Study On The Technology Of The On-line Monitoring And Fault Diagnosis For Transformer Insulation Using The Character Value Of Oil-dissolved Gases

The operation reliability of the power transformer as the key equipment in electrical power systems, influences operation security of electrical power systems directly. The components and contents of gases dissolved in transformer oil can be used to reflects internal insulation faults of operating transformer. In order to overcome the errors caused by complex handling procedure and man-made factors using general chromatogram analysis method, author brings forward an on-line detecting of gases dissolved in transformer oil by using macromolecule polymer to separate oil and gases automatically and an information fusion technology of multi-sensors; at the same time, in order to improve the accuracy and reliability ,author uses neural networks to diagnose transformer faults. The research works are shown mainly as follows:1) The actuality and outlook of on-line detecting and fault diagnosis technology of gases dissolved in oil are analysed deeply, and characteristics of insulation structure of power transformer, evaluation of operating reliability and the relationship between ordinary faults inside transformer and gases dissolved in transformer oil are studied too. Analysing the purposes of detecting gases dissolved in transformer oil on-line, we can know that primary indexes to evaluate the technology of on-line detecting gases dissolved in transformer oil are reliability of system, accuracy of detected data, validity of diagnosis, automatization and price of systems.2) On-line extracting gases technology using macromolecule osmotic membrane is studied. The method using an improved osmotic membrane F46(made by ethylene polymer) to separate gases from oil is brought forward. Capability of osmotic membrane F46 is discussed too. From a great of experimentations, we get the osmotic balance curve of gases,confirm balance conversion coefficient. All of them settle the foundations of applying on-line detecting gases dissolved in oil to electrical power systems. 3) Studying the principle and method of many sensitive elements which are used to detecting the gases dissolved in oil, author bring forward the viewpoint that sensors of MQ series have obvious selectivity and higher accuracy when they detect gases of H2, CO, CH4, C2H4,C2H2and C2H6 dissolved in oil. Especially, the accuracy of C2H2 which reflects the main faults of transformer can reach 1uL/L. 4) Multi-sensor information fusion technology is studied and used to settleintersectant sensitivity problems among the six gases when sensors of MQ series are used in system. The technology of sensor array and the method of NN model identification can enhance the detecting accuracy and reduce the requirement of selectivity and time stability.5) Studying the diagnosis principle and method of NN used in electrical system, author brings forward a model of transformer insulation diagnosis based on the NN of ART. The reliability of transformer insulation diagnosis based on ART of NN has been proved by actual samples in the dissertation.