High-Dimensional Adaptive Test Design Including Boundary Search

This article investigates an adaptive test design approach for the purpose of a model-based engine calibration. Two different new algorithms are presented to take engine boundaries during test execution into account and selectively calculate new test points to increase engine model quality and its input domain. The algorithms are implemented into an adaptive test design framework and evaluated by an engine simulation with artificial Brownian noise added. The results highlight an increase in input space evaluation volume and a decrease in engine model error, while meeting calculation time constraints.