New Fourier filter algorithm for series compensated transmission lines

Abstract

In protection relaying schemes, the discrete Fourier transform (DFT) is the most widely used algorithm for computing the fundamental frequency components. When the measurements only contain fundamental frequency and integer harmonic frequency components, the conventional DFT only need one-cycle post-fault data amounts to compute the fundamental frequency component. However, in series compensated lines, the voltage and current signals both contain large sub-synchronous frequency and decaying DC components during the fault interval. These abnormal components involved in measurements will extremely postpone the convergent speed of the conventional DFT algorithm. Different from the ideal one-cycle post-fault data amounts of the ideal cases, the conventional DFT need 5-10 cycles post-fault data amounts (for decay DC component) or 10-20 cycles post-fault data amounts (for sub-synchronous frequency component) to compute the convergent fundamental frequency component. The vital slow convergency will extremely reduce the accuracy and response time of the following fault locator or other installations in the relaying schemes. In order to overcome the above problems, this paper presents a new Fourier filter algorithm for series compensated transmission lines. Via the proposed algorithm, the convergent fundamental frequency components can be computed in only 2-3.5 cycles post-fault data amounts, even the decaying DC or sub-synchronous frequency components involved in measurements. Meanwhile, the resolution of the A/D converters and the effects of the low-pass filter are also considered in the investigations. Since the proposed algorithm effectively suppresses the abnormal frequency components, not only the fundamental frequency components but also the following fault location computations all can be achieved very fast. EMTP generated data using a 300 km, 345 kV series compensated transmission line has tested the performance of the proposed algorithm. The tested cases include various fault types, fault locations, fault resistances, fault inception angles, etc. Simulation results indicate that the proposed algorithm can achieve up to 99.95% accuracy for most tested cases.