Defence Xiaoyan Dai: "Nonlinear Data-driven Control-Output Feedback and Online Methods"

Promotors: Prof. Claudio De Persis, Prof. Nima Monshizadeh

Abstract: Learning controllers from data is of utmost importance and a fascinating topic, with foundations in both control theory and data science. The thesis contributes to analytically correct data-driven control methods with computationally efficient data-dependent LMIs whose feasible solutions provide controllers and performance certificates. In the first part of the thesis, we consider the output feedback control problem of poorly known systems from input-output data and some prior knowledge. We derive optimal output feedback control methods for unknown linear systems that bypass model identification or a separate design of a state observer and a state feedback controller. Then, we develop data-driven output feedback control methods for a general class of nonlinear systems by enforcing a closed-loop system dominated by stable linear dynamics. To this end, we use a growth condition on the basis functions and input-output data. Notably, we extend the methods to cope with the cases with input-output measurement noise and an incomplete dictionary of basis functions. Finally, we consider the online control of input-affine nonlinear systems via time-varying SDPs. Both model-based and data-driven solutions are derived where control gains are correctly adapted. We derive compact conditions that certify recursive feasibility from data.

Dissertation