LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Optimization algorithm for measured signals

Hi,

I'm wondering whether there are any Labview implementations of an efficient optimization algorithm for "real-world" signals (i.e. a control problem rather than a pure math problem). The goal is to automatically adjust a set of physical parameters (e.g. voltages, motor positions or the like), which maximize a measured "objective" signal. An intuitive example would be the alignment of a laser beam through a pinhole using a motorized mirror and the feedback is provided by a photodiode or powermeter behind the pinhole.

 

I thought about using one of the optimization VIs, but I think they would have an inherent problem with an objective function derived from a "real-world" measurement because of the noise and/or drift. Maybe it's possible to control the step size and tolerance to get a reasonable result but I couldn't figure out a robust procedure so far.

 

I'd rather treat it as a control problem with an approach similar to a PID controller but instead of locking the "process variable" to a setpoint on a monotonous curve, it should rather lock it to an extremum (which can drift over time). I'm not very familiar with control algorithms and I don't even know whether it's possible at all. Does anyone know a suitable solution?

0 Kudos
Message 1 of 5
(1,349 Views)

There is a PID toolkit that should come with LabVIEW full or professional.

 

http://sine.ni.com/nips/cds/view/p/lang/en/nid/209054

0 Kudos
Message 2 of 5
(1,339 Views)

I know, but a PID controller can't lock to an extremum...

0 Kudos
Message 3 of 5
(1,334 Views)

There is also a modulation toolkit that might help you make a phase locked loop implementation.

0 Kudos
Message 4 of 5
(1,255 Views)

That's the same as a limited PID controller (using only P or PI) and the signal is not periodic, so I don't see how I can extract a phase from that to lock onto...

0 Kudos
Message 5 of 5
(1,213 Views)