Research on Wind Power Prediction Model Based on Random Forest and SVR.

Zehui Wang; Dianwei Chi

Wind power generation is random and easily affected by external factors. In order to construct an effective prediction model based on wind power generation, a wind power prediction model based on principal component analysis (PCA) noise reduction, feature selection based on random forest model and support vector regression (SVR) algorithm is proposed. First, in the data preprocessing stage, PCA is used for sample data denoising; then the random forest model is used to calculate the importance evaluation value of each feature to optimize the selection of feature parameters; finally, The SVR algorithm is applied for training and prediction. Experiments show that the prediction effect of the model based on random forest and SVR is excellent, the root mean square error(RMSE) is 0.086, the average absolute percentage error(MAPE) is 23.47%, and the coefficient of determination(R2) is 0.991. Compared with the traditional SVR model, the root mean square error of the method proposed in this paper is reduced by 95.9%, and the prediction accuracy and the fit of the prediction curve are significantly improved.

WIND power; PREDICTION models; RANDOM forest algorithms; SUPPORT vector machines; NOISE control; FEATURE selection

EAI Endorsed Transactions on the Energy Web, 2024, Vol 11, Issue 1, p1

2032-944X

Academic Journal

10.4108/ew.5758