The Research And Implementation Of Transformer Diagnosis System Based On FCM Algorithm And Fuzzy PETRI Net

The transformer is one of the most important equiments in power system, which operation state direnctly relates to the system reliability and safety. Once the transformer breaks down, and the economic loss usually is caused. So, the monitoring of the transformer operation and faults must be strengthened. The transformer internal faults are directly reflected by the composition and content of the gases dissolved in transformer oil. With the transformer preventive test, it is used as one of the most effective means. The main contens are as follows:Firstly, based on the analysis of DGA, three-ratio method is introduced in the paper and its advantage and deficiency are also stated. To sovle the existing defifeincy, FCM Classification Algorithm of transformer faults diagnosis is proposed. According to actual needs, a new method is raised to perfect the FCM training parameters. Throug the programming by Matlab, the new method is proved to be feasible.Second, the Fuzzy Petri Net faults location model, based on the FCM Classification Algorithm and transformer preventive test, is firstly structed. The model can be used under complete conditions and incomplete conditions. Another important task is to complete the reasoning algorithm of the model and to simulate the model by Matlab. The model is verified to be correct and useful by the examples.The FCM algorithm based on DGA data, can quite good realize the transformer breakdown phenomenon division. However, this algorithm can't diagnose the falut reson. The Fuzzy Petri model is the right method to overcome the shortage of the FCM algorithm. But, this model can't process the lager amount data. So, the FCM algorithm and the Petri Net model are combined in this paper to complement each other, both to realize the classification of symptoms, but also to realize the cause of the fault location.Finally, with FCM classification algorithm and Fuzzy Petri network as the core, the mixing programming of Vc++ and Matlab as the foundation, the transformer fault diagnosis software is designed and developed. The software realizes the aim of fast-diagnosis, accurate positoning. The strong point of the software is easy operation, unambiguous diagnosis processing and visible diagnosis result.Through the use of the system in the transformer substation, it is verified to be accurate and practical.