Abstract:In order to improve the diagnosis of transformer fault and the effective realization of state maintenance, according to the 500 thousand volt transformer internal latent fault case describes the use of a variety of online monitoring and off-line detection means and fault diagnosis method for quickly and accurately find the reason and position of transformer fault. It is through the use of oil chromatographic detection, transformer vibration and sound detection, capacitive equipment, on-line monitoring of transformer partial discharge detection, charged test condition monitoring technology to carry out multidimensional evaluation and fault diagnosis of transformer, and helps to realize the accurate positioning and the development of condition based maintenance strategy for internal fault of the main transformer. The main transformer realizes the hidden trouble quickly and accurately after the condition monitoring and fault diagnosis. It can restore the normal running state through emergency treatment, so as to avoid a major equipment accident. The results show that the application of transformer oil dissolved gas chromatographic analysis and detection technology can accurately judge the nature and severity of transformer faults by eliminating method, which is a particularly effective way to detect latent faults in transformer at early stage. At the same time, the feature extraction method of transformer vibration signal based on Cross Wavelet Analysis and evaluation results shows that there is a large number of 50 Hz harmonic components after the grounding fault of transformer core. With the help of vibration and sound detection technology, it can help transformer fault diagnosis and accurate positioning, and improve the efficiency of equipment condition maintenance.