Set-membership identification of an inertial wave energy system

Within the process of designing effective control strategies able to maximise the energy absorbed by wave energy systems, one of the crucial tasks to be achieved is the development of representative control-oriented models. Usually, these models are developed following a physics-based approach based on first principles. However, with this approach, the simplifications needed to derive models compatible with real-time control applications inherently add a certain degree of uncertainty, which could result in suboptimal energy-maximising performances. For this reason, the applications of system identification techniques are becoming popular in the field of wave energy. The consequent modelling process is data-based, and, in this way, the uncertainty associated with the device model depends on the data fidelity and the assumed model structure. However, the quantification of uncertainty is not trivial, and is done most of the time a posteriori, i.e after the system model has been identified. Within this paper, we propose a data-based modelling procedure based on the concept of set-membership, which provides dynamical models consistent with a given level of uncertainty. To assess the capabilities of this technique, we detail its application to the problem of identifying the dynamics of an inertial wave energy converter from data obtained through system-identification-oriented tests in simulation. The identified model is then validated on a dataset generated by different simulations, to assess the capabilities of this methodology.