A combined approach for wind power generation prediction based on CNN and double BiLSTM neural network

Abstract

With the demand for electricity is increasing in the world. The power supply share of renewable energy generation is currently increasing and will require a large amount of power supply. However, renewable energy generation will cause large power fluctuations, which will affect the charging efficiency and security of the power system. Improving the prediction accuracy of renewable energy generation can effectively control and reduce power fluctuations in power supply. In this paper, the method based on convolutional neural network (CNN) and bi-directional long-term and short-term memory (BiLSTM) is applied to wind power generation prediction using the past time-series data sets. Experimental results on 12 different really world datasets show that the proposed model can reduce the prediction error by up to 77.8% compared with CNN-LSTM.