Wind power plant short-term power prediction method based on feature selection and multi-level deep transfer learning

• Main Authors: PENG XIAOSHENG, JIA SHIYUAN, WANG HONGYU
• Format: Patent
• Language: Chinese; English
• Published: 11.06.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FORADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORYOR FORECASTING PURPOSES; ELECTRIC DIGITAL DATA PROCESSING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE,COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTINGPURPOSES, NOT OTHERWISE PROVIDED FOR

The invention discloses a wind power plant short-term power prediction method based on feature selection and multi-level deep transfer learning, and the method comprises the steps: dividing collected data into two data sets which are respectively used as a target wind power plant and a source wind power plant for transfer learning; firstly, subjecting data samples of a source wind power plant to multi-level division according to the degree of correlation with a target wind power plant, then creating a multi-level deep migration learning model of the target wind power plant based on the multi-level data samples of the source wind power plant, and finally, optimizing the multi-level deep migration learning model by adopting a feature selection method. Through prediction of the method, the data training scale can be reduced, data overfitting is avoided, and the method has popularization value. 本发明公开一种基于特征选择与多层级深度迁移学习的风电场短期功率预测方法，将采集到的的数据划分为两个数据集，分别作为迁移学习的目标风电场和源风电场，首先根据与目标风电场的相关程度对源风电场的数据样本进行了多层级划分，然后基于多层级的源风电场数