Hydrodynamic performance and PTO optimization of a multi-body articulated wave energy device

The optimal damping allocation between different PTO groups remains an unresolved conundrum, hampering the full realization of their performance potential. Existing studies merely optimize single-side PTO damping or adopt identical damping for all joints and ignore the strong bidirectional coupling between inner and outer hinges, while simultaneously, combining multiple parameters will generate a large number of cases for calculation, requiring extremely long time and enormous computational costs. This study proposes an application-tailored iterative surrogate-assisted evolutionary framework coupling MLP and NSGA-II to undertake a systematic investigation. Leveraging the boundary element method, comprehensive simulations of wave energy devices operating under typical wave conditions are conducted. Rigorous validation against experimental data is subsequently performed, ensuring the reliability and accuracy of the numerical framework. With the random changes in wave conditions and damping distributions, a comprehensive database is constructed by numerical simulations, calculating the power, structural motion, cable force, and capture factors. These data serve as inputs and outputs, respectively, for training a prediction model, wherein a multilayer perceptron neural network is employed. Finally, the non-dominated sorting genetic algorithm II is applied to identify key damping parameters, which are used to supplement high-value samples and refine the prediction model in an iterative manner. This research successfully trained an accurate surrogate model and obtained optimized two-group inner-outer damping matching results with a relatively small amount of computational data, proving the feasibility and efficiency of the proposed framework. Under the optimal energy-capturing sea state, the capture factor has been improved by 16.7% compared to the equal inner-outer damping scheme, while the developed framework only requires a total of 542 sample sets to achieve optimal region refinement. The proposed framework provides an efficient optimization strategy for complex PTO parameter optimization problems of multi-body articulated WEC systems.