)
(see Figure
4-2). Computational delay is explicitly modeled in SIMLAB and
allows the assessing of trade-offs such as model complexity (for
better prediction capability) and computational delay. If the delay
is significant, for example, the generated control and guidance may be
obsolete when it is implemented. If the computational delay is longer
than the rolling horizon step size (
), then
the guidance generations need to be computed in parallel.
Specifically, a guidance generation task is started every s
minutes on the next available processor (one or more earlier tasks may
be still running on other processors). This ensures a new guidance,
which is based on information collected minutes
ago, and becomes available every s minutes.
As stated earlier, the default approach to predictive guidance generation is based on two models: state estimation/prediction and guidance generation. It is an iterative process which attempts to generate a consistent guidance.