Of Sofc Gas Detector In The Dga And Related Technology Research

Power transformers are the most important equipments in electric power system. At present, dissolved gas analyzing on transformer oil is the main method used by electric power department to secure power transformers'safe and stable running. Firstly, the status in quo of the online monitoring of the power transformer is discussed, and figures out that low veracity of the detector, poor exactness of quantitative method, low correctness and large calculation of expert algorithm are the choke point which restricts the popularization of the DGA online monitoring equipments. Then, the basic principal of the gas chromatography and its application on online transformer monitoring is introduced; the common gas chromatographic detector technique is analyzed and compared. According to the application request in practice, the SOFC sensors which are usually used to measure the Oxygen are originally utilized to detect the diagnostic gas dissolved in transformer oil. A lot of tests are implemented, including base line test, gas separation degree test and repeatability test, the paper comes to a conclusion:when the conventional method of peak height or peak area is used to quantitate the SOFC detector, the Relative Standard Deviation is larger than 50%,which leads to a big error. Based on the Nernst function and some tentative conditions, the mathematical model of the SOFC detector is established for quantitative analysis to solve the problem of the exactness. The exactness experiment of different concentration is conducted, and the results show that the repeatability which is expressed by Relative Standard Deviation of the model based on Nernst function is less than 10%,and the error of exactness is less than 16.5%,which complies with the requirement of the DGA online monitoring. Expert algorithms whose diagnostic results are directly used by substation workers to make decisions are one of the important parts in DGA on-ling monitoring system.Principles of the transformer faults which can be expressed by DGA data is expatiated, and the usual method of the transformer diagnosis'limitation are analyzed. Aiming at the limitation that conventional method has, and for the sake of increasing the legitimacy ration for diagnosis of the power transformer while lessen the calculation, a new algorithm based on linear classifier and BP neural network is proposed, whose realization step is expatiated. The results show that the new algorithm proposed by this paper can improve the legitimacy ration efficiently while the calculation is reduced considerably. By using the online SOFC chromatography equipments and LC-BP diagnosis method proposed in this paper, the status of the power transformer can be monitored and evaluated real time, while provides credible gist for the worker in transformer substation to maintain and manage the power transformer.