Has anyone constructed a working Simulink model of the Ultra-Local Model and accompanying iPID controller as presented by Michel Fliess and others?
In particular I am somewhat stuck on how to best implement the estimation of F in the pivotal y' = F + alpha*u equation, and where exactly to place alpha in the completed controller.
At this time I do have a structure which uses the derivative of the output, as well as the previous Plant input, via a ZOH, and an Alpha gain which does reduce a given NMP Plant into a single integrator. However I am unsure that this is in fact the correct way to accomplish this, as further sections of Fliess' papers provide the details of a sliding window integral estimator, which my model does not have.
Can anyone who has succeeded in this endeavor shed some light on the appropriate way to design and verify this system?
Hi Aleksandr,
I see, you still do not have response. The problem may be caused by the wrong citation in the title - as you write in the text, the author of iPID is Michel Fliess.
As you write and as e.g. described in
https://hal-polytechnique.archives-ouvertes.fr/hal-00828135/en/
reconstruction of F (equivalent input disturbance) leads to sliding window integral estimator. For me, the original papers about iPID are not easy to read. So, I preferred to derive the control equations by myself. Alpha may be considered as the plant gain and in controller it should be in the denominator. In tuning its equations, choose firstly alpha=1
I do not know, how far are you bound to this method - in our works we prefer approaches closer to disturbance observer known in motion control that may be interpreted quite easily - see e.g. https://www.researchgate.net/publication/269270731_Noise_attenuation_motivated_controller_design._Part_I_Speed_control
As one of the main advantages we may mention known influence of the filter parameters on the stabilizing controller gain.Of course, a detailed comparison of both approaches would be helpful...
Best wishes, mh
Conference Paper Noise attenuation motivated controller design. Part I: Speed control
- Badges
- Science topic