Research On Transformer Fault Diagnosis Based On Artificial Fish Swarm Improved Support Vector Machines

As an important equipment in the power system,power transformer can reliably operate,which is of great significance to the safe and stable operation of the power system.Because of the long running time and complicated operating environment of the transformer,it is inevitable to fail in the case of not timely maintenance.With the continuous development of China’s economy,people’s demand for power system reliability is getting higher and higher,it is particularly important to find the potential faults of transformer and repair it in time.In this paper,the method of dissolved gas used in transformer fault diagnosis is introduced firstly,and the fault diagnosis of traditional transformer fault diagnosis and artificial intelligence technology is based on this method.This paper analyzes the generation of dissolved gas in oil and the corresponding relationship between transformer fault type and gas content in oil,and the common fault types of transformer are determined.Then,the basic principles of artificial fish swarm algorithm,statistical learning theory and support vector machine are discussed,and the multi-classification method of support vector machine is studied deeply.The classification of support vector machine is improved by introducing improved method of directed acyclic graph performance,and through simulation experiments to verify.In addition,this paper describes the artificial fish swarm algorithm to optimize the support vector machine parameters of the process,and the theory applied to the specific transformer fault diagnosis.The transformer fault diagnosis model is established,and the performance of the diagnosis model is tested by the test set of samples.The method is compared with the BP neural network fault diagnosis method and the improved three ratio method,and the performance of these methods is compared.At last,this paper realizes the transformer fault diagnosis system by programming,the platform of the system is built based on MATLAB GUI(graphical user interface environment).The advantages of the system are that the users don’t need the command input only through the keyboard and mouse click on the button to achieve the corresponding functions.After the establishment of the system,two sets of specificexamples of transformer faults are selected,and the performance of the fault diagnosis system is tested,and the availability and practicability of the system are proved.