David Megginson <[EMAIL PROTECTED]> said:

> Jim Wilson writes:
> 
>  > I'd like to project (simplisticly) airspeed in 10 seconds,  primarily for
>  > autothrottle accuracy with jet engines, but also for display purposes.
>  > 
>  > The idea would be something like this:
>  > 
>  > airspeed_in_10secs = airspeed_kt + 
>  >        (airspeed_kt - last_airspeed_kt) * (10/dt);
> 
> You might want to increase your sample rate a bit: for example, you
> could sample every second and do a rolling weighted average of the
> acceleration (delta airspeed) over the past ten seconds.

That was my original thought,  but I was afraid that it would extend the
upper/lower limits (make larger predictions) and my readings would come up
short.  So it seemed to be safer, and of course easier, to base the data on 
what was happening at that instant, rather than over time.

Using this method I've been able to really zero in the selected speed for the
autothrottle on the 747.  The cessna does just as well,  so both extremes if
you will seem to respond well to the simplistic 10 second projection.

The thing I'm concerned with now is locking in an update frequency for the
autopilot.  Any suggestions on which method I should use?  Currently the
autopilot updates just before the fdm_states in the main loop...and doesn't do
that until after the FDM inits.

Best,

Jim

_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Reply via email to