Portfolio · Research
My work develops continuous-time multiparametric methods for dynamic optimization and explicit model predictive control.
I develop explicit control solutions directly from differential-equation models. The goal is to avoid unnecessary time discretization while keeping online control fast.
I study when constraints become active or inactive, how those switching times depend on the initial state, and how the state space can be divided into regions with the same control structure.
Some region boundaries are exact hyperplanes. Others are curved because the event that defines the boundary moves with the initial state. My work develops analytical tests and certified numerical methods for both cases.
The methods are being connected to the PAROC workflow for process control, including setpoint tracking, disturbance rejection, and intensified process applications.
In benchmark studies, the continuous-time map remained compact while the number of discrete-time regions grew as the sampling grid was refined. The continuous-time solution also provides the switching times directly.
Extending the framework to maintain feasibility and constraint satisfaction under bounded disturbances and model mismatch.
Extending continuous-time multiparametric ideas to systems with changing modes and discrete decisions.