Analysis of a Novel Three-Port Single-Stage Bidirectional DC–AC Converter for PV-ESS-V2G System

Abstract

Multiport DC–AC converters are widely used in photovoltaic-energy storage–charging systems, but traditional two-stage schemes face challenges in circuit cost and efficiency improvements. To address this issue, a novel three-port single-stage DC–AC converter is proposed for grid-connected applications. The proposed converter integrates two DC ports and one AC port through circuit multiplexing, eliminating the high-voltage DC bus and reducing system complexity. An unfolding bridge is employed at the AC port, and full bridge circuits are used at DC ports, reducing the number of high-frequency switches. The proposed single-stage topology inherently achieves galvanic isolation and bidirectional power conversion. To achieve accurate grid current regulation and wide-range zero-voltage-switching, a multiple-phase-shift modulation method is developed to ensure a sinusoidal current waveform. The effectiveness of the proposed converter and modulation method is verified through simulation results, demonstrating a peak efficiency of 97% and a total harmonic distortion of 2.91%.

1. Introduction

With the rapid growth of photovoltaic-energy storage–charging systems [1,2], such as the increasing demand for vehicle-to-grid (V2G), bidirectional power transmission technologies have attracted much interest. However, conventional schemes based on discrete converters suffer from limited efficiency, low power density, and insufficient interactions of multiple energy sources. Consequently, highly integrated multiport DC–AC converters have emerged as a promising solution [3,4]. By integrating EVs, battery energy storage, photovoltaic, and the utility grid, multiport converters enable efficient energy routing while meeting the requirements of high reliability, high power quality, and multifunctional integration in V2G applications.

Multiport high-frequency (HF) isolated DC–AC converters serve as power interfaces enabling multiport interconnection and galvanic isolation among EVs, energy storage systems, and the AC utility grid [5]. Considering the characteristics of distributed energy systems, multiport DC–AC converters are required to provide accurate bidirectional power flow control [6,7], high integration with low cost [8], high conversion efficiency [9], and galvanic isolation for safe and reliable operation [3,10,11].

Multiport DC–AC converters can be classified into four categories according to the number of power conversion stages, including two-stage DC–AC converters with independent configuration, two-stage DC bus-coupled DC–AC converters, two-stage transformer-coupled DC–AC converters, and single-stage DC–AC converters with independent configuration.

The simplified circuit of the first type is shown in Figure 1a. Each DC source employs an individual two-stage isolated DC–AC converter to achieve power conversion and interaction with the AC grid. The two-stage topology consists of a high-frequency isolated DC–DC converter, a DC bus, and a non-isolated DC–AC converter. The DC–DC converter provides buck-boost voltage regulation and galvanic isolation between the DC port and the DC bus, and the DC–AC converter regulates the grid current [12,13,14,15]. Various modulation techniques have been proposed to address common-mode voltage problems in non-isolated multilevel inverters [16]. For instance, an elimination method for three-level neutral-point-clamped (NPC) inverters was presented in [17]. However, this method focuses on non-isolated configurations and does not address the efficiency and component reduction challenges in isolated multiport systems. Since each DC source operates independently, existing isolated DC–DC and DC–AC schemes can be directly adopted. However, this structure relies on a high-voltage DC bus, which typically requires large-capacitance electrolytic capacitors for power decoupling. Although electrolytic capacitors are widely used due to their low cost, their lifespan is sensitive to environmental conditions, potentially reducing system reliability. Replacing electrolytic capacitors with film capacitors can improve power density and lifetime. Nevertheless, their high cost limits large-scale deployment [18].

Based on the first type shown in Figure 1a, the circuit structure can be simplified by sharing some specific circuit components. One approach employs individual high-frequency isolated DC–DC converters on the DC side, while sharing a single DC–AC converter for grid connection [19,20,21], as depicted in Figure 1b. Compared with the independent configuration, two DC ports are coupled through a DC bus, thereby eliminating one DC–AC conversion stage.

Compared with conventional schemes, a three-port HF isolated DC–DC converter employing a three-winding transformer enables higher integration by coupling two DC ports [22,23], as shown in Figure 1c. Overall system integration was further improved by eliminating four power switches and one magnetic core [24,25,26]. Coupling multiple DC ports via a multi-winding transformer can reduce the number of magnetic components, but this approach significantly increases transformer design complexity [19]. Moreover, the tight electromagnetic coupling between DC ports complicates the control algorithm and hinders achieving independent regulation for each port.

Recent studies have reported single-stage isolated DC–AC converters integrating isolated DC–DC and matrix-type converters [27,28], as exemplified in Figure 1d. Compared with conventional two-stage DC–AC converters, single-stage converters based on high-frequency isolated flyback, LLC resonant, and matrix-type structures can effectively eliminate the intermediate DC bus, thereby removing the bulky electrolytic capacitors and improving system reliability. However, existing studies on single-stage isolated converters primarily focus on single-port applications. When such converters are directly paralleled to accommodate multi-DC-source grid-connected applications, the total number of high-frequency switches increases substantially. Since the two single-stage DC–AC converters operate independently, further simplification through circuit multiplexing is possible.

