Register now After registration you will be able to apply for this opportunity online.
Primal-dual Feedback Optimization for Power Grid Operation
Feedback optimization is emerging as an important control method for modern power systems, thanks to its robustness and ability to steer the grid to an efficient operating point. In this project, we will design and evaluate novel feedback optimization schemes, based on Lagrangian dual methods, which can handle safety constraints and promise improved robustness to measurement noise.
Feedback optimization is a modern control method, where optimization algorithms are directly interconnected in closed-loop with a physical plant. The advantage is that the system is driven to an efficient point defined by an optimization problem, rather than to an already known reference; in this way, the controller can also quickly react to unpredictable changes in the environment. For instance, feedback optimization has been succesfully deployed in the control of power grids, where the ongoing energy transition is posing important new challenges, due to the volatility of renewable resources, distributed production and unknown power demand.
Feedback optimization is a modern control method, where optimization algorithms are directly interconnected in closed-loop with a physical plant. The advantage is that the system is driven to an efficient point defined by an optimization problem, rather than to an already known reference; in this way, the controller can also quickly react to unpredictable changes in the environment. For instance, feedback optimization has been succesfully deployed in the control of power grids, where the ongoing energy transition is posing important new challenges, due to the volatility of renewable resources, distributed production and unknown power demand.
The goal of this project is to extend the known feedback optimization schemes, based on primal dynamics as gradient descent, to primal-dual dynamics (i.e., methods based on Lagrangian duality as ADMM). The advantage is that output constraints and nonsmoothness can be easily handled, in contrast to classical feedback optimization.
We aim at studying analytically the stability of primal-dual feedback optimization schemes, and at
evaluating their performance in the control of power systems, in terms of robustness, speed, capacity of dealing with nonconvexity and handling constraints.
The student will be introduced to and taught on the concept of feedback optimization, and given a simple power system model and control problem that can be solved via primal-dual feedback optimization. After this initial phase, the student will search the literature for the most suitable models and problems, and will implement numerically the controllers to evaluate performance.
The project can be adapted on the run if new interesting research directions arise. If the results are promising, they can be turned into a publication.
The goal of this project is to extend the known feedback optimization schemes, based on primal dynamics as gradient descent, to primal-dual dynamics (i.e., methods based on Lagrangian duality as ADMM). The advantage is that output constraints and nonsmoothness can be easily handled, in contrast to classical feedback optimization. We aim at studying analytically the stability of primal-dual feedback optimization schemes, and at evaluating their performance in the control of power systems, in terms of robustness, speed, capacity of dealing with nonconvexity and handling constraints.
The student will be introduced to and taught on the concept of feedback optimization, and given a simple power system model and control problem that can be solved via primal-dual feedback optimization. After this initial phase, the student will search the literature for the most suitable models and problems, and will implement numerically the controllers to evaluate performance.
The project can be adapted on the run if new interesting research directions arise. If the results are promising, they can be turned into a publication.