Cyber–physical machine-learning-based grid-fault detection and location scheme for distance protection of MV distribution lines

In this paper, distance protection (DP) using advanced machine learning (ML) techniques, namely interval type-2 fuzzy logic (IT2FL), is proposed as a suitable solution for the mentioned problems. This paper proposes a novel method for classifying grid faults and delivering fault locations on 20 kV power distribution lines. This paper presents a fault location technique that improves time response and reduces power loss by avoiding impedance-based zoning and allowing for acceptable error. The proposed plan can be customized for different medium voltage (MV) power distribution lines with minimal changes, allowing for rapid feedback through specific inputs. Time-domain digital simulations based on a detailed model of an overhead power distribution line and the parameters of a real-life practical 20 kV line are performed in a MATLAB/Simulink software environment, and several tests are executed based on different scenarios and compared with previously reported techniques. The results prove the proposed method’s effectiveness, authenticity, selectivity, accuracy, precision, and superiority over other reported strategies. • A novel IT2FLC-based scheme is proposed for MV grid fault detection and location. • The method avoids impedance zones to improve fault location accuracy and speed. • The model adapts to various MV lines with minimal structural modifications. • Time-domain simulations are based on a detailed real-life 20 kV distribution line. • The proposed method outperforms existing techniques in accuracy and robustness.