Adaptive Model Predictive Control for MMC Current Control using Offline Gain Approximation and Online Weighting Adaptation
Authors:
- Rahul Rane
- Victor Daniel Reyes Dreke
- Wilhelm Winter
- Aleksandra Lekić
Published in: 2026 International Conference on Smart Energy Systems and Technologies (SEST)
Abstract:
This paper proposes an adaptive model predictive control (AMPC) framework for inner current control of modular multilevel converters (MMCs) in high-voltage direct current (HVDC) systems. The approach reduces the online computational burden of conventional MPC by precomputing the unconstrained feedback gain matrix Kmpc offline over a range of control effort weights R and approximating its elements using polynomial regression. This enables real-time evaluation of the control gain without repeated matrix inversions. The optimal control law is governed by the relative weighting between the state penalty Q and the control effort penalty R in the quadratic cost function. In this work, Q is kept fixed, while R is adapted online to regulate the trade-off between transient response performance and control effort. A precomputed lookup table, derived from closed-loop step-response simulations over a grid of initial states and reference transitions, maps operating conditions to the smallest value of R that satisfies the converter voltage headroom constraint at each sampling instant. Simulation results demonstrate that the proposed framework closely reproduces the performance of conventional MPC while significantly reducing computational complexity, making it suitable for real-time implementation in high-speed power electronic applications.
