Metrics and Guidelines for Rotorcraft Modeling Accuracy and their Relation to Model Uncertainty: A State-of-the Practice Survey

Abstract

The aim of this paper is to provide a survey on the validation metrics and criteria employed for validation of rotorcraft modeling accuracy, extending the present approaches to the uncertainty quantification (UQ) domain of validation. In this sense, the paper will review the current state-of-the-art metrics belonging to 1) the root mean square error method which are expressed as cost functions in frequency and time domain and 2) their enhancement towards the chi-square statistics consideration. The paper will extend afterwards the problem of validation towards uncertainty quantification (UQ) area with its distinction between aleatory and epistemic uncertainties. In UQ the emphasis is not on validation metrics but on the methods to validate the model, the choice of a metric depending on the problem under investigation. A simple example will be given for rotorcraft showing that the problem of flight controller modeling can become nondeterministic using different formulations for the equations of motion. A key message of the paper is that using UQ in the future can enhance our understanding of the simulation model fidelity.