Aircraft simulation validation using an instrumental variable approach

Abstract

A procedure is developed which offers the potential to validate aircraft simulation models using noisy flight test measurements. The proposed validation procedure is based on the instrumental variable parameter identification method. The instrumental variable method requires a choice of "instruments." For this research, the "instruments" are chosen using the response predicted by an available simulation model. With the “instruments” chosen from the predicted response, it is shown that the parameter estimates are correlated with only the measured input noise vector. In contrast, the generally used least-squares approach is shown to be correlated with both the state and input noise vectors.

Several studies are presented to demonstrate the utility of the validation procedure. These studies include input variations and noise variations. The method is demonstrated using longitudinal and lateral/directional axis cases derived from a nonlinear simulation of a high performance fighter aircraft. The results are presented using time response comparisons, eigenvalue comparisons, and identified stability derivative comparisons. The case study results confirm that the instrumental variable method performs better than the least-squares technique when the state noise level is high and the input noise level is relatively low.