Stochastic model updating using distance discrimination analysis

This manuscript presents a stochastic model updating method, taking both uncertainties in models and variability in testing into account. The updated finite element (FE) models obtained through the proposed technique can aid in the analysis and design of structural systems. The authors developed a stochastic model updating method integrating distance discrimination analysis (DDA) and advanced Monte Carlo (MC) technique to (1) enable more efficient MC by using a response surface model, (2) calibrate parameters with an iterative test-analysis correlation based upon DDA, and (3) utilize and compare different distance functions as correlation metrics. Using DDA, the influence of distance functions on model updating results is analyzed. The proposed stochastic method makes it possible to obtain a precise model updating outcome with acceptable calculation cost. The stochastic method is demonstrated on a helicopter case study updated using both Euclidian and Mahalanobis distance metrics. It is observed that the selected distance function influences the iterative calibration process and thus, the calibration outcome, indicating that an integration of different metrics might yield improved results.