Embedded learning and optimization for the next generation of smart industrial control systems





ESR9 – Politehnica University Bucharest

 

PhD Position in Embedded nonlinear MPC using learning techniques





 

Institution:
University Politehnica Bucharest, Romania

 





 
Content:
  1. Introduction
  2. PhD Project Descriptions
  3. Supervisors and Main Contacts
  4. Candidate Profiles
  5. Application
  6. Marie Curie Eligibility Criteria In Short

 


 

Introduction
 
The PhD position is part of the European Training Network ‟ELO-X – Embedded Learning and Optimization for the neXt generation of smart industrial control systems”. ELO-X will recruit altogether 15 PhD fellows at 6 research universities and 5 international companies from 5 European countries, who will meet regularly during exchange visits, training events, workshops, and summer schools organized by the network.
 
The position at University Politehnica Bucharest has a strong methodological focus in the field of computational control and mathematical optimization. It is based in the Optimization, Learning and Control Lab headed by Prof. Ion Necoara. The aim is the development of advanced optimization and learning-based control methods and open-source software and their application to industrially relevant problems. While these methods are generic and applicable in several branches of engineering, they shall be tested and used in close cooperation with the the other ELO-X PhD fellows, in particularly with those who are based in ODYS, KUL and EPFL- during mutual exchange visits of several months duration.
 
PhD Project Description
 
PhD Project: Embedded nonlinear MPC using learning techniques:
The aim of this PhD position is to develop and use advanced learning and optimization methods that are able to address the control design of nonlinear systems. For the control of linear systems in a receding horizon fashion (so-called model predictive control – MPC) the corresponding optimization problem is typically convex, for which very e”cient optimization algorithms exist. However, when the system is nonlinear the corresponding optimization-based MPC control problem is highly nonconvex. The structure of such MPC problem is very similar to the one encountered in training of deep learning networks. An optimization algorithm depends on several parameters (e.g., stepsize, penalty parameters, etc). The goal is to find the best parameters choice. Similarly, searching the parameter space of deep architectures is a cumbersome task, but learning algorithms, such as those for deep belief networks, have recently been proposed to tackle this problem with notable success. We plan to search the parameters of nonconvex optimization algorithms for MPC using deep learning. Moreover, though the optimization-based control problem has a similar representation as in training learning networks, it has specific structure. For example, we may know the reference signal we need to track with our embedded controller, which can help a learning algorithm. Moreover, in the optimization-based control problem we usually have a convex objective function but with nonconvex constraints, with or without sparsity, some variables are random, etc. In this PhD thesis we want to combine the best of these worlds, control, optimization and learning, to develop fast nonconvex optimization algorithms tailored for embedded MPC of SICS. We will analyse the structure of the nonlinear control problem and identify the characteristics that can help a learning algorithm (convexity, sparsity, stochasticity) to find the best parameters of the optimization algorithm. The optimization algorithms produced by the learning framework will also take into account limited computational resources, making them adequate for embedded frameworks including GPUs.

Timeline and remuneration:
The start time is September 2021. The PhD projects last for the duration of three years, and are carried out at the University Politehnica Bucharest. The PhD years include at least one longer visit – a so called ”secondment” – between one and six months to another group in the ELO-X network, depending on the project needs and the scientific interests of the PhD fellows. The first year is mainly dedicated to studying and getting acquainted with the relevant state of the art in optimization, model predicitive control and deep learning, the second year focuses on method development, and the third year on application problems and writing the thesis. The remuneration is generous and will be in line with the EC rules for Marie Curie grant holders. It consists of a salary augmented by a mobility allowance, resulting in a net monthly salary of about 1500-2000 Euro depending on family status.
 
Candidate Profiles
 
Ideal candidate has a master degree in one of the following disciplines or a related field: control, numerical mathematics or computer science. He/she should have a good background or interest in mathematical optimization, machine learning, control of dynamical systems, and programming (Matlab, python), as well as a desire to contribute to the development of open-source software and the success of real-world experiments. Proficiency in English is a requirement. The positions adhere to the European policy of balanced ethnicity, age and gender. Both men and women are encouraged to apply.
 
Supervisors and Main Contacts
 
Supervising team at the University Politehnica Bucharest
 
  • Prof. Dr. Ion Necoara (head of Optimization, Learning and Conrtrol Laboratory)
  • Daniela Lupu (senior PhD student working on Big Data Optimization)
 
Main Contacts at the ELO-X Partner Groups which could host secondments:
 
  • ODYS: Prof. Alberto Bemporad;
  • EPFL: Prof. Colin Jones;
  • KU Leuven University: Prof. Panos Patrinos
 
Application
 
To apply, send an email to elo-x@imtek.uni-freiburg.de in form of one single PDF attachment containing all contents or links (any other information within the email will not be processed). Subject of your email should be: “ELO-X PhD Application – ESR 9”.

Please include, in your single PDF document, the following items in this order:

  1. A cover letter incl. statement of research interests and career goals (max. 2 pages);
  2. An academic CV;
  3. Contact details of at least two referees incl. phone numbers and emails;
  4. Your diplomas and transcript of course work and grades;
  5. Sample of technical writing (publication or thesis);
  6. Proof of English language proficiency test results.

 

Please send your application before May 7, 2021.

 

Note that your PDF will be forwarded to several people in the ELO-X institutions and that in particular all Supervisory Board members of ELO-X will have access to your application material. If you want to apply to more than one ELO-X position, please create and send separate PDFs.
 
Marie Curie Fellowship Eligibility Criteria in Short
 
To be eligible, you need to be an “Early Stage Researcher” (ESR) i.e. simultaneously fulfill the following criteria at the time of recruitment:
  1. Nationality: you may be of any nationality.
  2. Mobility: you must not have resided or carried out your main activity (work, studies, etc) in Romania for more than 12 months in the 3 years immediately prior to your recruitment under the ELO-X project.
  3. Qualifications and research experience: you must be in the first 4 years of your research career after the master degree was awarded.