Power Network System Identification and Recovery Based on the Matrix Completion

Abstract

Health detection of power networks is a complex and important issue. In order to reduce the detection complexity of the power network, a matrix recovery method based on the matrix completion has been proposed in this paper. Matrix completion has attracted considerable attentions in computer vision, system identification, and machine learning in recent years. This paper presents a survey on algorithms about matrix completion including Singular Value Threshold (SVT), Alternating Direction Method (ADM) and so on. In order to reproduce algorithms about matrix completion and evaluate the performance of these algorithms, this paper is present. The numerical experiments are conducted to evaluate the performance of algorithms. An application about image denoising is also carried out.