基于稀疏自编码的智能电网海量运行数据清洗

Cleaning massive operation data of smart grid based on sparse self coding

为解决电网数据样本混合的问题并提升智能电网的运行与响应速率,提出基于稀疏自编码的智能电网海量运行数据清洗方法。按照稀疏自编码原则,建立自编码网络模型,联合数据样本稀疏矩阵,确定深度降维参数的实际取值范围,实现对智能电网运行数据的降维处理。根据数据仓库定义形式,估算数据样本运行状态,计算聚类清洗参量的具体数值,完成基于稀疏自编码的智能电网海量运行数据清洗算法的设计。实验结果表明,在稀疏自编码原则作用下,电网主机能够准确区分两类数据样本,从而解决电网数据样本混合的问题,达到提升智能电网运行与响应速率的实际应用需求。

In order to solve the problem of grid data sample mixing and improve the operation and response rate of smart grid,a method of Data cleansing for massive operation of smart grid based on sparse self coding is proposed.The self coding network model is established according to the sparse self coding principle,and the actual value range of the deep dimensionality reduction parameters is determined by combining the Sparse matrix of data samples to achieve the dimensionality reduction of the smart grid operation data.According to the definition form of data warehouse,the operation status of data samples is estimated,the specific values of clustering cleaning parameters are calculated,and the design of data cleansing algorithm for massive operation of smart grid based on sparse self coding is completed.The experimental results show that under the sparse self coding principle,the power grid host can accurately distinguish between two types of data samples,thereby solving the problem of mixed power grid data samples and meeting the practical application requirements of improving the operation and response rate of smart grids.