Classification of power quality disturbances using wavelet transform based artificial neural network / Mohamad Hafiz Mohamad Taib

• Main Author: Mohamad Taib, Mohamad Hafiz
• Format: Thesis
• Language: English
• Published: 2015
• Subjects: Neural networks (Computer science); Electric power failures
• Online Access: https://ir.uitm.edu.my/id/eprint/67094/2/67094.pdf; https://ir.uitm.edu.my/id/eprint/67094/

In electrical system, there are various power quality disturbances such as harmonic, voltage sag, voltage swell, and transient which may lead to power quality problems and will give adverse effects to the electrical supply performance. In this project, the wavelet transform based artificial neural network classifier is implemented under two types signal disturbances namely transient distortion and voltage sag to improve the classification process of the signal. The feature extraction is done by using wavelet transform and the details are collected and given to the neural network. The fifth order of Daubechies wavelet with the decomposition at level four is implemented in order to get optimum wavelet decomposition. The features from wavelet transform analysis will be an input to the neural network and the output will be two types of disturbances namely voltage sag and transient. The accuracy level and the advantage of integrating the wavelet transform with neural network will be discussed in this project by analyzing the performance measures such as accuracy, precision, sensitivity and confusion matrix specificity. The rate of classification is perfectly with 100% accuracy for two types of power quality disturbance signals.