基于dga与微水检测的电力变压器运行状态评估系统研究-物理电子学专业论文.docxVIP

ABSTRACTGrounded ABSTRACT Grounded on deep research on the transformer fault diagnosis technology and principle，a moisture content in transformer oil on—line monitoring system is put forward．Dissolved gas analysis has better performance in transformer fault diagnosis when combine with the moisture data．A new BP Neural Network algorithm based on Rank Ants System(ASRNN)is firstly brought forward．Firstly,the development and current state of Transformer Fault Diagnosis system iS analyzed．Then various measurements and diagnosis methods for transformer iS discussed． Based on this discussion，disadvantages of DGA application on transformer fault diagnosis iS summarized．Moisture content in transformer oil measurement iS very important SO that a moisture content in transformer oil on—line monitoring system based on water activity iS designed．Then a BP Neural Network model iS built for transformer fault diagnosis．The diagnosis result proves that BP Neural Network iS better than improved IEC three．ratio method．but it still has rather poor performance of convergence and quick approach to the global optical．To improve the performance，and Rank Ants System algorithm is proposed and applied to the exploration for better transformer fault diagnosis method．The design and implementation of BP Neural Network al gorithm based on Rank Ants System iS shown in detailin this thesis．The result of simulation of ASItNN system bv MATLAB proves that ASRNN system surpasses the BP Neural Network with a speed up in convergence，short time in training．better diagnosis accuracy and more actual reflection of practical fault situation．Finally,a large power transformer running state integrated evaluation system iS built．Based on the moisture content in transformer oil on．1ine monitoring system and ASRNN system．the evaluation system iS able to provide the maintainers in transformer substations or power plants a reliable basis for their maintenance and management of large power transfotruer． KEY WORDS：Pow