Development of a Hybrid Adaptive Neuro-Fuzzy and Artificial Bee Colony–Cuckoo Search Optimization Model for Enhanced Fault Detection, Isolation, and Protective Coordination in Smart Distribution Networks with Distributed Energy Resources
Abstract
The growing deployment of distributed energy resources is changing fault behaviour in radial distribution networks. Bidirectional power flow, inverter-limited fault currents and changing feeder configurations can undermine fixed overcurrent-relay settings, compromise primary–backup selectivity and delay fault isolation. This study develops an integrated adaptive neuro-fuzzy inference system–artificial bee colony–cuckoo search (ANFIS–ABC–CS) framework for fault identification, selective isolation, relay coordination and controlled restoration. The ANFIS module interprets voltage, current, symmetrical-component and network-status measurements to estimate fault state and likely protection zone. A hybrid ABC–CS search then identifies feasible relay-setting groups and restoration-switching decisions. Artificial bee colony search explores candidate settings, while cuckoo-search Lévy flights provide refinement and diversification. The design explicitly enforces directional sensitivity, coordination-time intervals, voltage limits, thermal ratings, switching interlocks and network-topology constraints. Autonomous protection and validated fallback settings are retained whenever the supervisory algorithm is uncertain or unavailable. The IEEE 33-bus, 12.66-kV feeder provides the proposed benchmark, augmented with photovoltaic, wind and battery-storage scenarios and an explicit protection model. A numerical demonstration based on 1,200 assumed fault and healthy-state observations yields a classification accuracy of 97.8% for ANFIS–ABC–CS and 91.0% for conventional logic. In the same example, mean fault-isolation time changes from 1.85 to 0.92 s and mean restoration time from 12.6 to 7.2 min. At full DER output, the lowest primary–backup coordination interval improves from 0.10 s under fixed settings to 0.32 s under adaptive setting selection, against a 0.30-s design threshold. These values are scenario-based numerical examples, not measured or executed simulation findings. The proposed study establishes mathematical models, validation requirements and controlled comparisons needed to determine whether the integrated approach offers meaningful gains under DER variability, uncertain faults, communications disruptions and changing network configurations. IJASMT E- ISSN 2489-009X ,
Keywords
References
More Articles from INTERNATIONAL JOURNAL OF APPLIED SCIENCES AND MATHEMATICAL THEORY
Author: Amala Isabella Smart, Amos Emeka, Davis Iyai, Anthony I. Wegbom, Da-Wariboko, Yvonne Asikiye
Author: Ganiyu Adedayo Ajenikoko, Andrew Oyekanmi Oke, Abdulbasit Oladayo Ajenikoko, Ayobami Olatunde Fawole, Muhammed Akintola Akinola, Kazeem Oyewale, Okunlola, Corresponding Author
Author: Ibrahim Jibrin
Author: Opara, P. N, Prof, S. I Onyeagu, Prof, G. A Osuji