To enhance the competitiveness of multi-port converters, this paper proposes a novel three-port single-stage bidirectional DC–AC converter for PV-ESS-V2G systems. The circuit topology is shown in Figure 2. The proposed converter topology is first analyzed in detail, followed by a comprehensive explanation of its operating principles under different power flow modes. An extended phase-shift modulation strategy is then developed to achieve independent power control for each DC port while ensuring zero-voltage-switching operation across a wide load range. Simulation results validate the theoretical analysis, demonstrating a peak efficiency of 97% and a total harmonic distortion of less than 3% for the grid current.

The main contributions of this paper consist of four points. First, a novel three-port single-stage bidirectional DC–AC topology is proposed, which enables single-stage power conversion. Second, all high-frequency switches achieve zero-voltage-switching across a wide load range, resulting in a peak efficiency of 97%. Third, the proposed topology reduces the number of high-frequency switches and magnetic components by 50% compared to conventional two-stage solutions, thereby significantly reducing circuit cost and volume. Fourth, an extended phase-shift control strategy is introduced to achieve independent control for each DC port.

This paper is organized as follows. The operation principle of the proposed converter is shown in Section 2. Simulation results are shown in Section 3. Section 4 discusses the losses, total harmonic distortion, comparison studies, and the application prospects in three-phase applications. Section 5 concludes the paper.

2. Operation Principle

2.1. Overview of the Proposed Topology

Figure 2 shows the circuit diagram of the proposed three-port single-stage bidirectional DC–AC converter. It features two DC ports and one grid connection interface, enabling the integration of various energy sources such as photovoltaics, energy storage, and electric vehicles, and facilitating single-stage power conversion and multi-mode interaction with the grid. The DC-side bridges of Port 1 and Port 2 are both full bridge circuits with four HF switches S₁–S₄ and S₅–S₈. The midpoint of the full bridge, denoted as v_p1 or v_p2, is connected to a transformer to form an independent port. The turn ratios of transformers T₁ and T₂ are denoted as n₁ and n₂, and leakage inductors are lumped in the secondary as L_k1 and L_k2, respectively. Subsequently, the outputs of Port 1 and Port 2 are connected in parallel to the grid-side bridge, which consists of switches S₉–S₁₂. The secondary midpoint voltage vs. is clamped by the utility grid. L_f and C_f form an L-C filter to bypass the high-frequency component of i_s and generate the grid current i_g. An unfolding-bridge circuit is used to reverse the polarity of grid voltage vg and current i_g, which consists of Q₁~Q₄.

During the positive half cycle of the grid voltage, switches Q₁ and Q₄ are turned on. During the negative cycle of the grid voltage, the switches Q₂ and Q₃ are turned on. To simplify the analysis, we assume that each component is ideal in steady state and the converter is lossless, meaning V_in × I_in = v_g × i_g. The switching frequency f_s is much higher than the line frequency f_g. Assume that the grid voltage v_g is a pure sinusoidal waveform as

$$v_g=V_g\sin ({\theta }_g)$$

(1)

where V_g and θ_g represent the peak value and phase angle of grid voltage, respectively.

Figure 3 shows the equivalent circuit of the proposed three-port converter. It is noted that the waveforms of v_p1, v_p2, and vs. can be controlled independently. Therefore, the currents i_s1, i_s2, and i_s can be controlled independently by adjusting the internal phase-angle α₁, α₂ of v_p1 and v_p2 and the external phase-angle β₁, β₂ between v_p1, v_p2, and v_s.

The switches S₉–S₁₂ operate at a high frequency to generate a high-frequency voltage vs. at the secondary midpoint. This full bridge generates high-frequency AC current, which, after being filtered by the filter network L_f-C_f, results in a pulsating DC current with double line frequency. The unfolding bridge (Q₁–Q₄) operates at line frequency and converts the pulsating DC current into a sinusoidal AC current.

The proposed converter supports bidirectional power transmission for each DC port independently. When power is delivered from the DC Port 1 to the AC grid, the primary-side full bridge generates a high-frequency voltage with a leading phase β₁ relative to v_s. As a result, energy flows from the DC source to the grid. Conversely, when power is absorbed from the grid into a DC port, the primary-side voltage lags behind v_s, indicating reverse power flow. This phase-shift control strategy allows each DC port to operate in either discharging mode (DC to AC) or charging mode (AC to DC) independently, enabling flexible power management in multi-source applications, such as simultaneous EV charging and discharging, or mixed renewable generation and storage coordination.

