Exploring novel methods for power quality disturbance and fault type classification.

摘要

The need for better power quality is more important now than ever before due to the increasing use of devices that are sensitive to power disturbances. This fact calls on utility companies to improve their power quality monitoring systems. Also, utilities are interested in fast fault location techniques to reduce the length of power outages and the unnecessary loss of revenue for end users. Fault type classification is a prerequisite for many fault location algorithms; therefore, it is important to have an effective fault classification algorithm. New and improved monitoring systems should be able to detect and classify the disturbances and fault types in a quick and efficient way.;The main focus of this research involves the development of automated systems that can automatically detect and classify different power quality disturbances and fault types. Advanced signal processing techniques such as Windowed Discrete Fourier Transform (WDFT), Wavelet Transform (WT), and S-Transform have been explored to detect power quality disturbances and faults, and to extract the unique features from the analyzed waveforms.;In the effort to distinguish power quality disturbances, various apposite parameters for describing specific types have been identified and the signal processing techniques mentioned above have been utilized in order to obtain the desired parameters. Once the features are extracted from the waveforms, intelligent techniques such as Adaptive Neuro-Fuzzy Inference System (ANFIS) and Binary Based Feature Matrices are utilized for the decision making step.;For the problem of fault type classification, the Discrete Fourier Transform is employed to obtain the zero sequence of the waveform during fault. Also, the inter-quartile range and correlation coefficients between each phase are also obtained in order to classify the ten different types of faults. The same ANFIS decision making method that we applied to the power quality classification problem is also used here. To verify the accuracy of the proposed ANFIS method we also called upon another intelligent technique with the name of Support Vector Machines (SVMs).;KEYWORDS: Power Quality Classification, Fault Type Classification, Adaptive Neuro-Fuzzy Inference System, Binary Based Feature Matrices.