Application of BP network based on free-weighting on transformer fault diagnosis

Applications of traditional back-propagation (BP) network and improved BP network on transformer fault diagnosis have outstanding effects and broad implementations. And the accuracy rate has reached 85%. However, BP network has disadvantages of local minimum and slow convergence speed. Improved methods can not fundamentally solve those disadvantages. BP network based on free-weighting can fundamentally solve these shortcomings. New algorithm which has been applied on transformer fault diagnosis has been proved that it could improve the accuracy rate. The accuracy rate has reached at the percentage of ninety five. It has a very good and wide application prospect in the power grid.