Two non-probabilistic methods for uncertainty analysis in accident reconstruction

Abstract 

There are many uncertain factors in traffic accidents, it is necessary to study the influence of these uncertain factors to improve the accuracy and confidence of accident reconstruction results. It is difficult to evaluate the uncertainty of calculation results if the expression of the reconstruction model is implicit and/or the distributions of the independent variables are unknown. Based on interval mathematics, convex models and design of experiment, two non-probabilistic methods were proposed. These two methods are efficient under conditions where existing uncertainty analysis methods can hardly work because the accident reconstruction model is implicit and/or the distributions of independent variables are unknown; and parameter sensitivity can be obtained from them too. An accident case is investigated by the methods proposed in the paper. Results show that the convex models method is the most conservative method, and the solution of interval analysis method is very close to the other methods. These two methods are a beneficial supplement to the existing uncertainty analysis methods.

Keywords: Accident reconstruction, Uncertainty analysis, Design of experiment, Interval analysis method, Convex models