第42卷第2期2023年2月ZhejiangElectricPowerVol.42,No.02Feb.25.2023基于CatBoost算法的短期光伏功率预测方法陈海宏1,易永利1,黄珅2,韩静怡2(1.国网浙江省电力有限公司温州供电公司,浙江温州325000;2.亿可能源科技(上海)有限公司,上海200090)摘要:光伏电站发电功率的间歇性与波动性对电网安全、稳定、经济运行的影响日益明显,因此需要不断提高光伏发电功率预测准确率,为电网灵活调度与规划提供准确信息。首先,介绍了短期光伏发电功率的预测算法、特征方程、预测流程以及评价指标。接着,通过SHAP方法对训练集所构造特征进行分析筛选,使用CatBoost算法进行训练。最后,通过与使用相同特征的其他机器学习算法模型预测精度的对比,表明所提方法有效提高了预测性能,证实了基于CatBoost算法、融合多维特征的模型在光伏功率预测中的优势。关键词:光伏发电;功率预测;CatBoost;SHAPDOI:10.19585/j.zjdl.202302009开放科学(资源服务)标识码(OSID):Researchonashort-termphotovoltaicpowerpredictionmethodbasedonCatBoostCHENHaihong1,YIYongli1,HUANGShen2,HANJingyi2(1.StateGridWenzhouPowerSupplyCompany,Wenzhou,Zhejiang32500,China;2.EQuotaEnergyTechnology(Shanghai)Co.,Ltd.,Shanghai200090,China)Abstract:TheintermittentandfluctuatinggenerationpowerofPVpowerplantshasanincreasinglyprominentim⁃pactonthesafe,stable,andeconomicaloperationofpowergrids.Therefore,itisrequiredtocontinuouslyimprovetheaccuracyofPVpowerpredictiontoprovideaccurateinformationforflexiblegriddispatchingandplanning.First,thepredictionalgorithm,characteristicequation,predictionprocess,andevaluationindexofshort-termPVgenerationpowerareintroduced.Afterward,thefeaturesconstructedinthetrainingsetareanalyzedandfilteredus⁃ingtheSHAP,andthetrainingisperformedusingtheCatBoost.Finally,bycomparingthepredictionaccuracywithothermachinelearningalgorithmmodelsusingthesamefeatures,thepaperindicatesthattheproposedmethodcanimprovethepredictionperformanceandconfirmstheadvantagesoftheCatBoostthatincorporatesmultidimensionalfeaturemodelsinPVpowerprediction.Keywords:PVpowergeneration;powerprediction;CatBoost;SHAP0引言目前,在“双碳”背景下[1],我国正...