Uncertainty Analysis Of Vehicle Accident Reconstruction Results Based On Latin Hypercube Sampling

Automobile industry in our country is entering a rapid development period,auto possession increases year by year,with traffic accidents becoming increasingly prominent problem. In order to assist the judiciary in investigating the accident and finding out responsibility,protect rights and interests of victims,the accident must be careful of investigation and scientific analysis. Accident Reconstruction as an important means of identification , has attracted wide attention. Therefore, investigating accident reconstructions has its important social benefits.First, this paper introduces three kinds of common traffic accidents: vehicle-Pedestrian Accident, two wheels-vehicle crash, the vehicle's rear-end crash accident's characteristics and mechanics models. Using these models can make the whole accident recreate the actual situation, to identify and analyse accident.Accident Reconstruction models involve many parameters, but not all the parameters have significant impact on the effect. To help the accident investigation experts focus on the sensitive parameters to improve the accuracy of the results. This paper proposed the accident reconstruction parameter sensitivity analysis based on Latin hypercube sampling (LHS) design. First, building test table according to the design principle of the LHS in the known accident reconstruction parameters uncertainty's premise; Second, the numerical calculation based on this table; Third, through the regression analysis of experimental results, obtaining the approximate response surface functions of accident reconstruction models; Finally, getting parameters sensitive degree about the results on the basis of response surface functions.The Surveying parameters of the accidents are not the only accurate values, but have certain changing range. To make the results more objective and fair, accuracy and reliability of accident reconstruction results need to be judged. Combined LHS with Monte Carlo method, the uncertainty of traces in the accident sites can be introduced into the accident reconstruction results. According to distribution of uncertain parameters in accident, Monte Carlo extracts a large number of random samples, combined with the response surface function by Latin Hypercube Sampling data, obtains the probability distribution of the results of accident reconstruction.Finally, selecting three accident cases, the case 1 has been studied by uncertainty analysis method is analyzed in paper's method and Monte Carlo method. Results obtained by three methods are very close. For case 2 and case 3 are studied by the same way, which is proved that the method in this paper is applicable to uncertainty analysis of all types of accident reconstruction results. The use of the uncertainty analysis in this paper obtains the speed distribution of the accidents, which is more reasonable and objective than a single accident speed in deterministic methods.