safety further . According to accident data in KunMing ,this paper analyses accident causes and draws a conclusion that human and automobile are primary causation producing road accident. This paper sets up associated model based on gray theory and studies the degree of all kinds of human and automobile factors affecting traffic accident in order to make countermeasures for traffic improvement pertinent. This paper analyzes the regularities of distribution of traffic accident in urban street in which there are time, space, situation, street condition, management condition, climate condition, and find out distributing characteristic and main problem in traffic safety. There are illegibility about traffic safety and diversity about thinking fashion, this paper discusses method of fuzzy evaluation discriminating black spot and establishes evaluation model regarding death rate per ten thousand vehicle as evaluation, equivalent comprehensive accident date and exponent of accident severity index, and discriminates black spot about intersections by means of this method in KunMing. Thanks to the drawback there are slow of training speed and very small spot of goal function, improved BP algorithm is advanced, that is, appending momentum item and auto adaptive training ratio, and Prediction of Traffic accident based on BP neural network is established. Owing to the fact of accident stat. in china, It makes occurring of traffic accident to become serial time event and winkles complex reasons leading traffic accident in urban street. The method that predicting the number of accident, death people and injured people is advanced based on serial time, and model is validated by data in KunMing. There are any relationship between the number of accident, accident fatalities , injuries and direct loss, so the model of direct loss prediction for street traffic accident is firstly put forward based on auto adaptive fuzzy-neural network system. According to the data, its model is established. Its thought is to provide a training way for achieve information from data muster as to design fuzzy model, and to effectively calculate the most parameter of subject function, which make inference system to simulate anticipant or actual input-output relationship. This would avoid situation that establishing model of prediction for accident loss is difficulty based on traditional math method and a great lot data is needed. At present, there are many methods for evaluating traffic safety in urban street. Because there are different situation, there are diversified methods. It includes macroscopic and microcosmic evaluation. This paper firstly analyses all kinds of evaluating methods as to find out merit or disadvantage and applied range of them, so that traffic engineers make impersonal and exact evaluation on account of each method. This paper advances different ways to apply vary objects including intersection, urban street and the whole city. There is illegibility that evaluating index affect to safety degree, so this paper firstly put forwar...
|
PDF Full Text Request