Evaluating the Maximum Region for Parametric Model Uncertainties in Variable-Oriented Photovoltaic Systems

Renewable energy has been a worldwide subject of interest in the last decades, because of the need to reduce greenhouse gas emissions. Solar is among the most widespread energy resources as it has almost limitless installation possibilities on Earth’s surface. Photovoltaic panels play a key role in the energy industry, having a dynamically evolving chemical structure of the cell and being implemented in many applications within various weather conditions. This paper will focus on introducing and analyzing an optimal control strategy for variable-oriented photovoltaic systems to address the issue of suboptimal generation. In contrast with the traditional Maximum Power Point Tracking (MPPT) approaches, a stability analysis will be performed and parametric model uncertainties will be introduced to emulate the real-world variable conditions, such as temperature and irradiance changes. These uncertainties will help determine the maximum region for which the control strategy still keeps the best performances, regarding reference tracking and regulation. The novelty of the paper comes from introducing in the photovoltaic control and optimization field, the approach of delimiting the parametric uncertainty margins, for which a designed Linear Quadratic Regulator (LQR) controller will keep the stability and performances, which was not previously addressed in this field. Accordingly, it will be shown that a robust controller can guarantee maximum power at output even though the model of the system evolved influenced by environmental factors, degradation of the system’s components or another system’s abnormal behavior. In this paper, it will be also introduced a series of simulation results, that will emphasize the stabilized system evolution for various uncertainties within the computed maximum stability margins. The simulation findings indicate that the implemented robust LQR will offer similar response time, without oscillations and, implicitly, similar performances for the entire family of parametric uncertainties added to the initial system. In other words, the approach can offer a cost-effective solution for sustainable large-scale energy production.