A Human-on-the-Loop Cyber-Physical Framework for Fault-Aware Service Restoration in Nigerian Distribution Feeders

Abstract

Distribution fault management depends on more than fault recognition. An operator or automation controller must establish the affected section, confirm switching-device states, find a feasible post-fault topology, respect protection and equipment limits, and verify that service has returned without creating a new hazard. This paper proposes a human-on-the-loop cyber-physical architecture for distribution feeders in Nigeria, with its requirements grounded in the author’s industrial training at the Enugu Electricity Distribution Company (EEDC)/MainPower. The training report records field exposure to right-of-way hazards, sagging conductors, transformer load assessment and commissioning, fuse-rating checks, relay work, insulation-resistance tests, earthing tests, power-quality inspection, and materials control. These activities motivate a design in which local protection remains independent of machine-learning inference and remote communications, while software proposes only switching actions that pass explicit electrical, topology, state-freshness, and authorization checks. The paper reconciles prior drafts against verified literature and separates published findings from author-proposed methods and claims that lack reproducible evidence. It formulates a constrained feeder-restoration problem, describes a candidate waveform classifier and policy-learning layer, and defines reliability, power-quality, latency, safety, cybersecurity, and economic evaluation measures. The proposed research has not yet been validated on an EEDC feeder; therefore, no numerical improvement in restoration time, SAIDI, SAIFI, or line losses is claimed. A reproducible evaluation plan using OpenDSS and a documented benchmark network is specified before any utility-specific calibration or field pilot. The result is a research framework and validation protocol, not a report of completed field or simulation experiments.