CoSi: Correlation of signals—A new measure to assess the correlation of history response curves

• New validation metric (CoSi) to assess history response curves of eg. test and simulation is presented.
• Assessment of curves similarity by numerical factors regarding the features shape, amplitude and phase.
• Consideration of uncertainties in both curves.
• Using of new corridor method, corridor size can be defined in direction of both abscissa and ordinate.
• Automatic curve alignment to achieve best agreement between curves.
• As comparison to other metrics show results of CoSi are easy to understand and to interpret in terms of visual meaning.

In the context of CAE work it is often required to assess the level of agreement of two curves objectively, e.g. measured against numerically computed results. Therefore a new comprehensive measure is proposed in this paper. The new measure ‘CoSi’ (Co rrelation of Si gnals) allows to account for uncertainties in both curves. This is achieved by constructing a corridor around one curve which considers deviations in direction of both abscissa and ordinate. Here CoSi differs from other common corridor approaches which consider only the deviation on the ordinate. It is explained how CoSi aligns the two curves taking the uncertainties of the second curve by scaling and shifting into account. This leads to the best theoretical achievable agreement between the two curves. Based on the aligned curves, quality factors are calculated to evaluate the results in terms of amplitudes of the curves, their overall match in shape, the phase between the curves, and all these combined into a comprehensive quality factor. The properties and results of CoSi are compared with other metrics from literature using various examples.