2.2. Modulation for Power Transmission from DC Port to AC Port

A modulation strategy based on extended phase shift (EPS) is introduced for precise control. Because of the independence of each DC port, Port 1 is selected as an example.

Figure 4 shows the theoretical waveforms of primary midpoint voltage n₁ × v_p1, secondary midpoint voltage v_s, and secondary current is1. The three-level voltage vp1 switches between V_in1, zero, and −V_in1 with a pulse width of α₁T_s. The two-level voltage vs. pulsates from |v_g| to −|v_g| with a pulse width of 0.5T_s. The phase-shift angle β₁ is defined as the leading time ratio between the center of v_p1 and the center of v_s. In this mode, the span of the primary voltage v_p1 is within the secondary voltage v_s. Therefore, the phase-shift angle β₁ is given by

$${\beta }_1>0.25-0.5{\alpha }_1$$

(2)

The slopes of the transformer current i_s1 in each interval can be derived as

$${\displaystyle \frac{\mathrm{d}{i}_{s1}}{\mathrm{d}t}}=\left\{\begin{array}{l}-v_g/L_{k1}, (t_0,t_1),(t_2,t_3)\\ \left(n_1{V}_{in1}-v_g\right)/L_{k1}, (t_1,t_2) \end{array}\right.$$

(3)

In addition, the current waveforms during a half switching period are symmetric along the t-axis, implying that i_s1(t₀) equals −i_s1(t₃). From this relationship and (3), i_s1(t₀) is obtained as

$$i_{s1}(t_0)={\displaystyle \frac{T_s}{2L_{k1}}}\left[0.5v_g-{\alpha }_1{n}_1{V}_{in1}\right]$$

(4)

Using (3) and (4), the average output current i_g1 generated by Port 1 is derived as

$$i_{g1}={\displaystyle \frac{1}{T_s}}{\displaystyle {\int }_{t_0}^{t_3}{i}_{s1}(\tau )d\tau }={\displaystyle \frac{n_1{V}_{in1}}{L_{k1}{f}_s}}\cdot {\alpha }_1\cdot {\beta }_1$$

(5)

Equation (5) shows that the relation between control variables α₁, β₁, and the average output current i_g1 is linear, so the output power can be precisely controlled by adjusting their amplitudes. To simplify the controller structure and obtain a sinusoidal grid-connected current, the phase-shift angle α₁ can be given as

$${\alpha }_1({\theta }_g)={\alpha }_{1m}\sin \left({\theta }_g\right).$$

(6)

where α_1m is the peak value of α₁. By substituting (6) into (5), the average output current i_g1 can be rewritten as

$$i_{g1}={\displaystyle \frac{n_1{V}_{in1}}{L_{k1}{f}_s}}{\beta }_1{\alpha }_{1m}\sin \left({\theta }_g\right).$$

(7)

From (7), it can be seen that a sinusoidal grid-connected current can be obtained by modulating the phase shift angle α₁ according to the sine law. In addition, the amplitude β₁ can be calculated based on the power reference value, expressed as

$${\beta }_1={\displaystyle \frac{\sqrt{2}{P}_{ref}}{\frac{n_1{V}_{in1}}{L_{k1}{f}_s}{\alpha }_{1m}{V}_g}}.$$

(8)

The control block diagram of the proposed solution is shown in Figure 5. The external power reference command comes from the battery management system or the charging control module. Since the focus of this paper is on the bidirectional power transmission function, the specific charging strategy is not presented. The power reference value is calculated using formula (8) to obtain the corresponding phase shift angle amplitude, and then the drive signal is obtained by modulating the phase shift angle according to (6). To achieve precise power control, a current loop was added to track the grid-connected current.

The switching-frequency f_s in (5) is designed as a constant (CSF) of 50 kHz. Using the control law (6), the power flows of DC Port 1 and DC Port 2 can be precisely controlled. Figure 6 shows the circuit transients from t₀ to t₃, consisting of ten operating stages. At Stage 1, Stage 4, and Stage 7, all of the switches S₁–S₄ and S₉–S₁₂ have achieved ZVS, which ensures the high efficiency of the proposed topology.

Stage 1: t = t₀. S3, S4, S9–S12 are turned off; S1 and S2 are turned on. The leakage current flows through the body diodes of S9 and S12, thereby fully discharging the parasitic capacitance. This leads to ZVS turn-on of S9 and S12.

Stage 2: t₀ < t < t₁ & i_s1 > 0. S3, S4, S10–S11 are turned off; S1, S2, S9, and S12 are turned on.

Stage 3: t₀ < t < t₁ & i_s1 < 0. Being similar to Stage 2, S3, S4, S10–S11 are turned off; S1, S2, S9, and S12 are turned on.

