Generalized Voltage-based State-Space Modelling of Modular Multilevel Converters with Constant Equilibrium in Steady-State
This work provides a linearizable state-space model for MMCs, enabling eigenvalue-based analysis and multi-variable control design, which is important for power electronics engineers working on MMC stability and control.
This paper presents a generalized state-space model for modular multilevel converters (MMCs) that achieves constant equilibrium in steady-state by using sum and difference of arm voltages and applying Park transformations at three frequencies. The model accurately captures internal dynamics and is validated through time-domain simulations against an average model and a detailed 400-submodule MMC model.
This paper demonstrates that the sum and difference of the upper and lower arm voltages are suitable variables for deriving a generalized state-space model of an MMC which settles at a constant equilibrium in steady-state operation, while including the internal voltage and current dynamics. The presented modelling approach allows for separating the multiple frequency components appearing within the MMC as a first step of the model derivation, to avoid variables containing multiple frequency components in steady-state. On this basis, it is shown that Park transformations at three different frequencies ($+ω$, $-2ω$ and $+3ω$) can be applied for deriving a model formulation where all state-variables will settle at constant values in steady-state, corresponding to an equilibrium point of the model. The resulting model is accurately capturing the internal current and voltage dynamics of a three-phase MMC, independently from how the control system is implemented. The main advantage of this model formulation is that it can be linearised, allowing for eigenvalue-based analysis of the MMC dynamics. Furthermore, the model can be utilized for control system design by multi-variable methods requiring any stable equilibrium to be defined by a fixed operating point. Time-domain simulations in comparison to an established average model of the MMC, as well as results from a detailed simulation model of an MMC with 400 sub-modules per arm, are presented as verification of the validity and accuracy of the developed model.