Neuro-fuzzy Approach for Reconstructing Fissures in Concrete's Reinforcing Bars

In concrete, metallic bars are used to reinforce the mechanic resistance of the structures. When the structural element is subject to strong traction stresses, the main efforts load just on the bars. Thus, they are mainly subject to problems of ruptures within the concrete. Therefore, a very useful application of Non Destructive Testing could be the implementation of decisional tool for characterizing the status of the bars, and the eventually existing breaks and cracks. This relevant inverse problem is solved by means of a system which extracts information on the specimen under test from the measurements and implements a priori constraints to facilitate the detection of defect, if any. A Neuro-Fuzzy approach is proposed in this paper to locate defects on reinforcing bars in concrete specimens applying eddy current-based measurements. The method exploits the concepts of fuzzy inference to localize and estimate the defect. A comparison with Neural Network estimators is presented.