Decentralized control methodology for multi-machine/multi-converter power systems

TL;DR

This work addresses synchronization of mixed machine/converter power systems with reduced inertia by employing a model-matching framework that links DC-side inverter modulation to SG-like dynamics via a secondary control layer. It establishes a bijective translation between network-level OPF setpoints $P^*,Q^*,\|v^*\|$ and local references $\{I_r^*,\theta_{dq}^*\}$ (reduced model) or $\{v_{dq}^*,\xi^*\}$ (full model), enabling topology-robust decentralized control through energy-function gradients. The analysis covers topology effects, particularly cycled graphs, and introduces a communication-graph approach to suppress circulating power and achieve angle consensus, including shifted-consensus variants. Decentralized control strategies are developed for both reduced and full models, including feedback linearization and incremental energy-function methods, with a voltage-based fully decentralized full-model controller demonstrated in simulations and discussed for robustness to clock drift and parameter variations. Overall, the framework provides practical pathways for imposing OPF-driven targets in grid-forming, mixed inverter–SG networks, offering insights into decentralization, topology handling, and robustness concerns for future power-system control design.

Abstract

In this project we evaluate a framework for synchronization of mixed machine-converter power grids. Synchronous machines are assumed to be actuated by mechanical torque injections, while the converters by DC-side current injections. As this approach is based on model-matching, the converter's modulation angle is driven by the DC-side voltage measurement, while its modulation amplitude is assigned analogously to the electrical machine's excitation current. In this way we provide extensions to the swing-equations model, retaining physical interpretation, and design controllers that achieve various objectives: frequency synchronization while stabilizing an angle configuration and a bus voltage magnitude prescribed by an optimal power flow (OPF) set-point. We further discuss decentralization issues related to clock drifts, loopy graphs, model reduction, energy function selection and characterizations of operating points. Finally, a numerical evaluation is based on experiments from three- and two-bus systems.

Figures (18)

• Figure 1: Circuit diagram of a 3-phase inverter Jouini
• Figure 2: Matrix representation of reduced network model
• Figure 3: Matrix representation of full network model
• Figure 4: Schematics of decentralized control setup GlobPVSyn
• Figure 5: Reduced power system model with cycled graph of generators