An excitation signal design for identification with the amplitude constraints

This paper proposes an optimal excitation signal design approach with amplitude constraints for open-loop system identification. For the convenience of imposing the amplitude constraints, generalized binary noise (GBN) is adopted as the excitation signal. Then based on the asymptotic principles of prediction error method (PEM), excitation signal spectral density and the upper bound of identified relative frequency-domain deviation are obtained via constructing and solving a set of linear matrix inequalities (LMIs). Owing to the statistical property of the obtained model, a probability constraint can be imposed to improve the model accuracy for robust control application. Simulation results demonstrate the advantage and effectiveness of the proposed approach.