Reliability of the Data-Driven Identification algorithm with respect to incomplete input data

This work focuses on the influence of the input parameters of the Data-Driven Identification algorithm developed by Leygue et al. (2018). This algorithm allows to measure stresses, from displacement fields and forces applied to a structure; the particularity is the absence of underlying constitutive equation. In the case of real experiments, the data are incomplete; but it is proven here that with appropriate data handling, stress fields can be identified in a robust manner. The incompleteness of input data is twofold: some missing displacement values (close to the edges or in a noise-affected area) and also a partial force information. The study proves that recovering those missing data has to be done smartly so that no assumptions except the balance equation is made.