Wind power prediction method based on space-time meteorological feature extraction and deep learning

• Main Authors: WANG NAN, PENG XIAOSHENG, YUAN SHUAI, JIA SHIYUAN, ZHANG YUANPENG, YANG ZIMIN, CHE JIANFENG, CHENG YAN, WANG BO
• Format: Patent
• Language: Chinese; English
• Published: 29.10.2021
• 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; 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 space-time meteorological feature extraction and deep learning-based wind power prediction method. The method comprises the following steps: based on wide-area space-time meteorological data and power data, researching cross-correlation characteristics of output of a new energy station and a weather process, establishing multi-level sub-region division based on different indexes, constructing a high-dimensional candidate feature library based on multi-dimensional meteorological data, constructing composite meteorological features based on data mining, finally constructing a multi-level-oriented deep learning model library based on high-dimensional deep feature mapping and high-dimensional deep data mining based on massive samples and optimized core features, and selecting an optimal model to perform cluster power prediction. Through prediction of the method, prediction of the wind power under the space-time composite data is realized, an effective matching relationship is established