Hybrid Modeling and Intelligent MPPT of Photovoltaic Panels Using Artificial Neural Networks and Interpolation Techniques

Creating accurate mathematical models of photovoltaic (PV) panels remains a formidable barrier to optimizing the operation of renewable energy generation systems. This paper presents a sophisticated model which solves the problem of PV parameter identification through the application of artificial neural networks (ANNs) combined with model-to-model linear interpolation. The necessary ANNs for PV characteristic forecasting are trained with manufacturers’ data or experimental data gathered at various irradiance and temperature levels, which ensures accurate forecasting of current–voltage (I-V), power–voltage (P-V) curves, and the maximum power point (MPP) of the PV. Model accuracy is tested by analyzing the performance of the ANN to accurately generalize to unmeasured irradiance level interpolated between models while achieving a low mean relative error on the training data. In an effort to boost the intelligent control of the PV system, its integration into the intelligent MPPT (Maximum Power Point Tracking) systems, especially the ones developed on the backbone of recurrent neural networks (RNNs), is incorporated for real-time and agile control. The complete model along with the MPPT strategy was implemented and tested against real data collected from field-deployed sensors and validated in the MATLAB environment. The results demonstrate the potential of this approach for developing flexible, high-performance, and cost-effective simulation and validation platforms for autonomous and hybrid PV systems.