Wavelet ANOVA approach to model validation

摘要: Abstract Model validation is that critical component in the simulation development process ensures a model truly representative of system it meant to model. Although there are numerous techniques described literature, many these still require some amount subjective analysis order assess validity. This particularly true with dynamic output. To reduce or eliminate this subjectivity, paper proposes uses wavelet variance (WANOVA) as an effective method statistically accept reject valid. WANOVA approach performs statistical inference time-frequency domain take advantage sparsity and decorrelation. test statistic based on thresholded coefficients null hypothesis set data equivalent. The technique illustrated using study empirical from automobile crash study.