A DNN-FISTA Hybrid Framework for Multi-Source Heterogeneous Data Cleaning in Smart Distribution Networks

The data of the distribution network has the characteristics of multi-source and heterogeneity, and is often accompanied by problems such as noise, missing values, and outliers, which seriously affect the stable operation of the power grid. This paper proposes a multi-source heterogeneous data cleaning technology that integrates deep learning and the Iterative Shrinkage-Thresholding Algorithm (ISTA/FISTA). By virtue of the powerful feature extraction ability of the Deep Neural Network (DNN) and the sparse optimization characteristics of the ISTA/FISTA algorithm, functions such as noise reduction, filling of missing values, and correction of outliers are achieved. The experimental results show that this technology can significantly improve the data quality, accurately and dynamically identify outliers, optimize the copy efficiency to improve the processing efficiency, and provide a solid guarantee for the stable and efficient operation of the distribution network.