Classification of Composite Power Quality Disturbance Signals Based on HHT and S-Transform

Abstract

Abstract-The correct classification of single and composite power quality disturbances is the premise and basis of governance and control of power quality problems. This article extracts the high-frequency and low-frequency characteristics by S-transform of the disturbance signals, combining HHT to extract the instantaneous amplitude of the signal before and after the disturbance. Some characteristic functions is defined as the classification criterions by analyzing the characteristics of the disturbance signals, as to achieve the classification of power quality disturbances. Experimental results show that the fusion of the two methods can accurately detect the categories and characteristics of power quality disturbance signals, the starting and ending time of transient disturbance signals.

Training of the Naïve Bayes Classifier for the Detection of the Power Quality Events (Voltage Dip, Voltage Swell and Voltage Interruption)

“…Commonly used feature extraction methods is Wavelet Transform (WT), Fourier Transform (FT), Hilbert Hung Transform (HHT), and S-Transform. When the feature extraction has been done, the classification is carried out by mathematical classifiers which includes Artificial Neutral Network (ANN), Fuzzy Logic, Support Vector Machine (SVM) and Bayesian classifier [7] [15]. One of the visible sign of a power quality problem is a distortion in the waveform of the voltage of the power sine wave or from the amplitude established reference level or a complete interruption.…”

Power Quality Events Classification on Real-Time Voltage Waveform Using Short Time Fourier Transform and Bayes Classifier

“…The use of nonlinear loads in industries nowadays is on the increase and it needs to be emphasized that these nonlinear loads is a threat to the power system. Also, usage of renewable energy sources with distribution system is a potential cause of variety of problems that must be detected and processed for corrective or preventive measures [Chang, &Chan, 2015;Fakolujo, Adejumobi, &Ogunyemi, 2012] To prolong the service of life sensitive industrial equipment, protection against steady state distortion and temporary transients in the distribution lines should be promptly addressed [Fernandez-comesana, et al,2012;Yu, et al, 2013]. Generally, power quality events in electrical power system could manifest inform of voltage sags, voltage variations, interruption, voltage swells, brownouts, blackouts, voltage imbalance, harmonics distortion, harmonic resonance, inter-harmonics, notching, noise, impulse, spikes (voltage), crest factor, electromagnetic compatibility, dropout, fault, flicker, transient, overvoltage, and under voltage [Khalid, & Dwivedi, 2011] among others.For a particular analysis, features of interest could be extracted using Wavelet Transform (WT), Fourier Transform (FT), Hilbert Hung Transform (HHT) and S-Transform [Khalid, & Dwivedi, 2011].…”

“…The next stage after feature extraction is classification which could be done using mathematical-based classifiers such as artificial neutral network (ANN), fuzzy logic (FL), support vector machine (SVM) and Bayesian classifier [Mahela, Shaik, and Gupta, 2015a;Yu, et al, 2013].…”