KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Parameter space
Multidimensional set of all possible combinations of parameter values of the physical model to be explored during calibration. The dimensionality of this space directly influences the complexity of optimization.
Gaussian regression
Probabilistic approach that models the relationship between parameters and physical model responses using Gaussian processes. It provides uncertainty estimates crucial for robust calibration.
Inverse calibration
Problem consisting in determining the input parameters of a physical model from output observations, often mathematically ill-posed. It requires regularization techniques and robust optimization methods.
Error metric
Standardized quantitative indicator measuring the deviation between calibrated simulations and experimental data, such as RMSE or MAE. The choice of metric strongly influences the identified optimal parameters.