基于深度嵌入聚类的风光水典型联合出力场景提取. (Chinese)

It is a prerequisite for resource characteristics analysis and scheduling optimization research to screen out the representative typical scenarios of wind-solar-hydropower integrated generation systems. Due to the influence of the “dimension effect”, traditional clustering algorithms cannot be directly applied to high-dimensional data cluster⁃ ing, and the existing technical route based on “dimension reduction before clustering” cannot guarantee that lowdimensional features after dimensionality reduction are suitable for clustering tasks, resulting in unstable clustering results. In view of the existing problems, this paper proposes a method for extracting typical output scenarios of windsolar-hydropower based on DEC （deep embedding clustering). The method can realize high-dimensional output data clustering and avoid that low-dimensional features after dimensionality reduction are not suitable for clustering tasks. First, with the help of the nonlinear representation ability of the stacked autoencoder, the high-dimensional windsolar-hydropower combined output data is deeply represented to achieve data dimensionality reduction. Then, the Kmeans clustering method is used to cluster the deep low-dimensional features, and the stacked encoder is optimized and adjusted at the same time in the clustering process to obtain the low-dimensional wind-solar-hydropower com⁃ bined output feature suitable for the clustering space. Moreover, the precise division of wind-solar-hydropower com⁃ bined output scenarios is realized. Finally, the DEC is performed on the wind-solar-hydropower output data of a re⁃ gion in south China. The PCA-K-means algorithm is used to set up a comparison example to verify the effectiveness of the DEC in selecting typical combined output scenarios of wind-solar-hydropower generation. [ABSTRACT FROM AUTHOR]

对于风光水联合发电系统筛选出具有代表性的典型场景, 是进行资源特性分析、调度优化研究的 基础. 传统聚类算法由于“维度效应”的影响, 无法直接应用于高维数据聚类问题, 而现有的“先降维, 后 聚类”技术路线无法保证降维后的低维特征适用于聚类任务, 导致聚类结果不稳定. 针对存在的问题, 提 出了一种基于深度嵌入聚类的风光水典型出力场景提取方法, 在实现高维出力数据聚类的同时避免了降维 后的低维特征不适用于聚类任务的问题. 首先借助堆叠自编码器对数据的非线性表征能力, 对高维风光水 联合出力数据进行深层表征实现数据降维, 然后结合K-means聚类方法对深层低维特征进行聚类, 并在聚 类过程中同时优化调整堆叠自编码器, 得到适用于聚类空间的低维风光水联合出力特征, 基于此实现对风 光水联合出力场景的精准划分. 最后以我国南方某区域风电、光伏、水电出力数据为研究对象对其进行深 度嵌入聚类, 并以PCA-K-means算法设置对比算例, 验证了深度嵌入聚类在风光水典型联合出力场景选 取上的有效性. [ABSTRACT FROM AUTHOR]

Copyright of Zhejiang Electric Power is the property of Zhejiang Electric Power Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)