Statistical methods, using the entire time-history, can be used to assess the impact response of an ATD (Anthropomorphic Test Device) in terms of its repeatability and reproducibility. In general, the methods generate a correlation relationship described as shape, magnitude and phase-difference between two time-histories’ in a given set of similar tests: for repeatability the relationship it is for the same ATD, for reproducibility it is for different ATDs of the same design and for biofidelity it is a relationship between ATDs and biomechanical response data from a series of human surrogate impact tests. The method uses the phase relationship to minimize the difference between any two time-histories through an alignment procedure and the magnitude and shape correlations are used to generate a parametric evaluation of the differences between any two time-histories, or set of time-histories.
This paper introduces a variance analysis using the entire time history to build additional foundation to the parametric evaluations using the magnitude and shape correlations and how they can be used to define repeatability and reproducibility ratings/criterion. The proposed methodology has been evaluated using two data sets based on HIII 50th dummy’s chest acceleration time histories observed in USNCAP tests. The first set consists of five tests from a single Lab. The second set consists of seven tests from labs different from the first set. A time-history parameter, V, (the normalized summation of squared point to point difference between a pair of signals) was introduced and used to perform statistical analysis of Variance (ANOVA) of the reproducibility of the time histories under investigation. In particular, the V-parameter has been analyzed using both ANOVA and T-test approaches.
The relationship between the parameter V and the parameters shape correlation and magnitude correlation is derived analytically. Using this relationship, criterions have been defined for reproducibility and/or repeatability with respect to the shape and magnitude correlations metrics. The criterions have been developed using a limited data set and may change as more data becomes available and is analyzed.