How to handle a controller that needs jerk and snap as inputs? 1

[HH space engineer]

I have developed a controller for spacecraft rendezvous using low thrust propulsion system. In the controller, it appears that we need to have a knowledge of jerk and snap (first and second derivative of acceleration) of the spacecraft, as inputs. However, we know that jerk, and more so, snap are hardly measurable, and differentiating acceleration to get these values would be full of noise. How can you handle such scenarios?

 

[IRstuff]

Huh? You developed the controller, so why did you include jerk and snap if you can't generate the inputs?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[rb1957]

If these acceleration is almost constant, so that these properties hardly change, it seems odd that these are needed for control.

Or is it that these are changing by a small amount but at a significant frequency ?

another day in paradise, or is paradise one day closer ?

 

[Sparweb]

Somehow this sounds like a re-reading of the spacecraft specifications, followed by the discovery that certain control parameters were expected or required in the design spec but omitted by the builder. If you're arrived at this situation, well, there's a new/old saying that any hardware problem can be fixed with software.

If you're getting too much noise on the first derivative of acceleration, maybe you need a better MEMS gyro?
Is the problem self-induced (vibration?)

No one believes the theory except the one who developed it. Everyone believes the experiment except the one who ran it.
STF

 

[VEBill]

Based on the history of US space flight, i.e. Gemini and Apollo, they used attitude and burn duration to achieve the desired delta V.

The engine thrust (acceleration) does turn on and off quickly, thus having squarewave edges which causes higher order derivatives, but the rectangular waveform of the thrust makes integration under the curve trivial. Which allows easy control of delta V.

Jerk and Snap are important considerations for Roller Coaster track design, for the comfort of the riders and their necks. Not normally part of navigation.

 

[GregLocock]

Hmm, did I write a post that got deleted? This may be too simplistic but it builds on the above.

Either - develop a filter for your jerk and snap signals that gets rid of the rubbish, or, synthesise jerk and snap from the known physical properties of your systems. Effectively, say for jerk due to firing the thruster, you'd need to know the mass of the thruster, and the compliance of its mounting system, and the thrust time signal. That's a SDOF system. In other words drive your program with synthesised signals if the real ones are too clumsy or expensive. This is used in cars to eliminate expensive sensors. For instance we know what the map of torque vs speed vs fuel injector pulse width is. So we don't need to measure torque, we already know the others. Cross correlating between the signals tells us when the system isn't behaving as expected and throws an engine code.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[charliealphabravo]

So you've put a button on it but you don't know what it does?

Anyway, I'd think that jerk and snap can be routinely ignored when you have the luxury of manual docking or when you have an independent means of checking position (laser ranging, photogrammetry, or whatever) that basically allows you to correct for (or ignore) errors induced by jerk/snap.

I imagine jerk/snap becomes more critical in automated docking situations where you might not have an independent means to verify position (realtime) or where you hope to keep the docking/positioning system as simple as possible (no sensors, accelerometers only).

In that case, I'd think the physical docking/mating hardware must be physically large enough, and flexible enough, to accommodate the expected errors in jerk/snap, or alternatively you must increase the quality of the accelerometer until your positional error is within the physical capability of the docking hardware to correct for, or alternatively you must reduce/smooth? the thrust release to bring the jerk/snap effects within the tolerance of your docking system.

This is not really my thing so the terminology might be off and sorry for all the ninja edits.

 

[rb1957]

what would happen if you told the controller that jerk and snap were "zero".

I'd've thought (and probably wrongly) that distance and velocity would be most important to a rendezvous ?

another day in paradise, or is paradise one day closer ?