Model Predictive Control Strategies in Power Electronics

Model predictive control (MPC) has emerged as a versatile and high-performance methodology for the regulation of power electronic converters and motor drives. At its core, MPC employs an explicit model of the power stage to forecast future system behaviour over a finite horizon and to optimise a cost function at each sampling instant. This predictive framework enables direct handling of multivariable interactions, constraints on voltages, currents and switching actions, and the incorporation of complex control objectives such as minimising losses or harmonic distortion. Two principal categories are distinguished: direct MPC, which computes and applies switching states without a subsequent modulation stage, and indirect MPC, which embeds a modulation algorithm to enforce a fixed switching frequency. Recent advances have addressed classical limitations of MPC, notably its computational burden and reliance on accurate models. Strategies such as space vector enumeration, duty-cycle modulation or extended deadbeat algorithms reduce online calculation, while neural-network-based surrogates and model-free identification schemes offer robustness to parameter uncertainty. Applications span grid-tied inverters, multilevel converters, static synchronous compensators and traction drives, with benefits including fast dynamic response, low total harmonic distortion and flexible constraint handling. The global impact of MPC in renewable integration, electric mobility and smart grids underscores its trajectory from a laboratory demonstration to widespread industrial adoption. Nevertheless, real-time implementation challenges persist, motivating ongoing research into hardware acceleration, solver design and adaptive model structures.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

Recent work has refined MPC for multilevel voltage source inverters by combining discrete space vector modulation with optimised switching-sequence selection. In three-level neutral-point-clamped converters, an external predictive layer evaluates a large set of basic and virtual vectors, distinguishing negative and positive sequences to maintain capacitor voltage balance without weighting factors. An extended deadbeat algorithm then selects the optimal vector to achieve sub-2 % current distortion while halving computational load compared with conventional schemes. Another study addresses the computational challenge of cascaded H-bridge converters by training a neural network as a predictive-control surrogate. Implemented on a field-programmable gate array and validated in hardware-in-the-loop, this approach delivers near-optimal switching decisions in real time, enabling high-level multilevel operation with low latency. Foundational surveys have also mapped the landscape of MPC in power electronics, categorising direct and indirect formulations, detailing the impact of weighting factors, horizon length and solver choice, and evaluating trade-offs between performance, stability and complexity. These reviews highlight emerging trends in hardware acceleration, application-specific solvers and integrated modelling tools that have accelerated MPC deployment in industrial power systems.

Model Predictive Control Strategies in Power Electronics publication trend

The graph below shows the total number of articles in model predictive control strategies in power electronics across all publications each year (not limited to Nature Index journals).

Technical terms

Model Predictive Control (MPC): A control strategy that optimises future control actions by solving an online constrained optimisation problem over a finite horizon.

Finite Control Set (FCS): A formulation of MPC for power converters in which discrete switching states are directly enumerated and evaluated within the optimisation.

Space Vector Modulation (SVM): A modulation technique that synthesises the desired voltage vector by combining inverter switching states to achieve a fixed switching frequency.

Total Harmonic Distortion (THD): A measure of waveform distortion, quantified as the ratio of the sum of harmonic components to the fundamental component.

References

1. Discrete space vector modulation and optimized switching sequence model predictive control for three-level voltage source inverters. Protection and Control of Modern Power Systems (2023).
2. Neural Network Model-Predictive Control for CHB Converters With FPGA Implementation. IEEE Transactions on Industrial Informatics (2023).
3. Model Predictive Control of Power Electronic Systems: Methods, Results, and Challenges. IEEE Open Journal of Industry Applications (2020).
4. Model-Free Predictive Current Control of a Voltage Source Inverter. IEEE Access (2020).