Wind turbine power generation prediction method based on correlation analysis

Wind energy has strong randomness and volatility, which poses difficulties for accurately predicting wind power generation. This article studies the prediction of wind turbine power generation through correlation analysis of various influencing factors. Firstly, preprocess the raw data by performing data cleaning, data normalization, and dataset partitioning. Then, the eXtreme Gradient Boosting (XGBoost) algorithm is used to select the feature data with the greatest impact on output power from all influencing factors, and a wind power prediction model based on Temporal Convolution Network (TCN) is constructed. Finally, ablation experiments and comparative experiments were carried out using public data sets to prove the effectiveness and progressiveness of the proposed prediction model. The results show that the prediction method based on feature extraction reduces the prediction error by 5%.