Surrogate-Assisted Differential Evolution for Wave Energy Converters Optimization

In the aftermath of the successes in wind and solar energy, wave energy has emerged as an exceptionally promising renewable resource. A pivotal aspect of wave energy development involves optimizing the placement of energy converters in constrained sea areas to achieve maximum output. However, the hydrodynamic interactions among buoys in wave energy arrays pose a formidable challenge, making power assessment modeling both expensive and intricate. This study introduces a surrogate-assisted chaotic differential evolution (SDE) algorithm to address this challenge. The proposed methodology replaces computationally demanding evaluations with trained neural network models. Significantly, in contrast to prior surrogate models, we employ a selection of elite individuals to guide the evolutionary algorithm's search direction through confidence ranking. We curated eight datasets from four real wave scenarios (Perth, Adelaide, Tasmania, and Sydney) with two buoy configurations (4 and 16 buoys). Subsequently, we trained five state-of-the-art deep network models and two numerical regression models to identify the optimal model. Experimental results demonstrate that SDE exhibits superior energy output and computational efficiency compared to the other four intelligent algorithms when optimizing the locations of a highly challenging 16-buoy wave energy converter array in authentic wave environments.