This thesis explores a different approach in the use of MPC applied to grid-tied converters, which aims to improve their performance, either during abnormal operating conditions or by exploiting the full potential of the converter in steady-state. The design of the MPC is firstly introduced in a single-phase converter. Although this device is mostly used in low voltage and low power applications, this analysis brings insight into the design of more complex converter topologies. Besides, the study of the single-phase converter presents interesting challenges due to the pulsating power of the single-phase system. It is found that the nonlinear behaviour of the converter, even at a high level of abstraction, leads to a formulation of the MPC that requires the use of nonlinear solvers. Consequently, the time-to-resolution of the optimisation problem becomes cumbersome, and a strategy to reduce the overall computation burden is sought.
Given that the steady-state operation of grid-tied converters is based on sinusoidal trajectories, a simplification approach where the converter is linearised over LTV trajectories is introduced. This approach is not common in the control of power converters but it has been successfully implemented in other fields such as in the automotive industry. The results obtained prove the good performance of this approach and the controller is validated in a laboratory test-bench. The benefits of the MPC strategy applied to a MMC-HVDC converter are also discussed. The algorithm is applied to a single-phase structure to facilitate its analysis. The results considering HVDC-scaled parameters suggest potential benefits in terms of hardware optimisation. It is shown that the predictive controller can enhance the operating area of the converter by naturally combining different known harmonic injection techniques.
Finally, this document also presents a preliminary investigation of the coordination of the different curtailment mechanisms integrated into a RE system using an MPC algorithm.