LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

optimization - where to start

Afternoon Gurus,

 

I have inherited some Excel code which I want to convert to Labview. it uses the Solver function to optimize a system.

 

I have some input parameters which are passed to a process which is non-linear and cannot be described by a single function. The process outputs a couple of values which I compare with measurement. I want to optimise the input parameters so the output matches the measured, ideally of course! In excel, Solver looks to minimize the fraction error between the measured and modelled  ((Modelled - measured)/measured)^2. It doesn't know that, it is just minimising a value in a cell by varying the input cells.


Which set of VI's do I need to replicate this system?

 

Many thanks,

 

Paul

"When I read about the horrors of drinking, I gave up reading"
0 Kudos
Message 1 of 6
(3,071 Views)

On the Mathematics Block Diagram Palette, there is an Optimization sub-Palette.  Optimization is a somewhat complex topic -- you should "know what you are doing" before blindly jumping in and choosing a method and options, particularly if dealing with non-linear problems.

 

Bob Schor

0 Kudos
Message 2 of 6
(3,052 Views)

Thanks for that. This is indeed - complex.

 

For my application, I have a function f(x), which is bounded between upper and lower limits, and has a single peak at X0. I process the function in a non-linear way and end up with a modified function. I then sum that modified function up over the bounds and end up with a single value, which I can compare with a measured value. I want to mininize the relative error between the measured and modelled by optimising X0.

 

I shall carry on thinking and Googling.

 

Paul

"When I read about the horrors of drinking, I gave up reading"
0 Kudos
Message 3 of 6
(2,900 Views)

I have had a lot of success using the "Constrained Non-Linear Curve Fit" and "Non-Linear Curve Fit" VI's located under Mathematics->Fitting. You will create a VI that is used as the non-linear function and outputs your error value, so it should translate very nicely from the Excel method.

0 Kudos
Message 4 of 6
(2,839 Views)

Thanks,

 

I will have a look, but the issue is that I am not actually curve fitting. But I will see if I can use them.

 

Paul

"When I read about the horrors of drinking, I gave up reading"
0 Kudos
Message 5 of 6
(2,823 Views)

Hi All,

 

Decided not to give up on this. Noticing a minor wiring error from a few days ago (max and min the wrong way around.....) I got this working. Attached is a ppt of the wiring diagram. Key is to: 1) use the template for the objective and constrains vi; 2) don't delete the unused items in the objective and constrains vi;

 

One note of caution. When I set the start point near to or at the max value, it did not find the minimum

"When I read about the horrors of drinking, I gave up reading"
0 Kudos
Message 6 of 6
(2,798 Views)