DAMAGE DETECTION OF STRUCTURES USING RESPONSE SURFACE METHODOLOGY

Abstract

Structural damage identification and quantification of damage using non-destructive methods is 41 an important aspect for any Civil, Mechanical and Aerospace engineering structure. This study focuses on response surface based approach for damage detection in structures. Performance of different conventional response surface methods in identifying damage has been explored first using three different structures. D-optimal design and central composite design have been concluded to be the most efficient in detecting damage. A damage detection algorithm has been developed using random sampling- high dimensional model representation approach which is capable of handling large number of input parameters. The performance of this algorithm has been found to be quite satisfactory. Two major key factors in response surface based damage identification methodology are the sensitivity analysis for choosing most important input parameters and the optimization algorithm used in the inverse process. Different sensitivity analysis methods and multi-objective optimization algorithms have been used in this thesis and their comparative assessment has been carried out. Performance of all sensitivity analysis methods used in this thesis has been found to be satisfactory. Multi-objective genetic algorithm works better than other considered optimization algorithms. A methodology has also been developed in this thesis to analyze the reliability of response surface based damage identification techniques.