IDENTIFICATION, CLASSIFICATION AND MITIGATION OF POWER QUALITY ISSUES IN DISTRIBUTION NETWORK USING STOCKWELL TRANSFORM AND DISTRIBUTION STATIC COMPENSATOR (A CASE STUDY OF YIRGALEM SUBSTATION)

Abstract

Power quality has become a crucial concern recently due to the increase of the consumption of electrical load and the increment in the use of sensitive devices connected to power systems. In spite of that, complexity in modern daily life and the increased usage of semiconductors make non linear load a real threat to power quality level. In order that maintain power quality and to ensure its reliability, power quality disturbances must be identified and classified correctly and precisely. Thus, identification algorithms support decision makers to identified and mitigate the disturbance, and protect the power network from a high level of financial loss. In this thesis study identification, classification and mitigation of power quality issues in Awada industry zone. The measured voltage and current harmonic distortion levels are compared with the IEEE 519-2014 and IEC 61000-2-2 / -3-4 standards. The harmonic voltage distortion level in the factory has found to be well under the limits set by these standards while the current harmonic distortion levels on one of the transformer among four transform exceeds the limits with a maximum percentage total harmonic distortion current value of up to 23.09%. First, an identification process covering the most important and common power quality issues for further analyzed and discussed. Then after, most of the powerful processing algorithms in addition to support vector machine technique was investigated and their results are discussed. SVM then classify complex data and enhancing the evaluation process. This method achieved a sufficient detection algorithm, which overcame the Wavelet, Fourier and Hilbert limitations and resulted in an overall accuracy of 91.08%, 88.91% and 86.8% respectively. This resulted in a substantial improvement in terms of overall accuracy, with more than 97.1% when using Stockwell transform. In addition to the average classification accuracy, other common performance measures computed from the confusion matrix also presented and highest average accuracy of SVM is 98.3%. For mitigating, the current harmonic distortion level in the industry a D-STATCOM in current control mode is designed. The performance of the D-STATCOM is evaluated by simulating the distribution network with and without D-STATCOM. The simulation results show that the source current becomes pure sinusoidal and in-phase with the source voltage within 0.02 second and THDI reduced to 4.36% after the enabled of the D-STATCOM in the system.