Simulink Design Optimization toolbox: Estimate parameters using several experiment data

Hello, All,

I'm trying to fit my simulink model's parameters to a real product.

Here, the product's characteristics varies due to its length (L).

So, I had measured the characteristics changing the length. (Experiment A, B, C)

I want to estimate other parameters, which fit to all experiment results, using Simulink Design Optimization.

For example)

Estimation A: fix L = 100, fitting to Experiment A

Estimation B: fix L = 300, fitting to Experiment B

Estimation C: fix L = 700, fitting to Experiment C

Here, other parameters must be common in all estimation.

How can I realize the estimation?

Answer 1

[Yifeng Tang]

Hi Kouta,

One solution I can think of:

Make three copies of your model, in one same Simulink model, all use the same parameters to be estimated, but different L. Now each copy will represent experiment A, B, or C. Then, take the difference between the output of each model and the reference signal, and then take the 2nd-norm sum of the three differences. Use the last quantity as the output to be optimized/minimized, by (for example) setting its reference to a zero signal, and run the SDO workflow. I believe this way the SDO will try to find a set of parameters that will minimize the difference between output and reference data for all three sets of experiements at the same time.

Hope this helps.

Yifeng