This paper introduces a novel control algorithm leveraging artificial intelligence to address the key defects of Direct Power Control (DPC) via Grid Voltage Modulation (GVM) strategy enhanced by Neural Network Control (NNC) for a three-phase inverter in a photovoltaic generation system. Conventional DPC-GVM techniques face major constraints due to the need for accurate tuning of gains (KP and Ki) in three PI controls, which affects the system's robustness, reliability, and stability. Besides, these conventional techniques suffer from incomplete decoupling between active and reactive powers, direct interdependence between the DC-link voltage and active power reference, elevated harmonic distortion, and suboptimal transient response. This paper introduces a progressive NNC-based algorithm, termed GVM-NNC (Grid Voltage Modulated-Neural Network Control), to overcome these issues. The proposed strategy effectively decouples the active and reactive power control, mitigates the dependence between DC-link voltage and active power reference, diminishes harmonic distortion, and improves transient response. The innovations and contributions of the GVM-NNC strategy include decoupling of active and reactive power for improved control precision, robust adaptability to sudden changes and external disturbances guaranteeing excellent dynamic response, enhanced stability and reliability by eliminating the need for precise tuning of PI controller gains, reduced harmonic distortion for cleaner power output, where the proposed algorithm reduces the THD until 0.98% compared to the THD of conventional DPC-GVM 1.20%. Numerical simulations executed in MATLAB Simulink demonstrate that the GVM-NNC method achieves superior comportment in each steady-state and transient state compared to the conventional DPC-GVM strategy.

The generation of variable energy from photovoltaic (PV) is significantly affected by unpredictable fluctuations due to weather changes. Additionally, the variability in load demand is a critical consideration for microgrid. Consequently, the implementation of an energy storage system is essential to address these challenges. This study presents a novel energy management technique (EMT) for hybrid energy storage systems (HESS). The innovative approach incorporates a low-pass filter (LPF) to optimize power distribution between batteries and supercapacitors (SC), thereby enhancing system stability and prolonging battery life. The proposed LPF-based EMT facilitates optimal power allocation, improves system stability by effectively filtering high-frequency power fluctuations, and extends battery life through reduced stress and optimized charge-discharge cycles. It developed a comprehensive system model to enable accurate simulation and analysis, supported by rigorous experimental validation that demonstrates the effectiveness of our method. This approach successfully redirects high-frequency power demands to the SC, stabilizing the DC link voltage. Comprehensive simulations indicate the system's capability, revealing quantitative improvements in battery performance and efficiency across various LPF time constants τ, representing a significant advancement in renewable energy control. The simulation results confirm that the proposed architecture and system representation achieve optimal DC link voltage stability. Furthermore, increasing the τ reduces the state of charge (SOC) of the battery, contributing to an overall increase in battery lifespan.