Please use this identifier to cite or link to this item: http://hdl.handle.net/10266/3289
Title: Development of An Intelligent Novel Algorithm for Transformer Fault Detection Using Various Emission Signals
Authors: Nagpal, Tapsi
Supervisor: Chudasama, B. N.
Brar, Yadwinder Singh
Keywords: power transformer;back propagation;Neural Network;probabilistic;fault diagnosis
Issue Date: 12-Nov-2014
Abstract: Transformers are an essential part of the electrical power system because they have the ability to change voltage and current levels, which enables the transformers to generate the electric power, to transmit and distribute electric power and utilize the power at economical and suitable levels. In electrical power system, voltage of electricity generated at the power plant is increased to a higher level with step-up transformers. A higher voltage reduces the energy lost during the transmission process of the electricity. After electricity has been transmitted to various end points of the power grid, voltage of the electricity is reduced to a usable level with step- down transformer for industrial customers and residential customers. Since power transformer is vital equipment in any electrical power system, so any fault in the power transformer may lead to the interruption of the power supply and accordingly, the financial losses will also increase. If an incipient failure of a transformer is detected before it leads to a catastrophic failure, predictive maintenance can be deployed to minimize the risk of failures and further prevent loss of services. To monitor the service ability of power transformer, many devices have been evolved such as buchholz relay, differential relay, over current relay, thermal relay etc. which are part of protection in terms of determination of faults in the transformer. But the main shortcoming of these devices is that they only respond to the severe power failures which require removal of equipment from the service. Even in normal operation, a power transformer is subjected to internal stresses that often, in time affect the performance and reliability of the transformer through the steady breakdown of its insulating materials. These materials include paper and oil. Such insulating materials after being subjected to a variety of stressful conditions that occur in a transformer, have been found to deteriorate, which results in generation of gases which are often combustible or harmful. During the operation of transformers, they are subject to electrical and thermal stresses, which can cause the degradation of the insulating materials. Stability and reliability of a power system in many respects depends on the condition of power transformers. Essential devices as power transformers are in a transmission and distribution system, a wide variety of electrical and thermal stresses often age the transformers and subject them to incipient faults. Being one of the most expensive and important elements, a power transformer is a xii highly essential element, whose failures and damage may cause the outage of a power system. Thus, techniques for early detection of the faults would be very valuable to avoid outages. The degradation of the insulating materials produces the degradation products, which are gases, which entirely or partially dissolve in the oil where they are easily detected at the parts per million (PPM) level by dissolved gas analysis. The transformer faults can be differentiated for their energy, localization and occurrence period. There are increased oil temperatures and generation of certain oxidation products such as acids and soluble gases, associated with the occurrence of the fault. In case of thermal and electrical stress condition caused by fault current in the transformers, the hydrocarbons molecules of mineral oil can decompose and form active hydrogen and hydrocarbon fragment can combine with each other to form gases like hydrogen (H2) , methane (CH4), acetylene (C2H2) ,ethylene (C2H4) , ethane (C2H6) , carbon monoxide (CO), carbon dioxide (CO2) etc. Such gases are considered as fault indicators and can be generated in certain. The deterioration of the insulation accompanies incipient faults, in the form of arcs or sparks resulting from dielectric breakdown of weak or overstressed parts of the insulation, or hot spots due to abnormally high current densities or due to high temperature in conductors. Irrespective of the cause, these stresses result in the chemical breakdown of oil or cellulose molecules constituting the dielectric insulation. Different conventional Dissolved Gas Analysis methods such as Roger’s ratio method, Dornenburg’s method, Duvel’s triangle method and key gas ratio methods are used to ascertain the exact amount of harmful gases dissolved in the transformer oil. The conventional methods have some drawbacks like some time the method is inconclusive on the specific fault-type and other time, it gives a false fault type. The present research work utilizes artificial intelligence techniques to detect, diagnose and classify transformer faults based on Dissolved Gas Analysis methods. The Dissolved Gas Analysis dataset encapsulates the data of gas concentration in ppm, collected from Punjab State Electricity Board, Patiala. The four different Back Propagation Learning Algorithms i.e. gradient descent method, gradient descent with adaptive learning method and Levenberg-Marquardt method have been designed and compared amongst themselves for their performance in transformer fault diagnosis and classification. The performance comparison has also been carried out between xiii Back Propagation Neural Network classifier and Probabilistic Neural Network classifier, which show that the later outperforms the former, for transformer fault classification. Additionally, a hybrid system i.e. Adaptive Neuro Fuzzy Inference System (ANFIS) has also been constructed to diagnose incipient transformer faults. It has been found that ANFIS outperforms the Artificial Neural Network (ANN) and Fuzzy Inference System (FIS) since it integrates the best features of both the expert systems. Entire thesis is divided into six chapters. Summary of each of them is provided here. Chapter 1 describes the significance of power transformer condition monitoring and fault analysis. Basic idea about research problem, the origin of transformer failure, transformer fault types and fault detection methods is given in this chapter. Chapter 2 delineates the basic construction of the transformer. It also describes the various types and properties of transformer insulating materials like mineral oil and cellulose paper. The significance of transformer condition monitoring and its advantages have also been described shortly. Chapter 3 summarizes the Dissolved Gas Analysis based techniques used for the power transformer incipient fault diagnosis. This chapter discusses the methodology and detailed operating procedure of Dissolved Gas Analysis. Chapter 4 focuses the implementation of artificial neural network based algorithms to classify different types of faults in a power transformer, meant particularly for Non-Destructive Testing of transformer fault classification. In this chapter, a knowledge based inference engine has been developed for transformer fault diagnosis, based on IEC 60599 ratio method (Dissolved Gas Analysis Technique). Chapter 5 employs the two soft computing techniques namely fuzzy logic and Adaptive Neuro-fuzzy Inference System (ANFIS), for detecting the faults in a power transformer. Chapter 6 summarizes all the major findings of the work. Future prospective and scope in field is also included in this chapter.
Description: PHD, EIED
URI: http://hdl.handle.net/10266/3289
Appears in Collections:Doctoral Theses@EIED

Files in This Item:
File Description SizeFormat 
3289.pdf3.11 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.