Harmonics Reduction for Four-Leg Distribution Network-Connected Single Phase Transformerless PV Inverter System Using Diagonal Recurrent Neural Network

摘要

This paper presents an analysis of the effects of solar irradiance on a single phase PV inverter system connected to a four-leg distribution system (FLDN). The proposed method aims to reduce current harmonic distortion (THDi) in the distribution system due to different and fluctuating irradiation. For this purpose, PV system with a capacity of 3.5 kW is connected to FLDN low voltage through an inverter voltage source (VSI) using intelligent control. This intelligent controller consist of: MPPT controller, constructive diagonal recurrent neural network controller (CDRNN), and synchronization and DQ transformation. The MPPT controller system automatically varies the voltage reference signal to obtain DC voltage, which extracts the maximum power from the PV string, while the DQ transformation block is used to convert the AC inverter output voltage and current into the synchronous reference frame. All of these signals are processed by the CDRNN controller to produce a PWM. Based on the simulation results using Matlab/Simulink, the THDi of FLDN was reduced from 5.09% into 4.63% in descent solar irradiance and 5.43% become 4.98% in ascend irradiance condition.