A Study On Comprehensive Evaluation For The Traffic Safety Of High-Grade Highways And Multifactorial Prediction Model For The Traffic Accidents

The development of the automobile industry and the transportation industry, promotes the society economic progress, at the same time, causes traffic accidents to occur is a serious menace to the social public safety. Owing to the severe traffic safety situation, analyzing and researching characteristics of the traffic accidents, controlling the continually occurrence of the traffic accidents, lowering the harm degree of the road traffic accidents are an important and urgent research work. In recent years, along with the high standard, large-scale construction of the highway networks, the occurrence context of the highway traffic accidents of our country has already had very great change, important characteristic of which is that the majority of serious traffic accidents occur at the high-grade highway above second grade. Therefore, to put up a research work for the traffic safety of the high-grade highway has profound realistic significance towards reducing the social economic loses associated with the road traffic accidents.The factors that affect the traffic safety are various, such as: the road user (driver and pedestrian), the vehicle, the road, the environment, and the traffic management level etc.. Further analysis indicates that the road factor is the profound cause for occurrence of the traffic accidents. International and domestic researches on the traffic accidents pay much attention to the human and the vehicle factors, have very few studies on the harmony of roads with the environment and management level factor, and even neglect the important function of the road factor in the traffic safety. Actually, accidents usually concentrate in those special road segments among the entire highway that doesn't harmonize with runing condition of the whole road. The surface causes of occurrence of the traffic accidents on these special road segments are some behaviors of breaking the traffic code, such as, break the speed limit, forcedly overtake vehicle, occupy roadway to drive etc., but inner cause of occurrence of the traffic accidents is that the designer doesn't be well aware of important function of the road structure based on the inducement principles and information of the transportation environments to the traffic safety, which confuses the cognition information of the driver by alignment characteristic, the road surface condition and the transportation environment etc., finally induces the driver to adopt wrong operations. According to statistical results, about 10 percent to 20 percent of the traffic accidents have direct relation with road condition, about 50 percent to 60 percent of the traffic accidents have indirect relation with defects of the road safety, related research also indicated that about 50 percent of the traffic accidents can avoid or reduce injury degree by improvement of the current road condition. Therefore, this dissertation takes the road factor that influences the traffic safety as emphasis to research, attaches importance to influence of road factor on the traffic safety, which insures the road transportation is unimpeded and safe.The dissertation is supported by sub-project A Study on Black-Spots of the High-Grade Highway and Prediction Model of Road Traffic Accident sponsored by science and technology development project of communication bureau of Jilin province (3L1057832417) A Study on Safety Technology Exploitation and Countermeasure of High-Grade Highway. On the basis of researches of the project, the dissertation aims at some important problem or difficult problem to research, particularly, accomplishes a great deal of in-depth study on traffic safety comprehensive evaluation for the high-grade highways and multifactor prediction model for the traffic accidents.Firstly, comprehensive evaluation for the traffic safety of highways is divided into two parts that are spot or section safety evaluation and holistic safety evaluation of highways respectively. In research of the spot or section safety evaluation of highways, considering integrality of statistics data of road traffic accidents and roads itself characteristics, two kind methods of identifying black-spots are researched. In course of field survey for the black-spots, we find that managements for traffic accidents archives of some high-grade highways in mountainous terrain are non-normative, so it is difficult to adopt the identification method based on statistics of accident history data to carry out analysis work for the black-spots. Aiming at this kind of circumstance as above-mentioned, and considering that because of complicated landform condition of the highways in the mountainous terrain, designer is hard to consider all indices simultaneously in the route design process, which often causes some parameter values selected are inappropriate or results in minimum values of some parameters are adopted simultaneously, easily causes the alignment combination indices change markedly, section-and-section joint is not smooth etc. cases for occurrence of traffic accident of the highway in the mountainous terrain. Therefore this dissertation carries through works for black-spots analysis by evaluating road alignment safety, searches profound reasons for forming the black-spots of the highway in mountainous terrain, which assures road safety, reduces menace of black-spots to traveling safety. Aiming at the high-grade highways that have entire accident statistics data, the dissertation mainly adopts the method based on accident data to identify black-spots, furthermore, improves existing filtration method for identifying black-spots from two aspects, namely confirming identification criterion and selecting reasonable filtration segment, which can attain better precision. Identifying black-spots, on the one hand, can find main influence factors on accidents, and put emphasis on these factors to reconstruct, consequently improve the traffic safety level of roads, on the other hand, can gain connection accidents with road traffic conditions, and accordingly identify limitations in design process for road traffic, research results of which can provide technology aid for traffic safety evaluation, is the first research intention of this dissertation.In research of the holistic safety evaluation of highways, because of complicated connection of various influence factors, for assuring veracity of identification and feasibility of identification means, aiming at characteristics of the