Structure damage identification in dams using sparse polynomial chaos expansion combined with hybrid K-means clustering optimizer and genetic algorithm

Structural damage identification plays a crucial role in structural health monitoring. In this study, a novelty method for structural damage identification is developed, which employs an advanced surrogate modelling technique to drive a new hybrid optimization strategy, namely a combination of K-means clustering optimizer and genetic algorithm (HKOGA). The core of this method is using the reliable sparse polynomial chaos expansion model as a cost-effective substitute for the computationally expensive structural finite element models, thus greatly improving the efficiency of the optimization strategy in finding the optimal value of the objective function. To evaluate the performance of this hybrid optimization strategy, seven optimization algorithms are selected and compared with it for 23 classical benchmark functions, and the comparative results show that the HKOGA has the best performance. Taking a small-scaled laboratory dam as an example, the efficiency and reliability of the proposed method to cope with the problems concerning finite element model updating and structural damage identification are explored. Two important findings are as follows. (i) For finite element model updating, compared to the conventional method based on iterative optimization, the proposed method improves computational efficiency by a factor of 59 while maintaining computational accuracy. (ii) For structural damage identification, leaving aside the huge leap in computational efficiency, the HKOGA has a faster convergence rate, stronger robustness, and higher accuracy than its sub-algorithm K-means clustering optimizer (KO). The results show that this method can be severed as an extremely efficient and potential tool to identify damage in large and complex structures.