Authors:
Chibuzo V. Ikwuagwu, Ikechukwu E. Okoh, Nwachukwu Eme-Okafor

Addresses:
Department of Mechanical Engineering, University of Nigeria, Nsukka, Enugu, Nigeria. Department of Mechanical Engineering, African Centre of Excellence for Sustainable Sower and Energy Development, University of Nigeria, Nsukka, Enugu, Nigeria. Department of Mechanical Engineering, Michigan Technological University, Houghton, Michigan, United States of America. 

Abstract:

Maximum Power Point Tracking, often known as MPPT, is an essential component for enhancing the efficiency of photovoltaic (PV) systems that are operating in environments that are subject to fast environmental change. Perturb-and-Observe (P&O) and Incremental Conductance (IC) are two examples of conventional maximum power point tracking (MPPT) algorithms. These algorithms often exhibit delayed convergence, oscillations around the maximum power point, and reduced performance under partial shading. Using deep learning, particularly Long Short-Term Memory (LSTM) networks, this study examines the potential for predictive Maximum Power Point Tracking (MPPT) control in solar photovoltaic (PV) systems. The suggested method uses past irradiance and temperature data to predict the optimal MPP reference voltage. This allows for tracking that is both steadier and more rapid. A data-driven modelling framework is built, and the predictive performance of the LSTM model is assessed using data representative of PV operating conditions. The findings indicate that LSTM-based predictive control is suitable for next-generation intelligent photovoltaic (PV) systems, achieving enhanced tracking accuracy, lower steady-state error, and higher MPPT efficiency compared to conventional techniques.

Keywords: Solar Photovoltaic Systems; LSTM Networks; Predictive Control; Renewable Energy; Solar Energy; Partial Shading; Tracking Accuracy; Incremental Conductance.

Received on: 12/12/2024, Revised on: 15/02/2025, Accepted on: 04/05/2025, Published on: 16/12/2025

DOI: 10.69888/FTSESS.2025.000555

FMDB Transactions on Sustainable Environmental Sciences, 2025 Vol. 2 No. 4, Pages: 200-209

• Views : 50
• Downloads : 7

Download PDF