Optimized ensemble of regression tree-based location of evolving faults in dual-circuit line

Abstract

In a dual-circuit transmission line, the location of evolving faults (EVFs) is more tiresome due to its multifaceted nature. In this paper, a novel data-mining-based scheme is proposed for location of EVFs by using an ensemble of regression trees, that is, bagged regression trees and boosted regression trees. This ensemble of regression tree modules is trained with optimized hyperparameters such as minimum leaf size, leaning cycles, and learning rate by using Bayesian optimization. A practical power transmission network of Chhattisgarh state is modeled/simulated in MATLAB software to employ the proposed fault location scheme. Exclusive datasets are provided by performing extensive simulation studies at a wide range of fault scenarios, thereby applying discrete wavelet transform as an explanatory signal processing technique. Further performance assessment is carried out by comparing different error metrics such as mean absolute error, mean absolute relative error, mean square error, and root mean square error. The simulation results confirm the applicability of the proposed scheme for fault location estimation, and it makes a research insight while designing relaying schemes to practical power transmission networks.