Online automatic and robust control configuration selection

This paper presents a complete method for automatic and robust control configuration selection for linear systems which relies upon acquired process data under Gaussian noise excitation. The selection of the configuration is based on the estimation of the Interaction Measure named Participation Matrix. This estimation is derived with uncertainty bounds, which allows to determine online whether the uncertainty is sufficiently low to derive a robust decision on the control configuration to be used or if the uncertainty should be reduced by e.g. prolonging the experiment to obtain more data.