Deep reinforcement learning based intelligent flow control of vortex-induced vibration for box girder

An intelligent active flow control method based on deep reinforcement learning (DRL) was proposed to suppress vortex-induced vibrations and enhance the wind resistance of bridges. This method employed synthetic jets to disturb the wake vortex shedding of an aeroelastic bridge section model, effectively suppressing vortex-induced vibrations. Wind tunnel tests were conducted to validate the aerodynamic performance of the model under uniform and steady wind conditions. Additionally, the relationship between the control voltage and synthetic jet flow rate was established. A systematic analysis showed that the control voltage was approximately linearly and positively correlated with the average jet velocity, with higher control voltages significantly improving the suppression effect. Subsequently, the synthetic jet control strategy was optimized using the Soft Actor-Critic (SAC) algorithm, which converged rapidly to the optimal control voltage, resulting in a maximum reduction of 83% in vibration amplitude. These findings demonstrate that combining synthetic jet technology with DRL algorithms provides an efficient and intelligent solution for suppressing bridge vortex-induced vibrations and offers an intelligent approach for wind-resistant bridge design.