Predicting Wind Power Generation Using Hybrid Deep Learning With Optimization

Abstract

Accurate prediction of wind power generation is complex due to stochastic component, but can play a significant role in minimizing operating costs, and improving reliability and security of a power system. This paper proposes a hybrid deep learning model to accurately forecast the very-short-term (5-min and 10-min) wind power generation of the Boco Rock Wind Farm in Australia. The model consists of a convolutional neural network, gated recurrent units (GRU) and a fully connected neural network. To improve performance, the hyper-parameters of the model are tuned using the Harris Hawks Optimization algorithm. The effectiveness of the proposed model is evaluated against other advanced models, including multilayer feedforward neural network (NN), recurrent neural network (RNN), long short-term memory (LSTM) and GRU. The forecasting model demonstrates around 38% and 24% higher accuracy as compared to the 5- and 10-min forecasting of the NN model, respectively.