Jopri, Mohd Hatta (2021) Harmonic distortion analysis in power quality signal using time-frequency distribution. Doctoral thesis, Universiti Teknikal Malaysia Melaka.
![[img]](http://eprints.utem.edu.my/style/images/fileicons/text.png) |
Text (24 Pages)
Harmonic distortion analysis in power quality signal using time-frequency distribution.pdf - Submitted Version Download (379kB) |
|
Text (Full Text)
Harmonic distortion analysis in power quality signal using time-frequency distribution.pdf - Submitted Version Restricted to Registered users only Download (13MB) |
Abstract
Harmonic distortion in the electrical power supply is caused by an increase in the number of power electronics devices. Harmonic distortion may have an effect on the production process, as well as economic losses and equipment failure. As a result, it is important to detect harmonic signals, identify, and to diagnose type of harmonic source in order to take precautionary measures to avoid the negative effects of harmonic distortion. Mostly, the power quality (PQ) analysis only focuses on the harmonic signal measurement, whereas it is also necessary to identify the location and type of harmonic sources with low complexity and high accuracy capability. Therefore, this research presents PQ signal analysis, detection, harmonic source identification and diagnosis method. The power quality signals consist of multi-frequency components and magnitude differences, thus, the time-frequency distribution (TFD) is very suited to present the signals within time-frequency representation (TFR) and to detect power quality signals accurately. The TFDs namely spectrogram, Gabor transform and S-transform are used in this study. The signal parameters are estimated and then are used to identify the signal characteristics based on the IEEE Standard 1159-2019. The best TFD in harmonic signal detection is identified in regards to the accuracy, calculation complexity, and memory size of signal analysis. Next, using the best TFD, the harmonic source is identified either from downstream and/or upstream of the point of common coupling (PCC) based on impedance spectral. Afterwards, five machine learning methods include k-nearest neighbour (KNN), support vector machine with linear function (SVM-L), support vector machine with radial basis function (SVM-RBF), linear discriminate analysis (LDA) and naïve Bayes (NB) are used to diagnose the harmonic sources. Three harmonic signal parameter groups which are harmonic voltage parameters, harmonic current parameters, and harmonic voltage and current parameters are examined. The performance of the detection method is verified by generating and detecting 100 multiple characteristics signals for each type of power quality signal. Meanwhile, 100 signals of harmonic sources, which are from rectifier and inverter loads with various characteristics in terms of firing angle, amplitude and frequency modulation indexes are evaluated in identification and diagnosis of the harmonic source method. The diagnosis results indicate that the LDA with harmonic voltage parameters offer the highest accuracy and fastest computation speed. To validate the proposed method, the real signals of field testing were recorded and analysed for detection, identification, and diagnosis methods. The results show that the proposed method provides high accuracy and fast computational analysis, making it ideal for use with an embedded device in detecting power quality signals, identifying, and diagnosing harmonic sources. The proposed method gives high-impact to the industry especially in reducing maintenance cost, and trouble-shoot duration of power system failure.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Uncontrolled Keywords: | Electric power systems, Stability Electric power systems, Quality control, Signal processing |
| Subjects: | T Technology > T Technology (General) T Technology > TK Electrical engineering. Electronics Nuclear engineering |
| Divisions: | Library > Tesis > FKE |
| Depositing User: | F Haslinda Harun |
| Date Deposited: | 13 Jan 2023 16:30 |
| Last Modified: | 13 Jan 2023 16:30 |
| URI: | http://eprints.utem.edu.my/id/eprint/26071 |
| Statistic Details: | View Download Statistic |
Actions (login required)
 |
View Item |