Izvorni znanstveni članak

https://doi.org/10.37798/2025741710

Accurate Photovoltaic Power Forecasting in 5GNetworks: A Novel Neural Network Approach

Mohammed Moyed Ahmed orcid.org/0000-0002-1743-0352 ; Jawaharlal Nehru Technological University Hyderabad, Hyderabad, India *

* Dopisni autor.

Sažetak

This study addresses the challenge of integrating photovoltaic (PV) power generation into 5G base stations to reduce energy consumption and promote sustainable energy integration in telecommunications infrastructure. A novel Improved Firefly Algorithm-Back Propagation (IFA-BP) neural network model is proposed for enhanced PV power prediction accuracy and reliability. The methodology combines Circle chaos mapping for optimized population initialization with nonlinear mutational perturbation to strengthen global search capabilities and improve convergence rates. Critical input parameters are systematically selected through grey correlation analysis to optimize model efficiency and reduce computational overhead. Comprehensive comparative analysis with conventional BP and FA-BP models is conducted using historical operational data from 5G base station installations across varying weather conditions. Experimental results demonstrate the model’s superior performance and statistical robustness, achieving a Mean Absolute Percentage Error (MAPE) of 4.79 ± 0.31% and coefficient of determination (R2) of 0.9895 ± 0.0012 under sunny conditions, while maintaining exceptional weather adaptability with a MAPE of 12.20 ± 0.87% and R2 of 0.9793 ± 0.0019 during cloudy weather. Statistical significance testing confirms these improvements are not due to random variation (p < 0.001). The proposed IFA-BP model demonstrates remarkable resilience in challenging weather conditions and provides a robust foundation for intelligent power management in next-generation wireless networks. However, the current evaluation is limited to two-day testing data and would benefit from extended validation across diverse seasonal variations and broader environmental conditions to establish comprehensive generalizability for practical deployment in real-time power management systems.

Ključne riječi

5G Base Station; Photovoltaic Power Prediction; Improved Firefly Algorithm

Hrčak ID:

338955

URI

https://hrcak.srce.hr/338955

Datum izdavanja:

1.3.2025.

Posjeta: 211 *