Latching control of a point absorber wave energy converter in irregular wave environments coupling computational fluid dynamics and deep reinforcement learning
Hao Qin, Haowen Su, Zhixuan Wen, Hongjian Liang
Abstract
This paper proposes a novel latching control model coupling Computational fluid dynamics (CFD) and Deep Reinforcement Learning (DRL) to improve the wave energy capture performance of a point absorber wave energy converter (WEC). Firstly, a numerical wave flume (NWF) is built to generate unpredicted irregular waves. That simulates the two-way coupling interaction between the WEC and waves based on CFD, which creates the nonlinear environmental state space for the DRL input. In the meanwhile, a training method based on the Soft Actor-Critic (SAC) algorithm without explicit parameter adjustment is designed to implement a non-predictive latching control agent. Secondly, using the CFD-DRL coupling model, training for the latching control strategy is conducted in parallel irregular wave environments, and three state space configurations are evaluated to enhance the agent's generalization ability. Lastly, the wave energy capture performance using the proposed latching control model is compared with a traditional real-time latching method, and comparative analysis of two different training approaches is carried out. Simulation results show that the proposed latching control model outperforms the traditional real-time latching method in tests under irregular waves with different wave heights and frequencies, stably achieving more than 30 % wave energy conversion efficiency. This paper highlights the applicability and advancement of the DRL method applied in intelligent control of WECs, which may provide new insights for the wave energy and ocean engineering industries.