Local polynomial method for structural identification

Frequency domain system identification is widely used in the area of structural identification. However, due to measurement noise and leakage errors in frequency domain, the identified model may be inaccurate. The local polynomial method is (LPM) is a recently developed procedure for estimation of the frequency response function (FRF) of a linear system considering the leakage errors. In this paper, we propose a new approach combining the LPM with the maximum likelihood (ML) for parameter identification of the system dynamics in a generalized output error framework. We show the benefits of the proposed approaches in a numerical example.