The global energy landscape is undergoing a profound transformation, driven by the imperative to combat climate change and achieve carbon neutrality. In this context, solar photovoltaic (PV) generation has emerged as a cornerstone technology due to its cleanliness, sustainability, and increasing cost-effectiveness. As nations, including China with its “Dual Carbon” goals, commit to ambitious energy transitions, the installed capacity of distributed solar PV systems continues to rise exponentially. This surge in penetration, however, introduces significant technical challenges for modern power grids, particularly concerning system stability, power quality, and operational resilience. Traditionally, grid-connected solar inverters operate in a current-controlled mode, relying entirely on the main grid to provide a stable voltage and frequency reference. During grid disturbances or outages, these inverters are typically mandated to disconnect via anti-islanding protection, ceasing power delivery to local loads and compromising supply reliability. This limitation underscores a critical need for advanced control paradigms that enable distributed energy resources to provide active grid support and enhance power supply resilience.
To address the challenge of uninterrupted power supply during grid failures, the integration of energy storage systems (ESS) with PV has become a common solution. In such hybrid setups, the storage inverter can provide the necessary voltage and frequency support in islanded mode, allowing the PV inverter to remain in a current-following mode. However, this approach is constrained by the finite energy capacity, high cost, and lifecycle limitations of batteries. In extreme scenarios where the storage is depleted, faulty, or simply unavailable, the system loses its voltage-forming source. Consequently, research is increasingly focusing on control strategies that empower the PV inverter itself to operate in a voltage-controlled, grid-forming mode independently, without the support of storage or the main grid. This capability defines a multi-modal solar inverter that must seamlessly transition between a grid-following (current source) state when connected to a healthy grid and a grid-forming (voltage source) state when islanded. The core challenge lies in executing smooth, transient-free transitions between these fundamentally different control modes while maintaining power balance and high power quality for the critical loads.
This article delves into the design and implementation of a multi-mode control strategy for a two-stage solar inverter enabling seamless transitions between grid-connected and storage-less islanded operation. We analyze the requisite operational modes, detail the control architectures for each, and propose a novel switching strategy that ensures continuity of supply. The performance and efficacy of the proposed method are validated through experimental results from a scaled-down laboratory prototype.
Operational Modes and System Topology
A versatile solar inverter designed for both grid-supportive and standalone operation must master several distinct operational modes, as illustrated in the following state diagram. The primary states are Grid-Connected Mode and Storage-less Islanded Mode, bridged by two transitional states: Grid-to-Island Transition and Island-to-Grid Transition.
Mode 1: Grid-Connected Operation. In this normal operating state, the inverter is synchronized to the main grid via a static switch (e.g., a relay or contactor). The grid provides a rigid voltage and frequency reference. The solar inverter operates as a controlled current source, injecting active and, if required, reactive power according to dispatch commands. Crucially, to provide grid support and mitigate the volatility of MPPT operation, the inverter typically operates in a constant power output (CPO) mode, tracking a power reference from a system operator rather than the absolute maximum power point.
Mode 2: Storage-less Islanded Operation. Upon detection of a grid fault (e.g., voltage/frequency deviation beyond thresholds), the inverter disconnects from the grid. In the absence of both the grid and a supporting storage unit, the PV inverter must immediately assume responsibility for forming the local AC voltage and frequency. It switches to a voltage-controlled mode, acting as a grid-forming source. The key challenge here is maintaining instantaneous power balance between the PV generation and the local load demand, which is achieved by using the DC-link voltage as an indicator of power imbalance to adjust the PV operating point.
Mode 3 & 4: Transition Modes. These are dynamic states managing the switch between Mode 1 and 2. The Grid-to-Island Transition must occur rapidly and smoothly upon fault detection to prevent load dropout. The Island-to-Grid Transition involves a pre-synchronization process where the inverter’s output voltage is adjusted in magnitude, frequency, and phase to match the restored grid voltage before re-closing the static switch, preventing large intrush currents.
The chosen hardware topology is critical for enabling these advanced control functions. A two-stage conversion architecture is selected for its flexibility in decoupling the PV panel operating point from the AC grid requirements. The system comprises a front-end DC-DC boost converter and a rear-end DC-AC inverter.
