Possibilistic NDT data fusion for evaluating concrete structures

M.A. Ploix (1) , V. Garnier (1) , D. Breysse (2) , J. Moysan (1)

1 LCND - Laboratoire de Caractérisation Non Destructive [Aix en Provence]
2 Ghymac - Géosciences hydrosciences matériaux constructions

A new application of data fusion is presented within the context of national research project named SENSO. The aim is to improve evaluation of indicators or pathologies of in situ concrete structures by combining measurements from different NDE techniques (radar, electrical resistivity and capacity, infrared thermography, impact echo and ultrasounds). Every non-destructive measurement is likely to provide an estimation of the unknown indicators with a certain confidence that is function of its reliability and its sensitivity to each indicator. When the estimations converge data fusion improves final confidence. In case of disagreement conflict can be managed by specific operators. We develop an adapted solution based on possibility theory that is particularly used in case of imprecise and uncertain data. This theory is very flexible in term of data representations and fusion operators and then requires adapted choices at every step of fusion process. Results are presented for simultaneous evaluation of water saturation and porosity ratio. They show that the chosen adaptative operator function of reliability is appropriate to the problem.

Domaines: Acoustique [physics.class-ph]; Acoustique [physics.class-ph]

M.A. Ploix, V. Garnier, D. Breysse, J. Moysan. Possibilistic NDT data fusion for evaluating concrete structures. Non-Destructive Testing in Civil Engineering, 2009, Nantes, France. ⟨hal-01300025⟩