
496 Optimal Estimation of Dynamic Systems
ance involves an off-line computation to satisfy the autocorrelation test in Equa-
tions (4.83) and (4.84). Since the state matrices are constant and the measurements
are assumed to be sampled frequently, then the steady-state discrete-time Kalman
filter shown in Table 3.2 can be used. The discrete-time state matrices, Φ and Γ, can
be numerically determined using Equations (A.123) and (A.124). An analytical solu-
tion for the discrete-time process noise covariance is difficult to determine for high-
order models. Therefore, Equation (3.183) will be used to determine this covariance
matrix. The steady-state error ...