Improving the prediction of wind speed and power production of SCADA system with ensemble method and 10-fold cross-validation

This research used machine-learning-based forecasting models to estimate a Supervisory control and data acquisition system's wind speed and electricity production. Six machine-learning algorithms were used to anticipate wind speed and electricity generation: a light gradient boosting machine, Gradient Boosting Regressor, an Ada Boost regressor, an elastic net, a lasso, and an ensemble (light gradient boosting machine and Ada Boost). To improve the performance of the light gradient boosting machine and Ada Boost algorithms, the ensemble (light gradient boosting machine and Ada Boost) was used to predict wind speed and production power of a Supervisory control and data acquisition system. Besides the ensemble method, the results of machine learning algorithms were obtained using 10-fold cross-validation, 5-fold cross-validation, and 4-fold cross-validation methods. The results of the algorithms were compared with each other. The results showed that the root-mean-square error of the ensemble method (light gradient boosting machine and Ada Boost) in predicting the three-month production power of the Supervisory control and data acquisition system was 11.78 with 10-fold cross-validation, and in predicting the wind speed was 0.2080 with 10-fold cross-validation.