We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Systems and Control

Title: A Heuristic for Dynamic Output Predictive Control Design for Uncertain Nonlinear Systems

Authors: Mazen Alamir
Abstract: In this paper, a simple heuristic is proposed for the design of uncertainty aware predictive controllers for nonlinear models involving uncertain parameters. The method relies on Machine Learning-based approximation of ideal deterministic MPC solutions with perfectly known parameters. An efficient construction of the learning data set from these off-line solutions is proposed in which each solution provides many samples in the learning data. This enables a drastic reduction of the required number of Non Linear Programming problems to be solved off-line while explicitly exploiting the statistics of the parameters dispersion. The learning data is then used to design a fast on-line output dynamic feedback that explicitly incorporate information of the statistics of the parameters dispersion. An example is provided to illustrate the efficiency and the relevance of the proposed framework. It is in particular shown that the proposed solution recovers up to 78\% of the expected advantage of having a perfect knowledge of the parameters compared to nominal design.
Subjects: Systems and Control (eess.SY); Machine Learning (cs.LG)
Cite as: arXiv:2102.02268 [eess.SY]
  (or arXiv:2102.02268v1 [eess.SY] for this version)

Submission history

From: Mazen Alamir Prof [view email]
[v1] Wed, 3 Feb 2021 20:01:25 GMT (783kb,D)

Link back to: arXiv, form interface, contact.