- Front-End Boost Converter: A three-level boost topology is employed. This structure offers advantages over the conventional two-level boost, including reduced voltage stress on the switching devices, lower inductor current ripple, and smaller filter requirements. It regulates the input from the PV array and maintains a stable DC-link voltage (
U_dc). The control of this stage is pivotal for power balance during islanded operation. - Rear-End Inverter: A three-level Neutral-Point-Clamped (NPC) inverter is used. NPC inverters are renowned for superior output voltage quality (lower total harmonic distortion – THD), reduced switching losses, and lower common-mode voltage compared to their two-level counterparts. This directly contributes to enhanced power quality for sensitive loads in islanded mode.
The overall system topology, including the LCL output filter and the grid connection switch, is shown in the following conceptual diagram. The PV array is represented by a DC source and series resistance emulating its I-V characteristic. The static switch K1 connects the inverter output to the Point of Common Coupling (PCC).
Control Strategy for Grid-Connected Mode
In grid-connected mode, the primary control objective for the solar inverter is to accurately inject commanded active (P_ref) and reactive (Q_ref) power into the grid while maintaining a stable DC-link voltage. The control structure is hierarchically organized across the two power stages.
1. Inverter Control (Current Source Mode):
The inverter uses a synchronous reference frame (dq-frame) control strategy, locked to the grid voltage via a Phase-Locked Loop (PLL). Assuming the PLL aligns the grid voltage vector with the d-axis (u_gq = 0), the instantaneous power theory gives:
$$P = \frac{3}{2} u_{gd} i_{d}$$
$$Q = -\frac{3}{2} u_{gd} i_{q}$$
where u_gd is the d-axis grid voltage, and i_d, i_q are the inverter output currents in the dq-frame. Therefore, the current references for the inner fast-current-control loops are derived directly from the power references:
$$i_{d\_ref} = \frac{2 P_{ref}}{3 u_{gd}}$$
$$i_{q\_ref} = -\frac{2 Q_{ref}}{3 u_{gd}}$$
A standard PI-based current controller tracks these references, generating the duty cycles for the NPC inverter modulation.
2. Boost Converter Control (DC-Link Voltage Regulation):
The stability of the DC-link voltage (U_dc) is paramount as it reflects the power balance between the PV input and the AC output. A constant DC voltage reference (U_dc_ref) is set. A voltage outer loop with a PI controller processes the error between U_dc_ref and the measured U_dc. The output of this voltage controller serves as the reference for an inner inductor current (i_L) control loop. By regulating the boost inductor current, the power drawn from the PV panels is controlled to match the inverter’s output power plus losses, thereby stabilizing U_dc. This control structure inherently causes the PV array to operate off its natural MPPT curve, settling at the operating point that delivers the power demanded by P_ref. Only when the available PV power is insufficient to meet P_ref does the system default to a current-limited MPPT mode. The key parameters for this control mode are summarized below.
| Control Loop | Reference | Measured Variable | Controller Output | Objective |
|---|---|---|---|---|
| Inverter Current (Inner) | i_dq_ref from Power Calc. |
i_dq |
Modulation Index | Fast current tracking |
| Inverter Power (Outer) | P_ref, Q_ref |
P, Q (calc.) |
i_dq_ref |
Accurate power dispatch |
| Boost Current (Inner) | i_L_ref from DC Voltage Ctrl. |
i_L |
Boost Duty Cycle | PV current regulation |
| DC-Link Voltage (Outer) | U_dc_ref |
U_dc |
i_L_ref |
Stable DC bus voltage |
Control Strategy for Storage-less Islanded Mode
When islanded, the solar inverter transitions from a current source to a voltage source. The control objectives shift to: 1) maintaining a stable, high-quality sinusoidal voltage at the load terminals with fixed amplitude (U_m_ref) and frequency (f_ref), and 2) ensuring real-time power balance between the variable PV generation and the variable load demand without an intermediate buffer like storage.
