Support Vector Regression for Hybrid Seismic Control of Nonlinear Base-Isolated Structure Equipped with an Active Tuned Mass Damper

The increasing demand for reliable seismic protection has motivated the development of intelligent control strategies that can address nonlinear structural behavior. In this study, support vector regression (SVR) is employed to control a nonlinear base-isolated structure equipped with an active tuned mass damper (ATMD) at the base. The framework builds upon the principle of hybrid control, where passive isolation is complemented by active damping to enhance vibration mitigation. The SVR model is trained using data generated by a Linear Quadratic Regulator (LQR) under a broad band input earthquake, enabling it to reproduce the optimal control strategy without explicit reliance on structural models. The primary contributions of this study lie in demonstrating that an SVR-based controller can achieve substantial seismic response reduction while relying on fewer sensors than traditional algorithms. In the considered framework, ATMD is integrated with the base isolation of an eight-story benchmark building to mitigate vibrations under three different earthquake signals. The performance of the proposed SVR-driven ATMD is compared to both the base-isolated structure alone and the LQR-controlled ATMD. The results demonstrate significant reductions in base displacement and inter-story drift, while simultaneously lowering the number of sensors required compared to traditional full-state controllers, thereby underscoring SVR as a practical and efficient approach for enhancing hybrid seismic control in smart structures.