Operational Modal Analysis (OMA) often relies on user-specified parameters, such as model order and input sequence lengths. These are typically selected based on expert knowledge. While stabilization diagrams effectively determine the model order, they are unsuitable for selecting other sensitive parameters. The existing criteria for the assessment of parameter sets in OMA fall short of meeting key requirements, including the independence from system dynamics, and applicability across different OMA methods.

This study proposes a common validation criterion based on input reconstruction and cross-validation techniques. The covariance- and data-driven stochastic subspace identification methods, are extended for the purposes of multi-block estimation and synthesis. Moreover, a modally decomposed formulation is selected to facilitate an additional assessment of the identified modal parameters.

The proposed method is verified through the use of synthetic and experimental vibration signals, as well as a full-factorial parameter study, which collectively demonstrate the robustness and consistency of the proposed criterion.

The key objective of the developed methodology is to enhance confidence in identified modal parameters, thereby facilitating more reliable assessments in e. g. structural health monitoring or life-cycle analyses.