An accelerated differential evolution algorithm with new operators for multi-damage detection in plate-like structures

Abstract This paper proposes an Accelerated Differential Evolution (ADE) algorithm for damage localization and quantification in plate-like structures. In this study, the inverse damage detection problem is formulated as a nonlinear optimization problem. The objective function is established through the alterations of the structure flexibility matrix weighted with a penalty-function, used specifically to prevent the detection of false alarms. The ADE algorithm is designed via the introduction of three modifications in the standard differential evolution algorithm. Firstly, the initial population is created using knowledge we usually have about the damage scenario of a structure. Such initialization technique assists the algorithm to converge promptly. Secondly, in the mutation phase, a new difference vector, created based on the dispersion of individuals through the search space, is used to ensure the automatic balance between global and local searching abilities. Thirdly, a new exchange operator is designed and used to avoid the untimely convergence to local optima. Finite-element models of isotropic and laminated composite plates are considered as numerical examples to test the efficiency of the proposed approach. Numerical results validate the performance of the ADE method, in terms of both solution accuracy and computational cost and highlight its ability to locate and assess damage, even for large-scale problems and noise-contaminated data.