high-grade highway in different terrains, different evaluation methods are adopted to evaluate road traffic safety. Aiming at fuzzy characteristics of road indices of highways in the mountainous terrain, Set value statistics of fuzzy statistics theory is adopted for holistic safety evaluation, which determine main influence factors on road traffic safety. Aiming at highways in rolling and flat terrain, on the basis of Grey clustering theory of whitenization weight function, Fuzzy consistency theory, and Analytic hierarchy process (AHP), a novel method for systematically evaluating road safety are developed. Grey Fixed Weight Clustering is adopted to confirm safety grade of roads. Weight of each clustering criterion is determined by applying the Fuzzy Consistency Matrix that transforms qualitative relation into quantitative computation to compute combined weight. Aiming at the highways with low safety grade, AHP is applied to analyze, establish an orderly hierarchy structure, make quantitative evaluation for qualitative indexes and confirm magnitude of different road factors to influence traffic safety. By thorough analysis to relationship road factors with accidents, can determine disadvantage factors that influence traffic safety, and by adjusting or perfecting these factors in indices system, can provide reasonable safety countermeasures for new-built road design, preventing and controlling occurrences of the traffic accidents, and rebuilding roads, which is the second research intention of this dissertation.Disharmony of various factors in road system is one of important causations that cause traffic accidents to occur. The factors have effect on one another by high non-linear relationship, which has biggish randomicity. Owing to characteristics of the road transportation system, for example, various factor, diverse relations, different time dimensions and high non-linearity, etc., and the complexity, the inconstancy and the inscrutability of environment, people very difficultly or impossibly obtain the complete information for system model under limited technique means and limited space-time condition, namely, the system is placed in a"poor"information condition, at the same time, information for system modeling is effected by noise, all the aforementioned reasons induce model to cannot obtain satisfying prediction result. Considering above-mentioned problems, and combining with quantified analysis of important road factors determined by road safety evaluation, multifactor model for road traffic accident prediction, namely the corn part of the dissertation are put forward, which realizes forecasting for the road traffic accidents influenced by various road factors and can effectual overcome the shortcoming that existing prediction models only consider influence of one factor on traffic accident, is the third research intention of this dissertation.A fit aspect for research is selected in the dissertation by carrying on thorough comprehension and analysis to research domain of the road traffic safety, inheriting the advanced research production of the international and domestic scholars, and combing with the concrete circumstances of the high-grade highway of our country, which provides novel research ideas and research means for science and research works of transportation domain in certain extent. Based on a great deal of thorough field survey, and accurate first-hand research data attained by applying advanced apparatus, this dissertation constitutes flexible and practical research scheme, uses combined methods of various theory to carry on thorough analyses for research objects. This dissertation does it's best to across-the-board explore the intrinsic laws of the road transportation system, enriches the approach of theories research and practice of the road traffic safety, provides practical and feasible safety guarantee to persistent and healthy development of the road transportation system. Innovative research productions of the dissertation are as follows:(1) Aiming at the condition of possessing traffic accident data and different characteristics of the roads, the dissertation constitutes different research strategies for the high-grade highways in mountainous terrain and flat and rolling terrain, and uses combined methods of various theories to research respectively.(2) Aiming at the high-grade highways with partial traffic accident history data in mountainous terrain, the dissertation puts forward the synthesized analysis method of using vehicle movement parameter that combines velocity with acceleration to identify the black-spots of the high-grade highway for the first time, breaks the limitation of existing identification methods for black-spots that only applies to the high-grade highways with entire traffic accident history data, which provides effectual research means for management and programming of road safety.(3) The dissertation applies Set value statistics theory to the high-grade highways in mountainous terrain for comparing the traffic safety of each highway section, simply, conveniently, and effectually confirms different safety problems that exist in each section.(4) The dissertation applies the synthesized method that combined Grey fixed weight clustering with Analytic hierarchy process to evaluate road traffic safety of the high-grade highway in flat and rolling terrain for the first time, quantitatively analyzes various road factors that influence traffic accident, and gives weight values according to effect magnitude of the road factors on traffic accident, which sufficiently reflects whole safety condition of roads.(5) Combining virtues of Grey theory that weaken undulation of data sequence with capability of neural networks that disposes of non-linear adaptability information, the dissertation develops multifactor prediction model for road traffic accident based on Grey-Radial Basis Function (RBF) neural networks, can effectual overcome the shortcoming which existing prediction models only consider influence of one factor on traffic accident. Aiming at the characteristics of different terrain, this dissertation put forward various quantitative methods for the road factors that influence traffic safety, which makes the prediction model to has more practicability.This paper researches of basic theory, method and actualizing means for black-spots identification, safety evaluation and traffic accident prediction by mode of combining systematization with logicality, which provides novel idea and new method for formation of system of traffic safety research and further research.