“DETECTION, CLASSIFICATION AND MITIGATION OF POWER QUALITY DISTURBANCE: A CASE STUDY ON THE 15KV DISTRIBUTION FEEDER 6(R4-G5) AT HAWASSA SUBSTATION”

Abstract

Modern power systems face several difficulties due to power quality disturbances, such as voltage sags and swells, which call for effective detection, classification, and mitigation techniques. In order to fully address these problems, this study offers an integrated strategy that combines modern machine learning methods with optimization techniques. Artificial Neural Networks (ANN) trained and refined with MATLAB's Classification Learner Toolbox are used for detection and classification after features are extracted from voltage/current signals. This thesis was carried out on one of the Hawassa Feeder-6 (R4-G5) 15 kV distribution feeders utilizing distribution network analysis and MATLAB simulation. The purpose of power flow analysis is to ascertain the active and reactive power flows on the distribution lines, as well as the voltage magnitude and phase angle at each bus (node) in the system. Power quality disturbances are divided into four distinct wavelet filter levels using Debechesh-4 (Db-4). This enables the improvement of an approximate and detailed coefficient distribution in addition to the extraction of features such as the mean, maximum, and lowest values of the disturbances for power quality disruptions. The classification efficacy of neural networks (ANN) and support vector machines (SVMs) is 100%. Finally, a dynamic voltage restorer (DVR) is positioned optimally using the Grasshopper Optimization (GOA) techniques. Power loss decreased from 1913.3 kW (active) and 1202.4 kVAR (reactive) to 295.534 kW and 261.803 kVAR when GOA was used, and the voltage profile was increased from 70% to 98.5% and lowered the voltage swell from 110% to 98%, and also, by applying different kinds of faults, easily tested the voltage sag and swell by using DVR integrated with wavelet transforms algorithm. This reduces the possible influence on delicate systems and equipment while simultaneously enhancing the power supply's quality