Non-linear Control of the Power Injected Into a Weak Grid by a Self-Synchronized Inverter

TL;DR

This work tackles robust grid-tied inverter control in very weak grids by developing a nonlinear, self-synchronizing strategy that does not require a PCC voltage sensor. A full-order observer estimates the PCC voltage $\vec{v}_p$ from the measured inductor current, enabling a nonlinear feedback-linearization controller coupled with current limiting and anti-windup, plus a startup synchronization mechanism. The control architecture includes a low-level energy-based FL/FSF scheme, a CLA with AW, and a slow outer-loop droop with input-power limitation, validated through simulations under start-up, normal operation, and grid disturbances with unknown grid parameters. The results show stable operation, effective disturbance rejection, and safe current and voltage trajectories, highlighting practical viability for grid-tied inverters in weak-grid scenarios while reducing reliance on direct PCC sensing.

Abstract

In this work, a non-linear controller designed using non-linear transformation linearization and feedback is proposed for an inverter connected to a weak grid through a single-stage inductive filter. The proposed strategy is self-synchronized, so that it is not necessary to have a voltage sensor at the Point of Common Coupling (PCC). The strategy allows to robustify, in the presence of a weak grid, a strategy that has already been demonstrated to allow a significant reduction in the size of the DC-link capacitor of the converter. For this purpose, a state observer is designed that allows estimating the voltage at the PCC from the measurement of the output inductor current. A start-up controller is also included, which allows synchronization even in the case of system start-up. Simulation results are presented for different operating cases, including start-up, normal operation, and grid-voltage sags and swells. In all these cases, it is considered that the exact parameters of the grid to which the inverter is connected are unknown.

Figures (5)

• Figure 1: Diagram of the system.
• Figure 2: Block diagram of the proposed controller.
• Figure 3: Start-up sequence. (a) $\vec{v}_p$ vs $\hat{\vec{v}}_p$. (b) $\vec{i}$. (c) $v_c^*$ vs $v_c$.
• Figure 4: Normal operation. (a) $\vec{v}_p$ vs $\hat{\vec{v}}_p$. (b) $\vec{i}$. (c) Anti-windups for FL control and CLA. (d) $p$, $q$ and $p_{i\text{max}}$. (e) $v_c^*$ vs $v_c$.
• Figure 5: Sag and swell. (a) $\vec{v}_p$ vs $\hat{\vec{v}}_p$. (b) $\vec{i}$. (c) Anti-windups for FL control and CLA. (d) $p$, $q$ and $p_{i\text{max}}$. (e) $v_c^*$ vs $v_c$.