A study of estimation techniques for frequency-relaying applications

This paper presents an evaluation and performance comparison of two estimating algorithms used for frequency-relaying applications. The first algorithm is based on least squares (LS) error estimation, while the second is based on the least absolute value (LAV) approximations. Two models for frequency estimation, namely the constant-frequency model (CFM) and the variable-frequency model (VFM), are used. CFM is used to measure the steady-state frequency deviation, while VFM is used to measure the transient frequency and its rate of change. For the constant-frequency model, which is suitable for use in microprocessor-based relays, the algorithms are tested using two data sets. The first involves a pure sine waveform, and the second is a wave generated from the electromagnetic transients program (EMTP) simulating a study system. The EMTP is used to generate the corresponding voltage samples at a relay location. Fault conditions applied on the study system to generate distorted waves with and without frequency deviations are considered. For the variable-frequency model, the algorithms are tested using a data set generated from a pure sine waveform. The effects of sampling rate, sample window size and bad data on the performance of each algorithm are studied. A comparison of the results is offered to evaluate the two techniques.