Fault Diagnosis for Oil Immersed Transformer Using Certainty Factor

Traditional dissolved gas analysis (DGA) methods for incipient fault diagnosis of oil immersed power transformers often give contradictory results. Due to the increased availability of methods, and procedures for fault diagnosis a reliable method is needed to reach a conclusion. The certainty factor (CF) method gives promising results where the available knowledge base and techniques are unable to arrive at the same conclusion. In this work, a traditional method based on a gas ratio method, two graphical methods, and a key gas (KG) method are applied to DGA samples. When the fault diagnosis by the traditional methods is contradictory, a computational method based on evidence and CF is used. The combined certainty factor (CCF) is applied to the CF results incrementally for confirming fault type. There is a special provision for partial discharge (PD) type of fault. The proposed method is checked on samples used in literature where the actual fault is known. The goal is to make available a reliable method that scientifically combines the diagnosis of three traditional DGA methods and confirms with an increased level of responsibility the diagnosis of the type of incipient fault. The proposed method may be used by utilities to reach a common consensus in case of contradictory diagnosis by traditional DGA methods.