Constrained non linear optimization in my VI doesn't work.

[getdlts]

Hello? everyone,

I have tried to make my own optimization program based on a example provided by NI, "Optimization and Calibration with Simulation - Calculate Parameter". The provided optimization program  generates a spring simulation with Gaussian noise and fitting it.

I have tried to apply a complex exponential function to it. However, although there are no errors, the "Optimize" button does not optimize at all(actually graph vibrates a little but, it became original shape). I do not know what the problem is. Please help me.

I attach my optimizer.

[altenbach]

• Does your model give any reasonable output for a given set of parameters?
• Why is your inner calculation carried out in SGL? (That severely limits the resolution of your calculations. Look at all the coercion dots!)

[altenbach · Accepted solution]

---

altenbach wrote:

• Why is your inner calculation carried out in SGL? (That severely limits the resolution of your calculations. Look at all the coercion dots!)

---

If you change the representation of the all the orange terminals inside "Correlation function (SubVI).vi" from SGL to DBL, the optimization succeeds. Try it!

(note that you should probably set the "spring constant(start&fit)" upper limit a bit higher (e.g. 10) for more reliable results. You are constraining too much!)

 

[getdlts]

Thank you very much Altenbach, it is really working! 

If it was not for your help, I would have been struggled for solving that problem.

[getdlts]

Thank you very much Altenbach, it is really working! 

---

altenbach 이(가) 작성함:

---

altenbach wrote:

• Why is your inner calculation carried out in SGL? (That severely limits the resolution of your calculations. Look at all the coercion dots!)

---

If you change the representation of the all the orange terminals inside "Correlation function (SubVI).vi" from SGL to DBL, the optimization succeeds. Try it!

(note that you should probably set the "spring constant(start&fit)" upper limit a bit higher (e.g. 10) for more reliable results. You are constraining too much!)

 

---

If it was not for your help, I would have been struggled for solving that problem.