基于CNN-XGBoost模型的光伏功率预测

Photovoltaic power prediction based ona combined CNN-XGBoost model

为提升光伏功率的预测精度,首先,对数据预处理,包括缺失值处理及归一化处理;其次,利用皮尔逊相关系数分析出与光伏功率最匹配的气象因素,减少模型的输入维度,降低模型复杂度;最后,基于卷积神经网络-极限梯度提升决策树(convolutional neural network-extreme gradient boosting, CNN-XGBoost)组合预测模型进行测试。测试结果表明本文所提模型可以降低预测值的均方根误差,有效地提升了光伏功率预测的精度。

To enhance the accuracy of photovoltaic power forecasts,this study begins with data preprocessing,which involves missing values and normalization.Pearson correlation coefficient is then used to analyze meteorological factors that best correlate with photovoltaic power,thereby reducing the model′s input dimensions and complexity.Finally,a combined predictive model is tested using a convolutional neural network and extreme gradient boosting(CNN-XGBoost)model.The test results show that the proposed model successfully enhances the accuracy of photovoltaic power forecasts by significantly reducing the root-mean-square error in predictions.