On March 31, we organized a scientific seminar on “Active Set Prediction using Machine Learning for The Complexity Reduction in Nonlinear Model Predictive Control“. The scientific seminar was led by Martin Klaučo, associate professor at the Slovak University of Technology in Bratislava. The scientific seminar was organized in the framework of the FrontSeat project, as part of the seminar series on “Research Seminar on Smart Cybernetics”.
Abstract:
We use Machine Learning (ML) methods to simplify the Nonlinear Programs (NLPs) arising in Nonlinear Model Predictive Control (nonlinear MPC, NMPC). Since every solution to an NMPC problem is directly connected to the set of active and inactive constraints, we propose to use predictions about these sets to reduce the complexity of the NLP to be solved. Recently, several approaches on how to determine active and inactive constraints have been proposed for both, linear and nonlinear MPC. Especially for the NMPC case, accurate predictions are still under research. Here, we propose to use Machine Learning methods to predict active sets before solving the NLP. Especially classification algorithms are simple enough to be evaluated online, i.e., during runtime of the controller, and can be trained to a high accuracy, qualifying as suitable candidates for an application to NMPC. The results are evaluated using extensive simulation for a Continuous Stirred-Tank Reactor.
This project has received funding from the European Union’s Horizon under grant no. 101079342 (Fostering Opportunities Towards Slovak Excellence in Advanced Control for Smart Industries).
0 Comments