Parameter identification, if the noisy measurements have a bias, too

[Peter230655]

I may have fond a solution.
What I did was this:
I doubled the set of the eoms, where the second set is simply a shifted copy of the first set.
No matter what bias I use, it finds the dampening and the spring parameters very accurately. ( made at least 25 trials, likely more)
For two reasons, I am unsure I got it:

• the estimation of the bias is very poor.
• The speeds are absolutely crazy,

[Peter230655]

I made a PR with the simulation.