Signal Processing-Based Artificial Intelligence Approach For Power Quality Disturbance Identification
No Thumbnail Available
Date
2022-04
Journal Title
Journal ISSN
Volume Title
Publisher
Department of Electrical and Electronic Engineering
Abstract
Power quality has become a vital concern recently due to the expansion of the consumption of electrical load and the growth in the use of sensitive devices connected to power systems. Nevertheless, complexity in modern life and the increased usage of semiconductors for non-linear load, make a real threat to power quality level. The modern power supply, based on developing renewable sources such as solar, wind and nuclear energy, has increased power quality disturbances to a significant level. In order to maintain good power quality and to ensure its reliability, power quality disturbances must be detected and identified correctly and precisely. Thus, detection algorithms assist decision makers to solve the disturbance, and protect the power network from a high level of financial loss. In this thesis, a new approach of a detection algorithm and classification technique is proposed for power quality disturbances based on the methodology of an Advanced Signal Processing algorithm and Artificial Neural Networks. Firstly, an investigation process covering the most important and common power quality disturbances is analyzed and discussed. Thereafter, most of the powerful signal processing algorithms in addition to Artificial Intelligence techniques are investigated, their results are discussed. Since power quality disturbances are non- stationary signals, a characterization process is built by simulating 7 power quality disturbances in power systems. As a result, a validation methodology is conducted basedon discrete Wavelet Transform and the extracted features are recorded. Artificial Intelligence techniques can classify complex data and enhancing the evaluation process.Therefore, these extracted features are fed to Artificial Neural Networks to train the database of the generated power quality disturbances. This method achieved a sufficient detection algorithm which overcame the Fourier and other transform limitation and resulted in an accuracy of near 98%. In this thesis, a comparison analysis for both detection algorithms combined with the Artificial Neural Networks classifier is presented, this shows the robustness and the effectiveness of the proposed methodology
Description
submitted by Md. Sadman Sakib bearing
Matric ID. ET 171023 and S.M.Sazzadul Haque Tanim, bearing Matric ID. ET 171024
of session Spring 2021