Grid Integration of PV Systems With Advanced Control and Machine Learning Strategies

In the pursuit of sustainable and efficient energy solutions, Photovoltaic (PV) systems have emerged as a prominent player in the domain of renewable energy generation. Particularly, grid-tied PV systems have gained substantial attention due to their potential to contribute to stability and reliability of existing power grid infrastructure. Accordingly, an innovative approach to enhance grid supply using PV systems with Machine Learning Strategy is proposed in this research. The primary objective is to optimize voltage output from PV system while concurrently maximizing power using a novel Modified Zeta-Cuk converter, coupled with Hybrid Maximum Power Point Tracking (MPPT) algorithm combining Incremental Conductance and Bat Optimization Algorithm (InC-BOA). The stabilized DC link resulting from this process is directed to a 3-phase Voltage Source Inverter (VSI) to facilitate conversion of DC supply to AC. To further improve the efficiency and accuracy of system, current produced by inverter is subjected to Discrete Wavelet Transform (DWT) analysis followed by Principal Component Analysis (PCA) for feature extraction. The final step involves implementation of Recurrent Neural Network (RNN) controller, enabling the generation of a refined reference current. The generated reference current is then compared with actual current using Hysteresis Current Controller (HCC). This comparison yields an output which is subsequently employed to Pulse Width Modulation (PWM) generator facilitating the achievement of effective grid synchronization, enhancing overall performance and stability of the system. The validation is performed using MATLAB Simulink software and the outcomes reveals the dominance of proposed work.