Power electronic systems are often underutilized with conventional control solutions and are being operated in a suboptimal manner. An attractive control alternative is model predictive control (MPC) due to its numerous advantages, such as explicit inclusion of design criteria and restrictions, design versatility, and inherent robustness. Thanks to these features, MPC can bring significant benefits by improving performance metrics (e.g., current distortion, power losses, settling time), and/or reducing the hardware requirements (or, equivalently, by fully utilizing the existing hardware).
Motivated by the above, the objective of this tutorial is to show the performance improvement that can be achieved with control algorithms designed in the framework of MPC. To this aim, different MPC methods (control problem formulations) will be discussed and analyzed. Moreover, it will be shown how these MPC strategies can bring tangible improvements, such as lower harmonic distortions, hardware reduction, increased efficiency, or increased output power. Furthermore, implementation-related issues will be analyzed, while methods to tackle them will be presented. In doing so, insight into the MPC-based algorithms and the associated challenges will be provided.
Overall, the tutorial aims at providing a balanced mix of theory and application-related material. Special care is taken to ensure that the presented material is intuitively accessible to the power electronics practitioner. This is achieved by augmenting the mathematical formulations by illustrations and simple examples.
By the end of the tutorial, the attendees will:
- understand the standard and emerging MPC methodologies, their design and real-time implementation as well as suitable embedded system architectures to implement them and to solve the underlying optimization problems,
- be able to understand what design options exist that maximize the system performance and how MPC-based controllers are to be designed to outperform conventional control techniques and to push the system performance to its physical limits, and
- appreciate the industrial relevance and benefits of MPC-based controllers.