Novel closed-loop identification algorithm based on the finite-dimensional signal subspace

Abstract This paper presents a novel approach for closed-loop identification of continuous-time systems. The method is based on the projection of measured signals onto a finite-dimensional signal subspace, which results in inherent robustness against measurement noise and capability for various structures for LTI models. The proposed method enables us to handle indirect closed-loop configurations as well as open-loop ones without any special manipulation. The effectiveness of the method is demonstrated through a numerical example for non-minimum phase and unstable plants in closed-loop.