Objective functions

Name

Name of the objective function.

Function type

AutoCal uses three different functions for aggregation of the defined output measures:

Weighted sum:

(1.7)   

Weighted sum of absolute values:

(1.8)   

Weighted sum of squares:

(1.9)   

where F_j is the output measure and n is the number of measures that are pooled.

Typically, output measures within a certain area that measure the same sta­tistic for the same physical variable are pooled to evaluate the average model performance for that variable in the specified area with respect to bias (Avg. Error), dynamical behaviour (St. Dev.) or an overall goodness-of-fit (RMSE). The event-based statistics are typically pooled into an aggregate error of maximum and minimum values, respectively.

Weight

The weight assigned to the objective function in the aggregation of the differ­ent objective functions into one aggregate measure to be optimised. The assigned weights should reflect the relative priorities given to the different objectives, depending on the specific model application being considered. For investigating the entire Pareto front between the objective functions in a multi-objective optimisation, the aggregated measure can be adopted by per­forming several optimisation runs using different weights.