Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Tuning of Online Feedback Optimisation Controllers for Process Control
Designing controllers for industrial systems, such as compressor stations or energy distribution networks, is usually challenging because of the complexity of the processes that are involved. A lot of existing controllers require solving difficult constrained optimization problems to find the optimal operating point. Online Feedback Optimization is a method for steering a nonlinear system to the solution of an optimization problem without explicitly solving nonlinear constrained optimization problems. It can be used to optimally share the load between gas compressors, optimally curtail renewable energy if necessary, or generally to operate a physical system at the optimal solution of an optimization problem. The idea behind this project is to analyze how Online Feedback Optimization should be implemented to facilitate its use in industry and expand the range of possible applications.
Online Feedback Optimization as a novel control method has recently gained traction in several research communities, is investigated by companies, and deployed in the European power grid. Even though it is now used in both industry and academia, the tuning of its parameters still requires in-depth knowledge of the method and characteristics of the physical system. By analysing the impact of tuning on the performance of the controlled system, the semester project will bridge the gap between the theory of feedback optimization and its implementation.
Existing applications of feedback optimization usually assume that the dynamics of the optimization algorithms are much faster than the dynamics of the controlled system. Lately, people started using Online Feedback Optimization for industrial systems characterised by complex yet slow dynamics. The goal of this project is to push the limits of applications of feedback optimization by providing guidelines for tuning the controllers for systems with slow dynamics. Good guidelines on tuning of parameters of Online Feedback Optimization would help expanding the range of application of the method to more industrial systems.
The semester project has two main tasks:
- T1. Computational analysis of the influence of tuning parameters and sampling time of the controller using provided models
- T2. Adjustment of tuning parameters to individual inputs in a multi-input system.
The two tasks are numerical, so a good knowledge of software such as Matlab is essential. The project can also be adapted on the run if new interesting research directions arise. If the results are promising they can be turned into a publication.
The semester project is well suited for students who enjoy applied mathematics and technology transfer:
- The student will learn the novel OFO controller;
- The student will work with realistic process control applications;
- The student will get an opportunity to expand their academic background.
The project could be done in person at the Automatic Control Laboratory, hybrid, or completely remotely depending on the current ETH rules. Most importantly, we can change between these forms whenever needed.
Finally, if the results are promising they can be turned into a publication.
Existing applications of feedback optimization usually assume that the dynamics of the optimization algorithms are much faster than the dynamics of the controlled system. Lately, people started using Online Feedback Optimization for industrial systems characterised by complex yet slow dynamics. The goal of this project is to push the limits of applications of feedback optimization by providing guidelines for tuning the controllers for systems with slow dynamics. Good guidelines on tuning of parameters of Online Feedback Optimization would help expanding the range of application of the method to more industrial systems.
The semester project has two main tasks:
- T1. Computational analysis of the influence of tuning parameters and sampling time of the controller using provided models - T2. Adjustment of tuning parameters to individual inputs in a multi-input system.
The two tasks are numerical, so a good knowledge of software such as Matlab is essential. The project can also be adapted on the run if new interesting research directions arise. If the results are promising they can be turned into a publication.
The semester project is well suited for students who enjoy applied mathematics and technology transfer: - The student will learn the novel OFO controller; - The student will work with realistic process control applications; - The student will get an opportunity to expand their academic background.
The project could be done in person at the Automatic Control Laboratory, hybrid, or completely remotely depending on the current ETH rules. Most importantly, we can change between these forms whenever needed.
Finally, if the results are promising they can be turned into a publication.
The objectives of the project are:
- Understanding the interactions between the parameters of feedback optimisation and the implementation of the algorithm
- Analysis of the impact of the parameters of feedback optimisation on dynamic properties of the closed-loop system
- Providing tuning guidelines to enable application of feedback optimisation to systems with slow dynamics
The objectives of the project are: - Understanding the interactions between the parameters of feedback optimisation and the implementation of the algorithm - Analysis of the impact of the parameters of feedback optimisation on dynamic properties of the closed-loop system - Providing tuning guidelines to enable application of feedback optimisation to systems with slow dynamics
Marta Zagorowska (mzagorowska@control.ee.ethz.ch)
Lukas Ortmann (ortmannl@control.ee.ethz.ch)
#IfA #NCCR
Marta Zagorowska (mzagorowska@control.ee.ethz.ch) Lukas Ortmann (ortmannl@control.ee.ethz.ch) #IfA #NCCR