DIFFERENTIAL PROTECTION OF TRANSFORMER BASED ON ARTIFICIAL NEURAL NETWORK AND PROGRAMMABLE LOGIC

Electricity is important for all sectors of society and must be delivered to the ultimate consumer within preset limits of quality and reliability. Among other factors, it is essential that the electrical system rely on an efficient protection scheme, for precluding damages to the system and disruption in the supply of electricity to final customers. Considering the importance of protective devices in the performance of the electrical system, this work shows the development of a differential protection algorithm for power transformers, focusing its implementation in a programmable logic device. The proposed protection algorithm is based on a multilayer perceptron artificial neural network (ANN), which presents as the main feature and the ability to recognize patterns and approximate functions. These characteristics enable the ANN to differentiate between normal operation conditions and critical conditions in a power transformer. As certain functions required by the algorithm would produce complex ANNs, this work utilizes a high capacity field programmable gate arrays in the differential protection implementation. In this work, we opted for the description language hardware VHDL to describe the protection algorithm to be implemented.