Electric Fish Optimizer-Based Hyperparameter Tuning of MLP for Fault Diagnosis in Grid-Connected Photovoltaic Systems

• Muhammed A. ADAWUDI

  Author

• Danlami MALIKI

  Author

• Ibrahim M. ABDULLAHI

  Author

This paper presents an optimized multilayer perceptron (MLP) model for fault diagnosis in grid-connected photovoltaic (PV) systems using the Electric Fish Optimizer (EFO). A real-world dataset comprising five fault classes was preprocessed and used to train a baseline MLP model. The EFO algorithm was employed to optimize key hyperparameters of the network, and a systematic random search was further conducted to determine optimal EFO configurations. The optimized model achieved an accuracy of 99.06%, improving upon the baseline accuracy of 98.63%, with enhanced per-class F1-scores across all fault categories. The results demonstrate that EFO is an effective tool for improving neural network performance in PV fault diagnosis, particularly when applied to large and imbalanced datasets.

• FJET_21_53_53

Copyright (c) 2026 FUDMA Journal of Engineering and Technology

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.