Performance Analysis of An Intelligent Adaptive Rbfnn Based Novel Mmc-Dstatcom in Grid Connected Pv System

This paper presents a novel Modular Multilevel Converter–DSTATCOM topology used to integrate PV system into the AC Grid. Unlike the conventional MMC topologies, the proposed topology can be applied in high power applications requiring more voltage levels using less capacitor count maintaining higher efficiency. The common problem caused within a MMC by the impact of circulating current is eliminated here by adding it with the input current of each arm. Photovoltaic (PV) System generally uses Power electronic converters to convert the generated DC into AC by switching ON and OFF of the Power electronic devices. This leads to the introduction of Harmonics into the power system causing deterioration in the power quality resulting conductors heating, malfunctioning of fuses, circuit breakers and relays. This paper introduces an Adaptive Radial Basis Function Neural Network (RBFNN) based on the Synchronous Reference Frame (SRF) theory which is capable of generating the compensating reference current needed to extract the total harmonics from the system. It uses a deep learning filtering algorithm based on a hybrid learning method, which is regarded as the computationally efficient training method with less complexity and negligible training time. Extensive simulations are carried out in MATLAB software and its performance in each aspect is analyzed by comparing it with conventional methods. The results obtained show the usefulness of the novel topology and the proposed algorithm in improving power quality.