{"number":"CN111985711A","kindCode":"A","documentType":"publication","title":{"original":"一种风电功率概率预测模型建立方法","english":"A method for establishing a probabilistic prediction model of wind power"},"abstract":{"original":"本发明公开了一种风电功率概率预测模型建立方法，包括以下步骤：剔除初始数据集中的异常值，并基于灰色关联理论，选取与风电功率关联度大于预设阈值的气象变量作为风电功率概率预测模型的训练数据集；建立改进自然梯度提升元模型：实现参数向量更新；对多个改进自然梯度提升元模型进行Blending模型融合，建立新的元模型进行训练，从而输出最终预测统计参数向量。本发明提供的风电功率概率预测模型建立方法，可提供完整风电功率不确定性信息，具有更高预测区域覆盖率和更小预测区间平均宽度占比，为构建高效智能的新能源电力系统提供更准确参考。","english":"The invention discloses a method for establishing a wind power probability prediction model, comprising the following steps: removing abnormal values in an initial data set, and based on grey relational theory, selecting a meteorological variable whose correlation degree with wind power is greater than a preset threshold as the wind power probability prediction The training data set of the model; establish an improved natural gradient boosting meta-model: implement parameter vector update; perform Blending model fusion on multiple improved natural gradient boosting meta-models, establish a new…"},"applicants":["North China Electric Power University Baoding"],"inventors":["李永刚","王月","吴滨源"],"filingDate":"2020-08-19","publicationDate":null,"grantDate":null,"verifiedWithOffice":false,"pdfUrl":"https://downloadpatent.com/pdf/CN111985711A.pdf","pages":{"en":"https://downloadpatent.com/patent/CN111985711A","zh-Hant":"https://downloadpatent.com/zh-Hant/patent/CN111985711A","ja":"https://downloadpatent.com/ja/patent/CN111985711A"}}