Harmonics Compensation of Grid-Connected PV Systems Using a Novel M5P Model Tree Control

This paper proposes a new fast machine learning approach based on a mixed control of fuzzy logic controller designed under MATLAB and a pruned M5P model tree technique designed under WEKA software. An investigation of an innovative maximum power point (MPPT) method and equivalent dc-link controller for a grid-connected PV system was undertaken to evaluate the potential performances in both dynamic and steady operating conditions. In addition, an optional power quality enhancement can be benefited from integrating the shunt active power filter functionality (SAPF). This topology allows the system to be more effective, which means that the combined system can perform dual functions of harmonic mitigation and grid-connected photovoltaic system. Obtained results of simulations and real-time implementations indicate that the developed tree algorithm based on simple MATLAB code has clearly revealed significant role to replace controllers suffer from modeling complexity and a long hardware implementation time. According to the harmonic standard IEEE 519–1992 limits, the novel control would result in very low total harmonic distortion of the grid current and a successful power transaction along with near unity power factor.