摘要
电网的运行需要大量电力大数据的支持,为了降低传输工作量,设计基于稀疏学习的电力大数据压缩与高精度重建方法。采用最优复杂度模型处理电力大数据的缺失值,通过基于残差学习方法的DnCNN去噪模型,对大数据去噪。根据向量主成分分析方法,对电力大数据进行压缩处理。基于稀疏学习构建大数据重建网络模型,实现电力大数据的重建。实验测试结果表明,设计方法的数据压缩比最高达到0.986,综合矢量误差整体低于0.3%,归一化均方误差整体低于0.8%。
The operation of power grid needs the support of a large amount of power big data.In order to reduce the transmission workload,a method of power big data compression and high-precision reco-nstruction based on sparse learning is designed.The optimal complexity model is used to deal with the missing values of power big data,and the big data is denoised through the DnCNN denoising model based on residual learning method.According to vector principal component analysis method,power big data is compressed.Build a big data reconstruction network model based on sparse learning to realize the reconstruction of power big data.The experimental results show that the maximum data compression ratio of the design method is 0.986,the overall integrated vector error is less than 0.3%,and the overall normalized mean square error is less than 0.8%.
作者
苏良立
王敏楠
余仰淇
肖娅晨
肖戈
SU Liangli;WANG Minnan;YU Yangqi;XIAO Yachen;XIAO Ge(Big Data Center of State Grid Corporation of China,Beijing 100052,China;State Grid Info-telecom Great Power Science and Technology Co.,Ltd.,Fuzhou 350003,China)
出处
《电子设计工程》
2024年第14期68-72,共5页
Electronic Design Engineering
基金
国网大数据中心-2021年中国电力消费指数建设(数经e)(二期)-数据工程项目(SGSJ0000FXXX2100134)。
关键词
稀疏学习
电力大数据
最优复杂度模型
向量主成分分析
sparse learning
power big data
optimal complexity model
vector principal component analysis