1. Inverter Control (Voltage Source Mode):
The inverter now employs a voltage-controlled strategy. An outer voltage loop generates the current reference for the inner current loop. The reference voltage signal (u_dq_ref) is directly generated from the amplitude and frequency setpoints. A standard approach uses PI regulators in the dq-frame:
$$i_{d\_ref} = (K_{p\_v} + \frac{K_{i\_v}}{s}) (u_{d\_ref} – u_{d\_meas})$$
$$i_{q\_ref} = (K_{p\_v} + \frac{K_{i\_v}}{s}) (u_{q\_ref} – u_{q\_meas})$$
where u_d_ref = U_m_ref, u_q_ref = 0, and u_dq_meas are the measured load voltages. The inner current loop remains identical to the grid-connected mode, ensuring fast dynamic response to load changes and disturbance rejection.
2. Power Balance Mechanism (Virtual Governor):
The absence of storage makes power balance the critical challenge. The mechanism hinges on the DC-link capacitor. The power balance equation at the DC link is:
$$P_{pv} – P_{loss} = P_{inv} + C U_{dc} \frac{dU_{dc}}{dt}$$
where P_pv is PV power, P_loss is system loss, P_inv is inverter output power (equal to load power P_load), C is DC-link capacitance, and U_dc is its voltage. If P_pv > P_load, the excess power charges the capacitor, causing dU_dc/dt > 0 and U_dc to rise. Conversely, if P_pv < P_load, the deficit is supplied by the capacitor, causing U_dc to fall.
This voltage deviation is used as a feedback signal to adjust the PV operating point. The DC-link voltage controller in the boost stage (similar to the grid-connected mode but with a different purpose) acts as a “virtual governor.” If U_dc rises above its reference (implying excess PV power), the controller reduces the power drawn from the PV array by lowering the PV operating voltage, moving it away from the MPP. If U_dc falls (implying deficit), the controller increases the PV power extraction. This creates a closed-loop system where the load demand ultimately dictates the PV generation level, ensuring stability. This strategy is viable as long as the available PV power exceeds the load demand. For scenarios where load exceeds available PV power, load-shedding protocols for non-critical loads must be incorporated.
Proposed Seamless Mode Transition Strategy
The abrupt change in control law from current-source to voltage-source (and vice versa) can cause severe transients, including voltage sags/swells and current spikes, potentially tripping protections or damaging equipment. The proposed strategy ensures bumpless transfer by managing the initialization of integrators and reference signals during the switch.
Grid-to-Island Transition
This transition is triggered by a grid fault detection algorithm (e.g., detecting abnormal voltage or frequency). The static switch is commanded to open. The key to smoothness is that the inner current control loop remains active and unchanged throughout. The challenge is seamlessly swapping the source of the current reference i_dq_ref.
- At the moment of switching: The control logic disconnects the power calculator (source of
i_dq_refin grid mode) and connects the output of the voltage controller (source ofi_dq_refin island mode). - Integrator Initialization (Bumpless Transfer): To prevent a step change in the voltage controller’s output, its integrators are pre-initialized with the value of the final current reference just before the switch. That is, if at the instant before transition
i_d_ref = I_d_gridandi_q_ref = I_q_grid, these values are fed into the corresponding integrators of the islanded mode’s d-axis and q-axis voltage PI controllers. This ensures the controller output starts from the correct operating point. - Phase Reference Handover: The phase angle
θfor the dq-transformations must also continue smoothly. The islanded mode’s voltage-forming oscillator (integratingω_ref) is initialized with the phase angle output by the PLL at the moment of grid disconnection. This maintains phase continuity for the load voltage.
With these measures, the inverter’s output current and voltage exhibit no discontinuity, and the load experiences an uninterrupted sine wave.
Island-to-Grid Transition (Pre-Synchronization)
Reconnection requires the inverter’s output voltage to be synchronized in magnitude, frequency, and phase with the restored grid voltage. A forced, fast synchronization is performed before closing the static switch.
1. Voltage Magnitude Synchronization: The reference voltage amplitude for the islanded controller (U_m_ref') is dynamically adjusted by adding a correction term derived from the error between the measured grid voltage (u_g_dq) and the inverter voltage (u_m_dq).
$$U_{m\_ref\_new} = U_{m\_ref} + (K_{p\_sync} + \frac{K_{i\_sync}}{s})(u_{g\_dq} – u_{m\_dq})$$
This closed-loop adjustment gradually drives the inverter output voltage to match the grid voltage.
