Artificial neural network-based approach for optimized photovoltaic grid injection using a three-phase multilevel inverter

The growing global demand for renewable energy has positioned photovoltaic (PV) systems as a key component of modern power generation. However, efficient grid integration of PV systems remains challenging due to issues such as harmonic distortion, poor power quality, and grid instability. This study presents an artificial neural network (ANN)-based PV grid-injection system designed to enhance energy conversion efficiency and ensure stable grid interaction under varying environmental conditions. The proposed ANN model is trained to accurately predict the maximum power point (MPP) of the PV array under fluctuations in solar irradiance and temperature. In addition, the ANN regulates the switching signals of a multilevel inverter to achieve optimal voltage control and seamless grid synchronization, ensuring efficient and stable power injection with minimal fluctuations. Simulation results demonstrate significant performance improvements. The PV and DC–DC boost converter output voltage is increased from 350 V DC to 800 V DC while reducing its output current, thereby minimizing power losses. Furthermore, the three-phase multilevel inverter with filtering produces a near-pure sinusoidal waveform with a 200 kW output, significantly reducing total harmonic distortion (THD) and improving the overall power quality. The proposed ANN-based control strategy enhances system robustness, improves grid power quality, increases energy conversion efficiency, and reduces stress on grid infrastructure, thereby supporting reliable and sustainable PV integration into the electrical grid.