Optimization of Inerter-Based Systems Using Intelligent Optimization Algorithms and BP Neural Network

In response to the varying dynamic control requirements of different vibration systems under external loads, displacement, velocity, and acceleration are selected as the control objectives. Intelligent optimization algorithms and BP neural network methods are employed to determine the optimal design parameters for series viscous mass dampers (SVMD), tuned inerter mass dampers (TID), tuned viscous mass dampers (TVMD), and tuned mass dampers (TMD). The effectiveness and robustness of these vibration control systems are subsequently compared. The results indicate that the optimal parameters obtained through intelligent optimization algorithms are more accurate than those derived from the analytical solution based on fixed-point theory. Among the four optimization algorithms — particle swarm optimization (PSO), wild horse optimization (WHO), snow ablation optimization (SAO), and differentiated creative search (DCS), the WHO algorithm demonstrate superior performance in terms of solution accuracy, stability, and efficiency. Furthermore, the use of BP neural networks for optimal parameter prediction offers high predictive accuracy, minimal error, and robust generalization capability, significantly improving prediction efficiency. Finally, under the same mass ratio, TVMD demonstrates superior vibration control effectiveness and robustness across separate optimization objectives, including displacement, velocity, and acceleration.