Detection and Feature Extraction of Single Power Quality Disturbances Based on Discrete Wavelet Transform, Energy Distribution and RMS Extraction Methods

Power utility providers and power industry service providers face a significant challenge in identifying the type of Power Quality Disturbances (PQD) automatically. This paper discusses a method to classify PQD using signal decomposition, statistical analysis and machine learning. Firstly, Discrete Wavelet Transform (DWT) is applied on the generated PQD signals to decompose the signal to obtain its representation in time and frequency domain. Secondly, first and second order statistical parameters are computed on the selected sub-band of DWT. These parameters are used as features vector for the machine learning based classifier. Our database consists of 2400 generated signals of PQD, which were divided into train and test set. Another set of noise corrupted signal database was generated to evaluate the capability of the system. SVM using quadratic kernel was selected as the classifier of the Power Quality Disturbances feature vector. Comparisons were also made with other types of classifiers and other types of mother wavelet filter functions. The results show that the combination of DWT and SVM managed to classify Power Quality Disturbances with high accuracy and has a strong resistance towards noise.

Download Full-text

Comparison of different feature extraction methods for the analysis of uterine magnetomyography signals to predict term labor

International Journal of Engineering & Technology ◽

10.14419/ijet.v7i3.29.18449 ◽

2018 ◽

Vol 7 (3.29) ◽

pp. 1

Author(s):

T Ananda Babu ◽

Dr P. Rajesh Kumar

Keyword(s):

Feature Extraction ◽

Wavelet Transform ◽

Discrete Wavelet Transform ◽

Wavelet Packet ◽

Extraction Methods ◽

Wavelet Packet Transform ◽

Svm Classifier ◽

Discrete Wavelet ◽

Term Labor

The prediction of term labor by analyzing the uterine magnetomyographic signals attempted in this research. The existing works did not focus on the classification of the signals. Publicly available MIT-BIH database records were divided into term-labor and term-nonlabor groups. This research presents two methods for feature extraction, discrete wavelet transform and wavelet packet transform. Energy, standard deviation, variance, entropy and waveform length of transform coefficients used in the first method. The normalized logarithmic energy of wavelet coefficients from each packet of the total wavelet packet tree used as the feature space for the second method. The labor assessment done through the classification of the features by using five different classifiers for different mother wavelet families. Discrete wavelet transform features extracted using coif5 wavelet with random subspace classification gives the accuracy, precision and FPrates of 93.9286%, 94.2014% and 5.7986% respectively. Using sym8 wavelet for wavelet packet transform features classified with SVM classifier performed well with 95.8763% accuracy, 95.9719% precision and 4.0281% FPrate. The results obtained from the research will be helpful in term labor assessment and understanding the parturition process.

Download Full-text