Application Of Extension Theory And Depend Analysis In Diagnosis Of Transformer Faults

The commonly used methods of fuzzy expert system, self-adopting and self-learning diagnostic system and artificial neural network system have various shortcomings. In order to overcome these shortcomings, this paper proposes a new diagnostic method. Above all, the extension theory based on the matter-element model and the correlation analysis method to the transformer fault diagnosis process are applied which provide a theoretical basis for establishing a control platform that could realize remote monitoring a fault. This paper deal with the following:A matter-element model for transformer fault diagnosis is established through studying the implication relationship between the reason and the manifestation for transformer fault and the Extension theory.By adopting mathematic logic theory and sets theory, the nature of the extension relationship equation, the interval distance and the interval location are studied as well as the establishment method for correlation function.According to the mathematics mechanism, physical meaning and nature of extension correlation degree, different mathematical treatment formula are determined, and its necessity, feasibility are analyzed as well as its advantage or disadvantage in actual practice.The natures of consistency, categoricalness and nonredundancy of correlative matrix are studied as well as the linear algebra derivation method for correlation degree. Then the paper evolves the correlation degree into the optimal algorithm which could be applied to actual practice.Based on the matter-element model, the paper considers the power transformer fault from the points of qualitative and quantification. It puts forward a kind of extension diagnostic method based on the DGA technique and with which power transformer fault types and degree could be determined simultaneously. The foundation, method and process of the provided diagnostic method are analyzed and discussed. The results of great many experiments and site tests show that the accuracy of the provided method is up to 91% and additionally, it could diagnose two or above kinds of faults existedsimultaneously.The paper researches the fault diagnosis technique based on correlation analysis and presents three step correlative algorithm. A three step association degree method for diagnosis of power transformer fault which combines the qualitative and quantification methods is established. The contents parameter of five kinds of gas which could dissolve into oil ( hydrogen, methane, ethane, ethylene and acetylene ) are taken as characteristic quantity to demonstrate the actual diagnosis effect of new method to the transformer fault. The results show that by using this method, not only the fault types could be found out, but also the fault location could be observed. Result also show that the effectiveness of the provided method exceeds the traditional correlation degree method, thereby the transformer fault diagnosis theory system is established.In this paper, the concepts of extension sets, matter-element concept, correlation function and correlation degree are introduced to the transformer fault information treatments. By adopting the extension sets and matter-element concept, the extents of objects belonging to certain sets can be judged according to the characteristic values of this object. By using the correlation function, the fault diagnosis information could be refined and quantify, additionally, by introducing the correlation degree, the running state and current or future fault mode could be deduced. These make the transformer fault diagnosis theory more abundant and advanced and thereby provide a new method for resolving the transformer fault diagnosis difficulties. The fault diagnosis technique based on the extension sets theory breaks through the limits of DGA diagnosis method and ANN diagnosis method. It could resolve the transformer fault troubles which the DGA diagnosis method and ANN diagnosis method could not overcome or could not overcome successfully. This may create a new world for power transformer fault diagnosis and a significant revolution of transformer fault diagnosis can be realized.