(a) The estimates of the voltage magnitude, the frequency drift, the
rate of change of the frequency and the voltage phase angle arc
greatly affected by the noise level, as the noise level increases and
the estimates getting poor. Indccd, that is true, because as the noise
level increases, the error in the measurements increases, and since
Kalman filtering algorithm has been built on the basis of the
recursive least error squares, which is adversely affected by the
presence of bad data. Note that these curves are obtained when the
whole set of measurements, the on-line measurements, is noise
contaminated. However, it has bccn shown, through extensive runs, that
if a zero mean Gaussian noise contaminates a numbcr of measurements,
which is the case in reality, where some of the measurements arc good,
and some of them are bad, the estimates arc greatly improved.