Study on the Tendency of Acquiring Hydrodynamic Interaction Effects Among Multiple PA-WECs Being Learned the Control Strategies by AI Under a Single State

Abstract When multiple PA-WECs (Point Absorber Wave Energy Converters) are arranged in rows or grids, hydrodynamic interactions among them can increase the expected power generation per unit by up to 1.5 times if they are placed at optimal intervals and their movements are controlled based on these interactions. However, the optimal control law depends on the location and number of PA-WECs, complicating the practical application and standardization of wave energy farm. This means that the control laws need to be adjusted each time the number of PA-WECs changes, which is a disadvantage for mass production and standardization. Recently, AI-based methods for learning control strategies have been proposed and tested. This study investigates whether an AI model trained on single-unit scenarios can effectively capture the increased power generation due to hydrodynamic interactions when multiple PA-WECs are deployed. Comparisons with theoretical optimal values confirmed that even AI models trained only on single-unit scenarios can somewhat capture the interaction effects among multiple units. Thus, the results indicated potentially simplifying the control strategy for PA-WEC farms. This approach could lead to more efficient and scalable wave energy farm, enhancing the viability of wave energy as a renewable resource.