Short-term wind power forecasting model based on random forest algorithm and TCN

Abstract

Wind power has strong stochastic volatility, short-term forecasting accuracy is not high. With the increase of grid connected capacity of wind farms, the overall prediction error has a greater impact on the system operation. This paper investigates the short-term power prediction problem for wind farms based on historical wind power and meteorological data. First, the preprocessed original data are divided into training set and test set, and the random forest (RF) algorithm is applied to screen out the feature data that have the greatest impact on the output power from all the feature data of the training set. Then, the training and the test sample sets are reconstructed based on the screening results, and the temporal convolution network(TCN) is built by finding the optimal hyper-parameter combination of the network with the grid search algorithm. Finally, comparative experiments for a wind farm in a certain area of North China are presented to demonstrate the effectiveness and advancement of the proposed prediction model.