Enhanced PV power optimization: integrating Jordan neural network MPPT with FOPID-controlled SEPIC converter for non-linear load applications

Abstract

In photovoltaic (PV) systems, maximum power point tracking (MPPT) technologies are used to constantly maximize PV output power, which is primarily governed by solar radiation and cell temperature. However, the typical MPPT approach wastes a significant amount of energy, and the efficiency is cumbersome and unstable. To solve these limitations, the Jordan neural network (JNN) MPPT with FOPID-SEPIC converter is employed in this research. As a result, the PV and non-linear load were constructed, and three scenarios were tested to validate the proposed PV design: normal, static, and dynamic. At first, the PV model with a non-linear load was designed for a certain range. This specially designed model is used to collect voltage, temperature, power, current, and irradiance under a variety of situations. The acquired parameters were given to JNN MPPT, which was specifically designed for maximum PV power tracking. The FOPID obtained the error value for JNN output and PV generator power. The FOPID consists of five parameters that are optimally chosen using brown bear optimization to produce a better process. FOPID generates a pulse signal to the SEPIC convertor, which powers the non-linear load after figuring out the optimal value. Consequently, the observed error of the JNN is 0.0033%, the accuracy rate is 0.99%, and the false positive rate (FPR) is 0.04%. The suggested JNN MPPT model functioned well in comparison to alternative strategies, resulting in appropriate implementation in actual tracking ways.