2. Frequency and Phase Synchronization: Similarly, the frequency reference for the islanded mode’s oscillator is adjusted based on the phase error between the grid PLL angle (θ_PLL) and the inverter’s internal angle (θ_inv).
$$ω_{ref\_new} = ω_{ref} + (K_{p\_ph} + \frac{K_{i\_ph}}{s})(θ_{PLL} – θ_{inv})$$
This phase-locked loop structure forces θ_inv to track θ_PLL. Once the voltage magnitude error and phase error are within acceptable tolerances (e.g., < 3%, < 5°), the static switch is closed. Subsequently, the control mode of the solar inverter is switched from voltage control back to the grid-connected current control, again using integrator initialization to inherit the final operating state from the islanded mode, ensuring a smooth takeover of power dispatch control.
Experimental Validation
A laboratory-scale prototype was built to validate the proposed multi-mode control strategy for the solar inverter. The test bench comprises a programmable DC source with a series resistor to emulate the I-V characteristic of a PV array, the two-stage 3L-NPC solar inverter, an LCL filter, local resistive load banks, and a grid emulator. A DSP (TMS320F28377D) serves as the central controller. Key system parameters are listed in the table below.
| Parameter | Symbol | Value |
|---|---|---|
| Grid Voltage (RMS, Phase) | U_g |
70 V |
| Island Voltage Reference (RMS) | U_m_ref |
70 V |
| Grid-Connected Power Command | P_ref |
210 W |
| System Frequency | f |
50 Hz |
| Load Resistance | R_load |
20 Ω |
| DC Source Voltage (Emulated PV) | U_pv |
70 – 80 V |
| Switching Frequency | f_sw |
10 kHz |
Experiment 1: Grid-to-Island Transition. The inverter initially operated in grid-connected mode, injecting 210 W. A grid fault was simulated by opening the grid connection. The inverter detected the fault, opened its static switch, and executed the proposed transition strategy. The results demonstrated a seamless switch. The load voltage and current waveforms showed no interruption or significant distortion. The PV emulator voltage dropped from ~74 V to ~70 V as the boost converter adjusted the operating point to meet the new load power demand of approximately 360 W. The DC-link voltage experienced a minor, well-damped dip during the transient and quickly recovered to its reference of 200 V, confirming effective power balance control.
Experiment 2: Island-to-Grid Transition. The inverter initially powered the load in storage-less islanded mode. The grid emulator was then enabled. Upon command, the inverter entered the pre-synchronization state. Within approximately six grid cycles, its output voltage synchronized perfectly with the grid voltage in all aspects. The static switch was then closed, reconnecting the system. The transition was smooth, with no observable current surge. The inverter successfully resumed grid-following operation, delivering the commanded 210 W. The DC-link voltage remained stable throughout. The quality of the inverter’s output voltage in grid-connected mode was analyzed, showing a Total Harmonic Distortion (THD) of 2.88%, which is well within standard limits (e.g., IEEE 1547), proving the control strategy does not compromise power quality.
Conclusion
This work has presented a comprehensive multi-modal control framework enabling a grid-connected solar inverter to transition seamlessly into a storage-less, grid-forming power source during utility outages. The proposed strategy addresses a critical gap in enhancing the resilience of distributed PV systems by eliminating the absolute dependency on energy storage for islanded operation. The two-stage topology with three-level conversion provides the necessary hardware foundation for high-quality voltage synthesis and efficient power processing. The control architecture elegantly separates the concerns: stable DC-link regulation for power balancing and dual-mode inverter control for grid-following and grid-forming operations.
The core innovation lies in the detailed transition strategy employing integrator initialization and reference handover techniques for the Grid-to-Island switch, and a dual-loop pre-synchronization controller for the Island-to-Grid reconnection. Experimental results on a functional prototype confirm the efficacy of the approach, demonstrating smooth, transient-free transitions without load interruption, stable voltage regulation in islanded mode, and maintained power quality.
This capability significantly advances the concept of the “grid-supportive” solar inverter, transforming it from a passive power injector to an active participant in maintaining grid stability and supply continuity. It offers a cost-effective pathway to improve reliability in microgrids and areas with weak grids, as it leverages the inherent controllability of power electronic interfaces without mandating additional capital expenditure on storage. Future work will focus on expanding this strategy to multi-inverter parallel operation in an islanded network, optimizing power sharing, and integrating advanced grid-support functions like fault ride-through and dynamic voltage regulation.