A novel wind power prediction model based on PatchTST and temporal convolutional network

Due to the unpredictable nature of wind, wind power forecasting still faces certain challenges. The accuracy of wind power prediction plays a crucial role in the stability of the whole system. To improve the accuracy of wind power prediction, this research proposed an innovative hybrid prediction model that utilizes a multi-layer perceptron, combined with a temporal convolutional network and PatchTST. Firstly, a multi-layer perceptron is introduced to capture higher-order features, and a temporal convolutional network is used to extract time-domain features from the dataset to capture the dynamic changes of wind speed; then, PatchTST is used to accurately forecast wind power. The results show that the proposed model performs well in terms of prediction accuracy and prediction speed. The minimal MAPE is 14.4%, the prediction accuracy is improved by 9.22%, and the power generation efficiency is increased by 0.31%. In addition, this research used Bootstrapping to estimate the probability interval of wind power to provide a more comprehensive wind power forecast. This study provides a new and effective tool in the field of wind power forecasting, helping to improve the stability of power systems.