Reinforcement Learning Based Constrained Control Applications

The main issue of the proposal hinges on the application of Data driven approaches for Model Predictive Control (MPC) applications. MPC is an optimization based control methodology well known in literature which is extremely popular when considering constrained control problems. The main drawback of MPC is the need of an accurate model plant to solve an optimal control problem on-line, under real-time constraints.
Data-driven control approaches mitigate the issue of model construction and tuning in two ways. In a more traditional way, using black (or gray) box system identification to come up with linear or nonlinear (such as neural) prediction models from data (model-based MPC). Alternatively, learning directly the control law without going first through an open-loop prediction model (model-free MPC). The main idea is to combine learning algorithms with control techniques like linear quadratic regulation and MPC, so to synthesize an optimal policy for the real process directly from data, without going first through time-consuming modeling and MPC tuning efforts, therefore reducing the overall numerical burdens to ensure safety, robust performance and overall stability.

Faculty Supervisor:

Luis Rodrigues

Student:

Partner:

University of Calabria

Discipline:

Engineering

Sector:

Education

University:

Concordia University

Program: