Yes you can; sometimes we suppose B=0, or u=0; in any case you can calculate it.

To understand better take a look at the example illustrated in the following link:

https://www.academia.edu/7547346/Subject_MI63_Kalman_Filter_Tank_Filling

Thanks a lot.

It's really helpful....

Mohammad

P is the predicted (a priori) error covariance. If you don't have control input(u=0) it becomes= P(k+1)=AP(k)A'+Q(k)

where Q(k) is covariance matrix of the noise process w(k)

In fact, most of the dynamic systems do not consider input, so I don't guess you have a problem.

For a comprehensive study, you can look at Simon's book about Kalman filter.

" Optimal state estimation: Kalman, H infinity, and nonlinear approaches "

Thanks