Fault diagnosis model of power transformer based on combinatorial KFDA

Fisher discriminant is a classical linear technique widely used in pattern classification and feature extraction. When fisher discriminant is used in classification of power transformer fault based on dissolved gasses analysis (DGA), it is transformed to its nonlinear version in high dimensional feature space by means of the kernel trick. The kernel fisher discriminant analysis (KFDA) is presented to diagnose faults in oil-immersed power transformer. In order to improve the classification accuracy, the conception of combination is introduced. The fault diagnosis of power transformer is consisted of 4 KFDA. KFDA1 is used to classify the normal and fault. KFDA2 is used to classify the thermal fault and discharge fault. KFDA3 is used to classify the general overheating fault and severe overheating fault. KFDA4 is used to classify the partial discharge fault, energy sparking or arcing fault. The example shows that the combinatorial KFDA have emerged with better performance, and more flexibility.