OUTPUT POWER PREDICTION OF WAVE POWER GENERATION SYSTEM BASED ON CONVOLUTIONAL GATED CYCLIC UNIT

In order to predict the wave output power efficiently and accurately, a mixture model of convolutional neural network and gated cyclic unit is proposed. The indirect prediction method is used to build a direct drive wave power generation system model, and CORREL function is used to analyze the correlation of different wave characteristics. Combining convolutional neural network to extract the relationship between characteristics and wave height in high-dimensional space, feature vectors are constructed. Through the gated cycle unit network for training, the output value of the full connection layer is inversely normalized to obtain the predicted wave height value. Input the built model to obtain the prediction value of wave output power. The simulation results show that the wave prediction algorithm of the mixture model is more efficient and accurate than that of other network models in the case of multiple feature inputs. © 2024 Science Press. All rights reserved.