Nonlinear MPPT optimization for PV system via RNN and condition-based super-twisting sliding mode controller

As the demand for energy rises and concerns grow over dwindling fossil fuel reserves and carbon emissions, it is increasingly critical to develop sustainable and efficient energy solutions. Solar energy has captured significant interest within the renewable energy sector, given its vast potential. However, the efficiency of photovoltaic (PV) systems remains heavily influenced by changing weather and load conditions, which present challenges to consistently harvesting maximum energy. To address this, maximum power point tracking (MPPT) techniques are widely applied to optimize PV output. Since PV systems exhibit nonlinear characteristics, advanced nonlinear control methods are well-suited for MPPT applications. Traditionally, popular methods for MPPT include Perturb and Observe (P&O). Recent advancements in artificial intelligence, have led to the adoption of controllers such as Artificial Neural Networks (ANN) to enhance MPPT. However, this study introduces a Recurrent Neural Network (RNN) controller, leveraging its memory-based architecture to adaptively track the PV system’s power reference point under fluctuating environmental conditions. To ensure robust MPPT performance, this research integrates the RNN controller with a condition-based super-twisting sliding mode control (C-STSMC) strategy, aiming to manage nonlinearities and enhance resilience to changing climatic factors. This paper details the design and implementation of the proposed RNN-based MPPT controller, alongside comparisons with traditional MPPT method P&O and Fuzzy logic controller. Experimental results for hardware in loop (HIL SETUP), confirm the real-time capabilities and enhanced accuracy of the proposed controllers. This comparative analysis demonstrates the improved precision and reliability of the RNN-based MPPT method in optimizing PV efficiency under dynamic conditions.