Study On Idetification Method Of Road Hazardous Locations

In recent years, with the great development of society and economy, the transportation industry has been growing in high speed. At the same time,traffic safety is coming to be the key problem which influences the development of transportation industry and even all the society. Therefore, study on road traffic safety certainly has great value。The paper studies three aspects of contents, including dividing road cell, developing safety level evaluation model and developing Bayes Statistic– road safety evaluation method. In the research, the first aspect is base of the total; the other two aspects are focus and emphasis.Dividing road cell is the necessary precondition to develop road safety evaluation method. Basic principle and method to divide road cell are studied, and the corresponding program is compiled by Visual Basic Application program. Calculation results of the program efficiently found the road cells that have more accidents. This paper analyzes limitation of using traditional linear regression model as accident regression model, and demonstrates the applicability of using generalized liner regression model as accident model. Using SPSS, a statistic analysis software, the paper develop and calibrate generalized liner accident regression model for three type of roads, including two motor vehicle lanes and two non motor vehicle lanes, two lane road with road marking and two lane road without road marking. The first two models are ideal, because the regression model and coefficient are significant. However, the third one is not ideal, because the regression model and coefficient are not significant. Reasons of the problem are confirmed as small sample size and randomicity of accident.Multivariate regression empirical Bayes approach can obtain excepted values and variance from accident model, hence it can overcomes the limitation of empirical Bayes approach which needs great quantity of similar road cells. The paper analyzed safety level of certain road in Tongzhou District of Beijing by multivariate regression empirical Bayes approach. Result of the analysis effectively overcomes the chanciness of historical accident data.