Machine Learning-Based Hybrid Optimization Method for Tuned Mass Damper Considering Seismic Soil-Structure Interaction

In this study, a machine learning-based hybrid optimization method is developed to optimize tuned mass damper considering soil-structure interaction. A total of 200,000 time history analyses of the tuned mass damper-structure-soil system are conducted and the results are used to construct the database of six machine learning models. Three meta-heuristic algorithms are used for optimization. A weighted optimization algorithm is developed based on the three individual optimization algorithms. Compared with three individual optimization algorithms, the weighted optimization algorithm can achieve better optimization results. The robustness of the single ground motion-based optimization applied to other ground motions is also investigated.