Optimizing Solar Photovoltaic Power Injection Using Machine Learning: Minimizing Harmonic Distortion Under Inertia Constraints

This paper elaborates on a machine learningbased technique for enhancing the Space Vector Pulse Width Modulation (SVPWM) for minimizing Total Harmonic Distortion (THD) in solar photovoltaic (PV) systems under a limitation of power system inertia. Since solar PV systems have become more prevalent in today's power systems, achieving a quality and stable power supply is a critical issue since PV systems do not inherently possess inertia. As the experimental data were not available, a synthetic dataset was generated based on the following key operating parameters of the converter: DC voltage, load conditions, switching frequency, duty cycle, modulation index, power factor, and temperature. This research used two machine learning algorithms, Random Forest (RF) and Decision Tree Regressors, to predict the Total Harmonic Distortion (THD) and to enhance the Space Vector Pulse Width Modulation (SVPWM). The RF model produced an MSE of 0.0006 and a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex> score of 0.9642 and so is superior to the Decision Tree model. The results therefore help to give more understanding of how to increase the efficiency of PV systems and minimize harmonic distortions for better integration into the grid.