A new poly-reference complex frequency-domain identification technique formulated in state-space model

When performing experimental modal testing in complex structural systems, engineers face the challenge of extracting their modal properties in an accurate, clear and robust manner. Several time and frequency-domain modal parameter estimation techniques exist for this purpose. Among these are the recently introduced poly-reference Complex Frequency (pCF) and its subspace-based counterpart know as Subspace-based pCF (SpCF). Both techniques are formulated based on the assumption that responses are contaminated with white noise and that the input forces can be modeled as (ideal) impulses or white noise processes, which implies that their corresponding spectra are constant in frequency range of DC to the Nyquist frequency. This paper describes the performance assessment of a new variant of the pCF method where the process noise is no longer assumed to be white and can be modeled with a polynomial function. In order to assess its performance, the technique is applied to estimate the modal properties from heavily noise contaminated simulated Frequency Response Functions and from the responses measured in an output-only vibration test of a tower-like steel specimen.