Data-Based Approach to Optimizing the Ocean Wave Energy Carpet Using Deep Neural Network

Abstract In this paper, we present the Neural Network-based Optimization Method, applied to optimizing the wave energy converter “wave carpet”. The proposed method can be applied to optimizing the computationally expensive objective function that other sequential optimization approaches fail to do. The results show that, in the simple case of single-frequency unidirectional incoming waves, this optimization method achieves the optimal carpet shape that can absorb 2.18 times more energy than the baseline circular shape, and in its best performance the neural network can optimize the carpet shape that absorbs 7 times more energy than the baseline, after being trained on a medium data set. Thus, the proposed method can be considered an effective approach to solving the optimization problems involving computationally expensive objective functions.