Continuous control of structural vibrations using hybrid deep reinforcement learning policy

Jagajyoti Panda, Mudit Chopra, Vasant Matsagar, Souvik Chakraborty

Abstract: Highlights•A novel deep RL policy is introduced for active control of structures.•Proposed algorithm fuses the concepts of data-driven and physics-inspired RL.•The approach outperforms individual control policies.•Hybrid RL-controller works in digital and analog regime.•Designed controller showcases superior performance and faster execution time.