Study On Traffic Safety Risk Analysis On Highways In Mountainous Areas Under Unfavorable Weather

Highways in mountainous areas are one of the important infrastructures to drive the economic growth of mountainous areas and to fully construct a well-off society.However,due to the limitations of objective conditions,traffic accidents easily occur on highways in mountainous areas as a result of many tunnels,small curvature radius,and many long downhill sections.In particular,under unfavorable weathers such as rain,snow and fog,the highways in mountainous areas are of reduced road friction coefficient,lowered visibility,and increased braking distance of vehicles,further increasing the risk of traffic accidents.Therefore,with the scientific and efficient methods coupling mechanism of traffic accident risks on highways in mountainous areas under unfavorable weather is analyzed,so as to work out main factors resulting in traffic accidents.Such factors are significant to reduce traffic accidents,lower severity of the accident and to guarantee the traffic safety on highways in mountainous areas under unfavorable weather.First of all,we have had a literature review on current domestic and abroad study the progress of traffic safety on highways in mountainous areas under unfavorable weather,comprehensively illustrated the main study method Bayesian Network theory used,and analyzed the characteristics and advantages of Bayesian Network in the traffic risk analysis study as the theoretical foundation.Secondly,we have analyzed the investigated traffic accident data on highways in mountainous areas of Shanxi Province under three typical unfavorable weather of rain,snow and haze,further analyzed,studied and judged the difference of effects of related risk factors on each traffic accident result under two types of weather.At the same time,taking section K600+500~K682 of Beijing-Kunming Highway as an example,we have calculated the traffic accident rate under unfavorable weather(rainy),completely analyzed the traffic safety situations on highways in mountainous areas under unfavorable weather.Thirdly,based on the actual traffic accident data acquired and statistical analysis,we have identified the risk factors affecting traffic accidents and selected 11 highly related nodes.The chisquare test was applied to judge the independence and dependence between nodes,meanwhile,PC constraint algorithm was employed to learn about the structure of the independence group aiming to obtain the structure of Bayesian Network.Expectation Maximization(EM)algorithm was used to estimate parameters and finally determined the Bayesian Network model for traffic safety risk analysis on highways in mountainous areas under unfavorable weather.Through the error analysis of the Bayesian Network model,the results show that the model possesses excellent fitting precision.Therefore,the Bayesian Network model is used for sensitivity analysis and rational analysis to obtain main factors affecting traffic accidents.Finally,based on the inferential analysis results of Bayesian Network model,the paper proposes the risk prevention and control measures of such aspects as the construction of traffic safety facility,optimization of traffic signs and markings,and accident rescue for the four main nodes affecting traffic accidents,namely,vehicle type,driving behavior,road alignment,and intersection,which provides certain references for traffic authorities on highways in mountainous areas dealing with the traffic accidents due to unfavorable weather.