Frequency Response Function Estimation in the Presence of Missing Output Data

Frequency response function (FRF) estimation is part of nonparametric system identification in the frequency domain. The FRF measurements give a quick but deep insight into the dynamics of complex systems. Data samples can get lost in some applications due to sensor failure and/or data transmission errors. We want to overcome this problem without having to repeat the measurement and/or the experiment because this can be either impossible or too expensive. In this paper, we propose a nonparametric method for estimating the FRF and its variance of a single-input single-output system from known input and noisy output measurements with missing output samples. No particular pattern of the missing data is assumed. Moreover, the proposed method provides an estimate of the missing data and its uncertainty.