Fault diagnosis of a converter transformer based on the FOA-LSSVM

摘要

In order to improve the accuracy of fault diagnosis of a converter transformer, a model has been proposed based on the Least Squares Support Vector Machine (LSSVM). Optimization of traditional LSSVM parameters is aimless and inefficient. In order to solve this problem, this paper uses fruit Fly Optimization Algorithm (FOA) to optimize the penalty parameter C and the kernel function parameter γ in LSSVM and build up the converter transformer fault diagnosis model based on the FOA-LSSVM. It is verified through a case analysis that FOA has higher precision on classification and global search ability compared with Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). The simulation results indicate that the fault diagnosis model of the converter transformer based on the FOA-LSSVM has a faster convergence speed and a higher diagnostic accuracy compared with the traditional LSSVM, GA-LSSVM, and PSO-LSSVM.