Vector fitting for estimation of turbine governing system parameters

Abstract

With the introduction of more and more renewables into the power system both the inertia and the primary frequency reserves are expected to decrease. It is therefore a growing concern that the frequency quality will deteriorate. One way of mitigating these problems may be a more detailed monitoring of the generators providing the primary reserves. A promising approach for monitoring the generators is to identify turbine governing system parameters using system identification. This will allow for estimating the droop and the bandwidth of the governor, parameters that are important for the primary control. Furthermore, if this can be reliably done on ambient data, updated estimates of these parameters can be obtained relatively fast. In this paper we will look into how vector fitting can be used for this purpose. The algorithm possesses some interesting properties for automatically constructing models from ambient data. How this can be done will be presented together with results obtained using real data from the Norwegian power system. A simple criterion for reducing the obtained model order is also proposed.