Regression Models with Graph-Regularization Learning Algorithms for Accurate Fault Location in Smart Grids
IEEE Systems Journal
Fault location, graph theory, least-squares approximation, modal analysis, regression analysis, smart grids, transmission lines (TLs), voltage measurement
This article focuses on the design of a hierarchical framework for locating faults in smart grids by resorting to only modal components of three-phase voltage measurements. The search space for identifying the faulty lines is first limited to the impacted regions by the fault, which is determined through an improved graph analytic-based algorithm by contributing the system topology and attribute affinities. The faulty lines within the faulty regions are then identified by employing a heuristic index extracted from the wavelet multiresolution analysis of corresponding modal components. The fault location on the faulty lines is finally estimated by the regression analysis of two novel graph regularization-based learning models. This fault location proposal has been evaluated over numerous simulated scenarios on the IEEE 39-bus system with the measurements subject to sampling rate, fault resistance, and noise issues. The attained results validate the efficiency of the proposed framework.
Hassani, Hossein; Razavi-Far, Roozbeh; Saif, Mehrdad; and Capolino, Gerard Andre. (2021). Regression Models with Graph-Regularization Learning Algorithms for Accurate Fault Location in Smart Grids. IEEE Systems Journal, 15 (2), 2012-2023.