Comparison of Several Error Metrics for FE Model Updating

Other conference contribution, 2007

The choice of error metric is one of the most crucial choices of model updating. Many articles have been published advocating the use of certain such metrics, as have comparative studies of various methods based on different error metrics. This article describes the differences between a number of error metrics and their corresponding objective functions using many different evaluation methods, such as plots of objective functions, Cramer-Rao lower bounds and condition numbers of the Hessian. This approach has the advantage that it gives the user a way of assessing error metrics in terms of solvability of the optimization problem and identifiability of parameters of the model updating problem.