Damage localization of structures using adaptive neuro-fuzzy inference system

The identification of cracks or similar damages at their initial levels in the load carrying members of structures is an important problem in civil and earthquake engineering. In this paper, adaptive neuro-fuzzy inference system (ANFIS) method was employed as a new approach to make prediction on damage locations in beam like structures. Single and multiple damages in some parts of numerical samples are caused by concrete elasticity modulus reduction. Mode shapes and modal strains of damaged beams were analyzed through ANFIS approach and the results were compared. Analyses suggest that decision making based on a particular mode shape or modal strain would not be applicable and using some vibrational modes seems to be essential to damage detection. Results also show that modal strains are significantly more sensitive for damage localization than mode shapes and the locations of damage scenarios can be predicted precisely