Research On Transformer Remote Faults Diagnosis Based On Support Vector Machine

Transformer is one of the most important electrical equipment of power system, and is also the important electrification power supply equipment. With the sustained development of national economy, transformer is developed along the direction to be high pressure, high capacity, automated, and high reliability, thereby it is important to improve the operation reliability of transformer. It has practical significance to develop the technique of fault diagnosis and improve the level of operation and maintenance of transformer, which are important contents of achieving state maintenance of transformer. Dissolved Gas Analysis is the efficient technique to measure the fault of transformer.The transformer fault diagnosis using support vector machine based on DGA is carried out, through case analysis to prove its effectiveness. Transformer remote fault diagnosis control interface based on LabVIEW technology is designed to realize the transformer gas data transmission using GPRS wireless communication technology, the main contents and conclusions are as follows:1. First of all, the common faults of power transformers are analyzed in this paper, the common methods of transformer fault diagnosis are described and then the traditional fault diagnosis methods based on DGA are reviewed. A fault diagnosis method based on support vector machine is proposed, as the traditional fault diagnosis methods have many deficiencies.2. It introduces statistical learning theory and the basic principles of support vector machine, and derived that support vector machine has a good nonlinear mapping ability from the theoretical and practical.Also it describes the deficience of traditional normalization methods, then presents an improved one,which compares with the traditional normalization methods.Describing the characteristics of different kernel functions, and then the most appreciate kernel functions would be selected through comparing the different algorithms of different kernel functions.Make a two-classification and multi- classification contrast to the support vector machine and the traditional ratio method, in order to validate the advantages of support vector machine. The superiority of the support vector machine was proved through the example for transfer fault diagnosis.3. Selecting wireless communication modules MC55 to complete the data transfer between on-site and diagnostic center, and ensure the data transmiting to the diagnostic center timely and accurately, then to make the appropriate diagnosis. The hardware and software designment of wireless communication module is described in detail, the important AT commands is introduced, and makes a reliability design to the system.4. Transformer remote fault diagnosis based on support vector machine control interface is designed through Lab VIEW technology.Based on the above research results, a transformer remote fault diagnosis system is developed, through integrated debugging, the on-site collectional gas data will be trsferred to the remote fault diagnosis center in real time, the higher classification accuracy would be obtained through the diagnostic system of supporting vector machine.