摘要
针对K均值算法无法直接应用于高维时间序列相似性分析的问题,基于相似洪水动态识别理论,结合K均值算法和投影寻踪模型的特点,构建了基于投影寻踪-K均值的相似洪水动态识别方法,并以池潭水库历史洪水为例进行验证。结果表明,基于投影寻踪-K均值的相似洪水动态识别方法具有一定的适用性,能够克服洪水间的相位问题,相似性识别结果合理,可为防洪调度提供决策支持。
To deal with the problem that K-means algorithm cannot be directly applied to the high-dimensional time series for the similarity analysis,a new similar flood dynamic recognition method is built that coupling K-means algorithm and the projection pursuit model based on the similar flood dynamic recognition theory.This new method was applied and verified to the historical floods of Chitan reservoir.The results show that the similar flood dynamic recognition method based on K-means algorithm and the projection pursuit model has certain applicability in recognizing similar floods,which can overcome the phase problem between floods.The reasonable recognition result can provide useful information for making decision of flood control operation.
作者
李映辉
钟平安
杨敏芝
刘畅
蔺宏岩
LI Ying-hui;ZHONG Ping-an;YANG Min-zhi;LIU Chang;LIN Hong-yan(College of Agriculture and Water Conservancy Engineering,Suihua University,Suihua 152061,China;College of Hydrology and Water Resources,Hohai University,Nanjing 210098,China)
出处
《水电能源科学》
北大核心
2020年第7期78-82,共5页
Water Resources and Power
基金
黑龙江省大学生创新创业训练计划项目(202010236002)。
关键词
相似洪水
动态识别
投影寻踪
K均值算法
similar flood
dynamic recognition
projection pursuit
K-means algorithm
