Sistemas de Seguimento de Máxima Potência do tipo Current Sensorless para Sistemas Fotovoltaicos em Condições de Sombreamento Parcial

Abstract

The impacts related to global warming are becoming increasingly frequent. Countries around the globe are seeking solutions to contain the rise in Earth's temperature. The forecast predicts a considerable increase in the demand for electricity, driven by the transportation sector and electronic devices, exacerbating the problem. In this context, photovoltaic systems have emerged as a key solution to meet the energy demand in a clean and sustainable manner, with an increase in installed capacity around 345 Gigawatts (GW) worldwide in 2023. For the maximum extraction of the available energy from photovoltaic systems, it is desirable to always operate the generation at the maximum power point. To achieve this, it is necessary to use maximum power point tracking (MPPT) algorithms, especially in situations with partial shading when the system has multiple local maximum power points (LMPPs), which complicates the task since conventional algorithms like Perturb and Observe (P&O) usually converge to the local maxima. In this sense, techniques based on artificial intelligence, metaheuristics, and hybrid approaches are presented in the literature with the aim of developing systems capable of finding the global maximum power point (GMPP). Another point worth highlighting is that, typically, regardless of the method used, there is a dependency on current and voltage sensors for the algorithms to function. Therefore, this work aims to address intelligent algorithms for tracking the GMPP in photovoltaic systems with partial shading, using only a voltage sensor, aiming to reduce costs and increase reliability. The proposed method will be compared with the conventional one that uses both sensors, and metrics such as GMPP search time, steady-state power oscillation, and tracking efficiency will be addressed for simulation and experimental results. Finally, it is noted that the implemented sensorless algorithms can maintain efficiency very close to, or even superior to, the methods that use both sensors, with particular emphasis on the Particle Swarm Optimization (PSO) method with a sensorless methodology, which experimentally achieved a Tracking Factor (TF) of 99.08%, compared to conventional PSO with a TF of 99.03%, considering the power profile analyzed in this work.