Intelligent Hybrid MPPT Control Using ANFIS with Real-Data Training for a Photovoltaic System: A Case Study in Cucuta.

Abstract

An intelligent Maximum Power Point Tracking (MPPT) system based on an Adaptive Neuro-Fuzzy Inference System (ANFIS), trained using real irradiance and temperature data, is proposed and experimentally validated for a photovoltaic system operating under highly variable climatic conditions. Unlike most approaches reported in the literature, which rely on synthetic datasets and deterministic real-time platforms, this work integrates a double-diode photovoltaic model with field measurements acquired using a Class A pyranometer to construct a training dataset that captures the stochastic nature of tropical environments. The controller is implemented on a Raspberry Pi 5 running a Linux operating system and governs a 100 kHz DC–DC Boost converter connected to a 340 W photovoltaic array installed in Cúcuta, Colombia. The ANFIS estimates the optimal operating voltage based on irradiance and temperature input vectors. Experimental results demonstrate a tracking time of 100 ms compared to 240 ms achieved by the conventional Perturb and Observe (P&O) method, reaching a maximum steady-state efficiency of 96.4% and a dynamic efficiency of 94.5% under real fluctuating irradiance conditions. These results confirm that intelligent MPPT controllers trained with real data can operate robustly on low-cost, non-deterministic embedded platforms, enabling their deployment in practical photovoltaic microgrid applications.