Comparative Analysis of Artificial Neural Network and Adaptive Neuro-Fuzzy Inference System Techniques for Fault Detection and Mitigation on the Nigerian 330 kV Transmission Network

Transmission-line protection in modern power networks faces growing challenges from high-impedance faults, current-transformer saturation and power swings that degrade the performance of settings-based conventional distance relays. This paper reports a comparative simulation study of Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS)-derived intelligent protection techniques, benchmarked against Deep Learning (DL) and Genetic-Algorithm (GA) optimized variants, for fault detection, classification and mitigation on the 330 kV Onitsha–Enugu transmission corridor in Nigeria. A distributed-parameter, twenty-section π-model of the 250 km line, coupled with a ±150 MVAr Static Synchronous Compensator (STATCOM) under fuzzy-logic control, was developed in MATLAB/Simulink. Line-to-ground, line-to-line, double line-to-ground and three-phase faults were simulated at varying fault resistances, inception angles and locations to build a training and test dataset. A Multi-Layer Feed-Forward ANN trained with the Levenberg–Marquardt algorithm achieved 96.0% fault-classification accuracy with a fault-location error below 2.0%, while GA optimization raised classification accuracy to 97.2%. The fuzzy-logic-controlled STATCOM reduced voltage-recovery time from 0.25 s to 0.1 s (a 60% improvement) and limited the peak fault-current surge by 74%. The results confirm that ANN- and ANFIS-derived architectures markedly outperform conventional threshold-based protection and reveal an accuracy–interpretability trade-off that motivates hybrid intelligent-relay deployment.