Frequency Estimation using Curve Fitting

Abstract

Frequency is an important parameter for monitoring a power system for control and is needed as part of the protection settings for grid connected distributed generation. Frequency is an indirect measurement that is typically estimated over time using an averaging technique. However, grid events such as phase jumps or faults and changes in distributed generation are sufficient to trigger changes to frequency calculations that could cause incorrect tripping of plant. In particular, fast acting distributed generation or energy storage are known to frequently trip off at the point of common coupling due to rate of change of frequency (ROCOF) protection. A variety of methods have been proposed to improve the frequency estimation or ignore bad ROCOF and frequency measurements during events for use in relays. This paper builds on previously published work with a curve fitting method in conjunction with a goodness of fit calculator to improve the curve fitting process to include the impact of step change events by modifying the curve fit equation. The paper uses real data sets to analyse and compare how the curve fitting works under different grid events. The paper shows that this method offers an advantage over traditional methods of frequency determination and would minimize tripping of distributed generation and energy storage systems on ROCOF settings as they ramp up and down quickly.