Structural Damage Identification Using Response Surface-Based Multi-objective Optimization: A Comparative Study

Abstract

Non-destructive structural damage identification (SDI) and quantification of damage are important issues for any engineering structure. In this study, a comparative assessment of the damage identification capability of different design of experiment (DOE) methods (such as, 2 k factorial design, central composite design, Box–Behnken design, D-optimal design and Taguchi’s OA design) used in response surface methodology (RSM) has been carried out. Three different structures (simply supported beam, spring mass damper system and fibre reinforced polymer composite bridge deck) have been used for various single and multiple damage conditions to access the comparative ability of the aforementioned methods in identifying damage addressing two critically important criteria: accuracy and computational efficiency. The study reveals that central composite design and D-optimal design are most recommendable among the five considered DOE methods for SDI. Two different input parameter screening methods (sensitivity analysis using RSM utilizing 2 k factorial design and D-optimal design, general sensitivity analysis) have been explored in this study, and their comparative performance is also discussed. It is found that both the methods used in sensitivity analysis for the purpose of input parameter screening in the damage identification process work satisfactorily. Performance of RSM-based damage identification algorithm for different DOE methods under the influence of noise has also been addressed in this paper.