Data-driven prediction and control with stochastic data: a system identification perspective (Yin Mingzhou, Automatic Control Laboratory, ETH Zürich)

Systems & Control Seminar

• Donnerstag 21.09.2023, 15:00 Uhr
• Raum A247, Gebäude 3403, Appelstr. 11

Abstract

Recently, direct data-driven prediction has found important applications for controlling unknown systems, particularly in predictive control. Such an approach provides exact prediction using behavioral system theory when noise-free data are available. This presentation discusses extensions of the approach to stochastic data by leveraging tools from system identification, the "classical" data-driven control paradigm. In particular, it tries to answer the following questions:

1. What are the paths going from noise-free data to stochastic data?
2. Is there an optimal predictor we can use?
3. Can we quantify the prediction error and use it to robustify the controller?
4. Where is the observer in data-driven predictive control?
5. Does the algorithm hold in practice with nonlinearity?

Biographical information

Mingzhou Yin is a doctoral student supervised by Prof. Roy S. Smith in the Automatic Control Laboratory at ETH Zurich since February 2019. His research interests include data-driven modeling, simulation & control, sparse learning theory, system identification with subspace and regularized methods, model predictive control, and periodic system theory. He received his MSc degree cum laude in control & simulation at the Faculty of Aerospace Engineering, Delft University of Technology, the Netherlands in 2018. His master's thesis research is on envelope-protected non-linear control of over-actuated aircraft in collaboration with Lockheed Martin. He received a joint bachelor’s degree in Mechanical Engineering at Shanghai Jiao Tong University, China, and the University of Hong Kong, China with first-class honours in 2016. He was the recipient of the IEEE Control Systems Society Swiss Chapter Young Author Best Journal Paper Award and the Systems Identification and Adaptive Control Technical Committee Outstanding Student Paper Prize in 2023.