Performance evaluation of solar PV integrated with custom power device under various load conditions

Abstract

The non-linear properties and rapid switching of power electronic equipment are the primary causes of power quality issues in power systems, particularly in the power distribution systems. The widespread use of delicate equipment, which continuously pollutes the environment, is making power quality problems worse. The increasing integration of renewable energy sources into the generation mix and the decarburization of the economy have created new challenges for smart grid technology, requiring creative solutions such as energy storage systems and smart transformers. This article describes a solar photovoltaic integrated Unified power quality conditioner (UPQC) that uses a deep learning method based on neural networks and a novel compensating technique. Here, two Deep Neural Network algorithms are used, one in the solar PV system to obtain the maximum power under various irradiance situations, and the other to manage the UPQC under various load conditions. When compared to the conventional UPQC based on PQ Theory, DNN-UPQC produces superior results in terms of reducing total harmonic distortion. This iterative strategy, which is focused on soft computing, provides faster convergence to the target condition while maintaining the updating weight within a predetermined limit. PV-based UPQC has been connected individually to reduce voltage sag, swell, and unbalance in variable load conditions. The system's dynamic and steady state performance are assessed by modelling it with a MATLAB-Simulink in different load condition.