“…A particular class of identification schemes, relies on adoption of an observer setup, often in the form of Bayesian filtering methods, such as the Observer/Kalman filter Identification (OKID) [7,8], which aim at identifying the system dynamics by recursively, or even in batch mode [9], assimilating data into the underlying model structure of the system at hand. These filtering formulations are particularly fitting within the context of virtual sensing, i.e., the task of inferring response quantities at unmeasured locations [10,11], or even unknown system properties. Virtual sensing is essential for tasks such as digital twinning, diagnostics of condition in critical yet unreachable locations and control.…”