Dynamic Pulsed Load Mitigation in PV-Battery-Supercapacitor Systems: A Hybrid PI-NN Controller Approach

DC microgrids are becoming more and more popular, however there are still difficulties involving ongoing instability that are mostly caused by imbalances in energy supply and demand, particularly in situations with pulse load or variable pulse load patterns. Innovative solutions are required since these disruptive load dynamics significantly decrease the effectiveness of conventional PV-Battery control systems. conventional Proportional-Integral (PI) controllers, although widely employed, may encounter limitations when confronted with the unique challenges posed by specific load patterns, notably pulse load scenarios. These dynamic load profiles, characterized by rapid and unpredictable changes. The inherent non-linearity and variability in such load conditions can result in suboptimal control performance, potentially leading to voltage instability and transients within the microgrid. To overcome these issues and improve control performance, the introduction of Neural Network (NN) controllers promises to be a potential approach. NN controllers have a remarkable capacity to learn and adapt from data, making them particularly well-suited for situations characterized by non-linear and dynamic load behavior. In this study, a hybrid PI-NN (Proportional-Integral Neural Network) controller is introduced to reduce load effects, especially pulsed loads, in PV systems with batteries and supercapacitors. This method efficiently optimizes energy storage. For voltage control, it utilizes a PI system, which divides the reference current into low and high-frequency components for battery and supercapacitor control. For enhanced sensitivity and dynamic responsiveness, neural network control replaces conventional PIs. The performance of the hybrid PI-NN controller is evaluated in this work, with an emphasis on its potential to improve PV system efficiency in severe load circumstances.