The Role and Impact of Online Deep Neural Network Training and Fuzzy in Seismic Structural Control

In recent decades, seismic structural control has developed significantly as an important research area in earthquake engineering, aiming to reduce earthquake-induced damage. Recently, the application of artificial intelligence techniques, particularly deep neural networks (DNNs) and fuzzy control, has garnered attention as innovative approaches in this field. The interaction between soil and structure is recognized as a fundamental challenge in earthquake engineering and structural design, significantly influencing the dynamic behavior of structures and necessitating advanced analysis and control methods. This research proposes an innovative online deep neural network fuzzy control (DNNFC) training method for more accurate modeling of structural behavior under complex and realistic conditions, accounting for nonlinear structural behavior and earthquake-induced uncertainties. The control process consists of three major phases: offline neural network training, online DNN training, and seismic response control of the structure considering soil-structure interaction effects. Results demonstrate that the proposed method effectively enhances structural responsiveness to severe earthquakes and serves as a valuable tool for structural system design and optimization. The potential of this approach for significantly improving safety and sustainability in the face of seismic hazards is evident.

Keywords:

Online training , deep neural network , fuzzy control , earthquake , SSI effect