摘要：AbstractThe aim of this work is to develop quantitative approaches to manage uncertainty of variables or functions modeled by complex systems where both experimental and simulation data are available. We present the case of variables, where in current practices, an uncertainty study could be understood as a quantity of interest study of the model output: mean, quantile, threshold probability, etc…This situation is very common in engineering, where complex models exist and where the experimental data are difficult to obtain. First, we propose a method for model calibration based on experimental and simulation data, then we prove the consistency of this calibration procedure. The main tool used here is the empirical processes theory. Our final purpose is to incoporate simulated data from a complex model into an estimator (of a quantity of interest) based on experimental data, and then to compare the performance of our estimators to the classical estimators based on experimental data only.
关键词：Model calibration;experimental and simulation data;M-estimation;empirical processes;consistency
