Wind power prediction based on CNN-LSTM
Huiting Zhang, Li‐Xun Zhao, Zhipeng Du
Abstract
Accurate wind power prediction is important to formulate power generation plans, reduce the impact of wind power integration into the power grid, and maintain the safe and stable operation of the power system. However, wind power output is affected by various factors such as wind speed, wind direction, temperature, air pressure, and humidity. So, it is obviously hard to accurately predict wind power over a long period of time only by relying on the historical power generation data. Therefore, this paper takes into account the influence of wind speed, wind direction, temperature, and humidity, and proposes a wind power prediction method based on CNN-LSTM. The method first uses the good feature extraction ability of convolutional neural network (CNN) to extract multiple features of wind power and its influencing factors data, then input them into the long and short-term memory network (LSTM) for predict, and finally realize the time series prediction of wind power. To prove effectiveness of the proposed method, use the method predict power generation of a wind farm in a certain area, and compare the prediction results with other methods. At last, the results show that the proposed method is effective and feasible.