Damage identification for functionally graded materials using improved experience-based learning algorithm based on sparse regularization

Abstract

This article proposes a new objective function based on sparse regularization for damage identification on functionally graded materials using an improved experience-based learning algorithm. The new objective function is developed to improve the accuracy and robustness of optimization analysis and thus to obtain better performance for damage identification. Functionally graded beam specimens of poly(methyl methacrylate) polymers with different damage situations are fabricated to reveal the effectiveness and robustness of the proposed method. Moreover, the influence of artificial noise on the identification performance is investigated. The damage identification results from both numerical and experimental validation on poly(methyl methacrylate) polymer specimens demonstrate the superiority of the new objective function for damage identification of functionally graded materials compared with the traditional objective function.