Research On Traffic Safety Characteristics And Accident Prediction Methods On "Highway Segments Passing Through Villages"

China is rapidly urbanizing as a result of the process of "Reform and Opening". This influences traffic safety significantly. As most small cities and towns are developed along national or provincial highways, traffic volumes in these areas are increasing and there is a growing incidence of traffic disturbances and crashes. The mixture of vehicles traveling through these areas with those that of local residences is a key cause crashes on such roads.Crashes are not only cause a large loss of social wealth, but create deep emotional scars. They often lead families into poverty and slow down the economic development of middle-small-size cities and towns. Therefore, it is important to analyze traffic safety during rapid urbanization, and particularly assess crash distribution characteristics and vehicle travel and disturbance characteristics quantitatively. Crash prediction models can help direct such safety analysis and can also improve the effectiveness of safety countermeasures.The main contents of this paper are as follows:(1) defining the concept of "Highway Segments Passing Through Villages" and giving the technical grade classification basing on systemic analysis and a summary of previous research;(2) designing trial plots based on analysis of key factors that influence safety on highways that pass through urban areas;(3) analyzing the influence of urbanization on traffic safety using the'fractional polynomial method'and generalized estimating equations (GEE), and developing urbanization rate Accident Modified Coefficients (AMC) using the safety level of Beijing as a standard;(4) analyzing the crash, speed, orbit and acceleration data, using the classical approach of principal-component factor and dynamic cluster methods;(5) constructing a crash prediction model using panel data for all crashes, fatal crashes, roadside crashes, rear-end crashes and collision crashes; defining a crash reduction factor for key safety countermeasures; and defining the safety classification criterion based on the prediction results.