|Classic control techniques are typically designed to achieve stability and good performance, but not constraint satisfaction. Model predictive control is widely adopted in practice since it is able to deal with hard constraints. For the control problem of large-scale systems however, centralized model predictive control is impractical due to a number of limitations, including the computational burden and limited communication, which motivate our research of distributed and hierarchical model predictive control methods.
In this project, we aim to develop new methods and algorithms for distributed and hierarchical model-based predictive control of large-scale, complex, networked systems. The algorithms should be able to deal with couplings in the dynamics and the constraints between subsystems, and guarantee feasibility and stability of the closed-loop system.
The research is based on different optimization techniques, which could be implemented in a distributed or hierarchical architecture. The properties of the controller are analyzed based on the convergence, performance, and feasibility properties of the optimization tools.