A maximum power point tracking control system leveraging artificial neural network for photovoltaic systems based on a Raspberry Pi embedded under various weather conditions

Introduction. Maximum power point tracking (MPPT) is essential for maximizing the energy harvested from photovoltaic (PV) modules, especially under fast irradiance and temperature variations. Problem. Conventional MPPT methods like perturb and observe and incremental conductance suffer from slow convergence, steady-state oscillations, and reduced tracking accuracy when operating under fast and unpredictable environmental variations. Goal. To develop a fast, accurate, and robust MPPT strategy capable of improving power extraction efficiency and reducing power fluctuations under realistic and dynamic meteorological conditions. Methodology. This paper proposes an artificial neural network (ANN)-based MPPT scheme that uses two estimated reference inputs – reference current (Iref) and reference voltage (Vref) to directly update the duty cycle of a DC-DC boost converter. The algorithm is designed using LabVIEW for supervision and co-simulation and is implemented in real time on a Raspberry Pi 4 using Python to control the converter switch. Experimental validation is carried out using a laboratory prototype with measured meteorological data. Results. Experimental results show that the proposed ANN-based MPPT achieves superior performance compared to fuzzy logic control and particle swarm optimization techniques. The system reaches MPPT efficiencies of up to 99.5 %, exhibits faster transient response, and significantly reduces power oscillations around the MPP. Scientific novelty. The novelty of this work lies in the direct duty-cycle control using ANN with dual reference inputs (Iref and Vref), combined with real-time deployment on a low-cost embedded Raspberry Pi platform, bridging the gap between data-driven MPPT algorithms and practical hardware implementation. Practical value. The findings demonstrate that low-cost embedded platforms can deliver accurate and robust MPPT performance for PV energy conversion under realistic weather conditions, and confirm the efficacy of the recommended Raspberry Pi system, which is programmed with the proposed data-driven ANN-MPPT, in optimizing the power output from the panel across varying meteorological conditions. References 20, tables 4, figures 11.