Deep Learning Based Controller for Parallel DSTATCOM to Improve Power Quality in Distribution System

Modern utility systems were being heavily strained by rising energy consumption and dynamic load variations, which had an impact on the quality and reliability of the supply. Harmonic injection and reactive power imbalance were caused by the widespread divergence. Power quality (PQ) issues were mostly caused by renewable energy powered by power electronic converters that were integrated into the utility grid. Despite the fact that a range of industries required high-quality power to function properly at all times. Several solutions were frequently created, but continuing effort and newly improved solutions were needed to solve these problems by operating to various international standards. Distributed Static Compensator (DSTATCOM) was created in the proposed model to enhance PQ in a standard bus system. A standard bus system using the DSTATCOM model was initially developed. A real-time dataset was gathered while applying various PQ disturbance conditions. A deep learning controller was created using this generated dataset, which examined the bus voltages to generate the DSTATCOM pulse signal. Two case studies such as the IEEE 13 bus and the IEEE 33 bus system, were used to analyse the proposed work. Performance of the proposed deep learning controller was verified in various situations including interruption, swell, harmonics, and sag. The outcome of THD in IEEE 13 bus is 0.09% at the sag period, 0.08% at the swell period, 0.01% at the interruption period, and in IEEE 33 bus have 1.99% at the sag period, 0.44% at the swell period, 0.01% at interruption period. Also, the effectiveness of the proposed deep learning controller was examined and contrasted with current methods like K-Nearest Neighbour (KNN) and Feed Forward Neural Network (FFNN). The validated results show that the suggested method provides an efficient mitigation mechanism, making it suitable for all cases involving PQ issues.