Prediction of Offshore Wind Power Grid Connection Based on Correlation Analysis and BOXGBoost

There are many factors that affect offshore wind power generation, among which the load has the greatest impact on the power generation system. It can generate adverse factors such as harmonics when connected to the grid, causing damage to the safety of the power grid. This article proposes a hybrid model that combines correlation analysis and Bayesian optimization (BO) with Extreme Gradient Boosting (XGBoost). The model first uses correlation analysis to clarify the impact of various influencing factors on the load of offshore wind power grid connection. Subsequently, the factor with the highest correlation is selected as the input for the XGBoost model, which can achieve grid connected load forecasting after training. Simulation and analysis evaluated the performance of the XGBoost model enhanced by Bayesian optimization through model accuracy analysis, and the results showed that Bayesian optimization is effective in improving the prediction accuracy of the XGBoost model. The model's MAE for predicting grid connected loads decreased by 50.58%, MSE decreased by 36.85%, and R2 increased by 3.80%, indicating excellent performance of the proposed BOXGBoost model.