<p>This study investigates the problems associated with the nonlinear power–voltage characteristics of photovoltaic (PV) systems, especially under partial shading conditions (PSC), which reduce energy efficiency and tracking accuracy. To overcome these limitations, two improved maximum power point tracking (MPPT) controllers based on Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) techniques are proposed. The controllers are designed with a suggested architecture that uses the power-voltage derivative (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\:\frac{dV}{dt}\)</EquationSource> </InlineEquation>) and the voltage time derivative (<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\:\frac{dP}{dV}\)</EquationSource> </InlineEquation>) as input features, enabling predictive, non-iterative control. This approach eliminates the steady-state oscillations inherent in conventional perturb-and-observe (P&O) algorithms and achieves superior dynamic response under rapidly changing environmental conditions. Simulation results demonstrate significant improvements compared with the traditional P&O method. The proposed ANN and ANFIS controllers achieved average tracking efficiencies (<InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(\:{\eta\:}_{\text{avg}}\)</EquationSource> </InlineEquation>) of 99.4% and 99.75%, respectively, with a response time reduction of about 55% and steady-state oscillation suppression exceeding 70%. The ANFIS controller exhibited higher stability, reducing the duty-cycle fluctuation index (<InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(\:{\sigma\:}_{\text{DCy}}\)</EquationSource> </InlineEquation>) by approximately 20% compared with the ANN controller, resulting in smoother and more reliable power extraction. A comparative evaluation with recently published metaheuristic and hybrid AI-based MPPT approaches confirmed that the proposed ANFIS model achieves equal or better performance while maintaining very low computational complexity. The average execution time per control step remained below 0.2 ms, confirming the suitability of both controllers for real-time deployment on low-cost digital signal processors (DSPs). These findings demonstrate that the proposed intelligent MPPT framework provides a fast, accurate, and computationally efficient solution for improving the reliability and energy yield of PV systems operating under dynamic and partially shaded conditions.</p>

Energy optimization of PV systems under partial shading conditions using various technique-based MPPT methods

期刊论文
出版模式：

开放获取
发表日期：

2026年1月13日

查看全文

Naima Benabdallah,
Belkacem Belabbas,
Ahmed Tahri,
Riyadh Bouddou,
Ayodeji Olalekan Salau,
Anna Pinnarelli,
Alireza Soleimani

摘要

This study investigates the problems associated with the nonlinear power–voltage characteristics of photovoltaic (PV) systems, especially under partial shading conditions (PSC), which reduce energy efficiency and tracking accuracy. To overcome these limitations, two improved maximum power point tracking (MPPT) controllers based on Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) techniques are proposed. The controllers are designed with a suggested architecture that uses the power-voltage derivative ( \(\:\frac{dV}{dt}\) ) and the voltage time derivative ( \(\:\frac{dP}{dV}\) ) as input features, enabling predictive, non-iterative control. This approach eliminates the steady-state oscillations inherent in conventional perturb-and-observe (P&O) algorithms and achieves superior dynamic response under rapidly changing environmental conditions. Simulation results demonstrate significant improvements compared with the traditional P&O method. The proposed ANN and ANFIS controllers achieved average tracking efficiencies ( \(\:{\eta\:}_{\text{avg}}\) ) of 99.4% and 99.75%, respectively, with a response time reduction of about 55% and steady-state oscillation suppression exceeding 70%. The ANFIS controller exhibited higher stability, reducing the duty-cycle fluctuation index ( \(\:{\sigma\:}_{\text{DCy}}\) ) by approximately 20% compared with the ANN controller, resulting in smoother and more reliable power extraction. A comparative evaluation with recently published metaheuristic and hybrid AI-based MPPT approaches confirmed that the proposed ANFIS model achieves equal or better performance while maintaining very low computational complexity. The average execution time per control step remained below 0.2 ms, confirming the suitability of both controllers for real-time deployment on low-cost digital signal processors (DSPs). These findings demonstrate that the proposed intelligent MPPT framework provides a fast, accurate, and computationally efficient solution for improving the reliability and energy yield of PV systems operating under dynamic and partially shaded conditions.