Research On Traffic Accident Judgment Method Based On D-S Evidence Theory

With the rapid development of traffic,there are more traffic accidents.Fast handling and accurate judgment of traffic accidents is import to reduce the impact of the accident and the social burden.To judge the traffic accident,the responsibility division are made by professionals which includes traffic police and judges.In order to improve the accuracy of decision and solve the problem that the judgment document is not standard and accurate,it is necessary to make a unified decision on the judgment results of several professionals.At the same time,as the judgment results made by professionals are highly uncertain,a data fusion method is required to express and process the uncertain information exactly.Basing on evidence theory,the main work is as follows:(1)In order to solve the problem of evidence conflict in traffic accident judgment,a traffic accident judgment method is proposed based on the conflicts of evidence sources.In the traffic accident judgement,the evidence for decision-making is given by experts according to their own experience.There may be large differences between the evidences given by different experts.In order to eliminate the impact of conflicting evidences on the results,a data fusion method is proposed based on the conflicts of evidence sources.This method defines the reliability of the evidence and the discount factor based on the conflicts between the evidences,and use the Shafer discount method to discount the original evidences to obtain the revised evidences.At the same time,in order to solve the problem of fusion of multiple evidence sources,the Dempster combination formula is used to perform iterative fusion.(2)In order to solve the problem of fuzziness in the traffic accident judgment,a method of road traffic accident judgment is proposed based on reliability construction.In order to improve the efficiency of classification,text classifiers are used instead of experts as evidence sources to provide evidences for data fusion.In order to solve the problem of the inconsistency between the classifier results and the reliability,the basic probability assignment is carried out by belief construction.The method gets uncertain information from the output results of the classifiers,and constructs the belief according to the type of the result.A belief construction method for soft classifiers based on information entropy is proposed,and a belief construction method for hard classifiers based on classifier reliability is proposed.The classifier results can be used to make unified decision by evidence theory after belief construction.In traffic accidents,to make full use of the judgment document,the text classifier is selected as the sub-classifier,and a method of generating word vectors in the classification process is given.This paper summarizes the research results in the end,and prospect the development prospects of traffic accident judgment.This paper includes 12 figures,13 tables and 52 references.