Maximum power point tracking enhancement for PV in microgrids systems using dual artificial neural networks to estimate solar irradiance and temperature

• Maximum power point tracking based on artificial nural network. • PV modeling. • Solar irradiance estimation. • Temperature estimation. • Microgrid based on PV and hybrid energy storage system. This paper presents an artificial neural network-based maximum power point tracking (MPPT) method. Where dual ANNs predict solar irradiance and temperature. Next, an adaptive computation block determines the maximum power point (MPP). The proposed MPPT method stabilizes output power at the MPP, unlike traditional MPPT methods, which oscillate. The suggested tracker does not require expensive solar irradiance and multiple temperature sensors, unlike trackers that do. Additionally, the MPPT can handle rapid solar irradiance changes. The ANNs integrated with the adaptive block were evaluated to validate the system's capacity to estimate solar irradiance, temperature, and maximum power point. The solar system was integrated into a Microgrid and subsequently simulated using MATLAB/Simulink to evaluate the robustness of the suggested approach under steady-state and during sudden changes in solar irradiance and temperature. The proposed solar system achieves a steady state within 30 ms following a sudden solar irradiance or temperature change. The method achieves pest estimate at 1000 W/m 2 and 45 °C with an efficiency of 99.9 %. Additionally, at 1000 W/m 2 and 25 °C, the system achieves 99.8 % efficiency. Under the worst-case scenario, the proposed system demonstrates a relative error of approximately 6.1 % when estimating solar irradiance at 600 W/m 2 and 45 °C and a relative error of around 21.1 % when estimating temperature at 700 W/m 2 and 25 °C. Nevertheless, the efficiency under the worst-case scenario is 94.2 % when the power density is 500 W/m 2 and the temperature is 45 °C.