What is the best way with small error to solve ODE depends on experimental data(numerical data) ??

I got 'displacement and accleration' data from device.

And I taked out 'velocity' with kalmanfilter method through those two datas above.

My objective is 'https://en.wikipedia.org/wiki/Bouc%E2%80%93Wen_model_of_hysteresis'

optimizing the parameters of boucwen model.

So, I tried to solve it using ODE45 and 'lsq curvefit'.

When I did this with sine velocity data not numerical data, it be optimized well.

But when I put 'experimental data(like 1000x1 vector) into boucwen model', it didn't work well.

like Not optimizing every parameters.

I've already uploaded my code in matlab community a lot. So I dont want to bother you by that code that I made.

Now, I'd like to ask what the good way in this case to solve.

I will apreciate if I teach a way or show me similiar case...

Thank you so much.

Answer 1

[Amir Kleiner]

You should try using the optimization app.

You can either open it from the "apps" tab in your workspace, or from the code.

Try checking out the documentation for optimtool.