FUZZY BASED UNIT PROTECTION SYSTEM FOR POWER TRANSFORMERS

Abstract

This thesis presents a new protection algorithm for three phase power transformers utilizing fuzzy logic. A transformer model was created using Alternate Transient Program (ATP). Various kinds of faults were generated using the program and these include interwinding faults, three phase faults, phase to ground faults, phase to phase faults, primary to secondary faults, as well as, external faults. Inrush conditions and over-excitation were also simulated. These fault data were used to test the protection algorithm. The algorithm first extracts the fundamentals and harmonic components of the transformer currents. It then computes the difference currents and restraint indices. These are inputs to the fuzzy logic. These inputs are fuzzified and the centre of gravity method is employed for defuzzification for all the three phases. The output of the defuzzification is a trip signal which indicates if the respective phase is at fault. The algorithm analyses the trip signals from all the three phases and determines if an internal fault has occurred and consequently identifies those phases that are at fault. The algorithm has been simulated successfully on MATLAB environment, as well as, in real time on a 486 personal computer. In most cases, the algorithm issues a trip signal within half of a power cycle after the occurrence of an internal fault. It does not trip under inrush and over-excitation conditions. Besides, the phases at fault are identified correctly.