Stage 4: t = t₁. S2, S3, S4, S10–S11 are turned off; S1, S9, and S12 are turned on. The leakage current flows through the body diodes of S4, thereby fully discharging the parasitic capacitance. This leads to ZVS turn-on of S4.

Stage 5: t₁ < t < t₂ & i_s1 < 0. S2, S3, S10–S11 are turned off; S1, S4, S9, and S12 are turned on. The primary current flows in the opposite direction.

Stage 6: t₁ < t < t₂ & i_s1 > 0. S2, S3, S10–S11 are turned off; S1, S4, S9, and S12 are turned on. The primary current is flowing in the forward direction.

Stage 7: t = t₂. S1, S2, S3, S10–S11 are turned off; S4, S9, and S12 are turned on. The leakage current flows through the body diodes of S3, thereby fully discharging the parasitic capacitance. This leads to ZVS turn-on of S3.

Stage 8: t₂ < t < t₃, i_s1 > 0. S1, S2, S10–S11 are turned off; S3, S4, S9, and S12 are turned on. The secondary current is flowing in the forward direction.

Stage 9: t₂ < t < t₃, i_s1 < 0. S1, S2, S10–S11 are turned off; S3, S4, S9, and S12 are turned on. The secondary current is flowing in the opposite direction.

Stage 10: t = t₃. S1, S2, S9–S12 are turned off; S3 and S4 are turned on. The leakage current flows through the body diodes of S10 and S11, thereby fully discharging the parasitic capacitance. This leads to ZVS turn-on of S10 and S11.

2.3. Modulation for Power Transmission from AC Port to DC Port

The operation principle of AC to DC power delivery is similar to that of the DC–AC mode, but the geometric center of the voltage v_p1 lags behind v_s, as shown in Figure 7. The negative sign of β₁ indicates the lagging angle. Therefore, the average input current i_g1 for Port 1 is derived as

$$i_{g1}={\displaystyle \frac{1}{T_s}}{\displaystyle {\int }_{t_0}^{t_3}{i}_{s1}(\tau )d\tau }=-{\displaystyle \frac{n_1{V}_{in1}}{L_{k1}{f}_s}}\cdot {\alpha }_1\cdot {\beta }_1$$

(9)

The negative sign of i_g1 in (9) indicates that power flows from the AC grid to DC Port 1. Therefore, bidirectional power control is achievable, and the proposed control strategy enables independent bidirectional power regulation for all ports.

2.4. Analysis of ZVS Constraint

The ZVS condition for the primary switches is determined by the polarity and magnitude of the transformer current at the switching instant. Taking the switches S1–S4 of Port 1 as an example, to achieve zero-voltage turn-on for S1 and S4, the current i_s1(t₀) must be negative to discharge the output capacitances of the switches before turn-on. For S2 and S3, ZVS requires that i_s1(t₃) be positive.

$$i_{s1}(t_0)={\displaystyle \frac{T_s}{2L_{k1}}}\left[0.5v_g-{\alpha }_1{n}_1{V}_{in1}\right]<0$$

(10)

From Equation (10), it can be observed that as long as the external phase-shift angle β₁ is properly selected within its feasible range, i_s1(t₀) maintains a sufficient negative value to ensure ZVS operation for S1 and S4, while the positive amplitude of i_s1(t₃) is also adequate to complete the discharge of the output capacitances of S2 and S3.

For the primary switches S5–S8 of Port 2, the ZVS condition is completely symmetric to that of Port 1. Specifically, to achieve ZVS for S5 and S8, the current i_s2(t₀) must be negative. For S6 and S7, ZVS requires that i_s2(t₃) be positive. Due to the circuit symmetry and the consistency of the modulation strategy, when the external phase-shift angle β₂ is selected within a reasonable range, i_s2(t₀) maintains a sufficient negative value to ensure zero-voltage turn-on for S5 and S8, while the positive amplitude of i_s2(t₃) is also adequate to complete the discharge of the output capacitances of S6 and S7.

In practical implementation, these non-idealities, including parasitic components and timing constraints, have an impact on system efficiency. Among them, two critical factors significantly affect switching losses: the switch’s parasitic output capacitance and the effect of the dead-time interval of high-frequency switches.

First, while ZVS aims to discharge the junction capacitance before turn-on, the non-linear and voltage-dependent characteristic of the parasitic capacitance complicates this process. If the energy stored in the parasitic capacitance is not completely discharged by the leakage current during the switching transition, it leads to additional dissipation. Therefore, the improved ZVS constraints can be expressed as

$$\left|i_{s1}(t_0)\right|\ge u_{dc}\times \sqrt{C_{oss}/L_{k1}}$$

(11)

Second, the dead-time must be long enough to allow the leakage current to fully charge or discharge the parasitic capacitances. If the dead-time is too short, the transition is incomplete, resulting in hard-switching. Conversely, if the dead-time is excessive, it can lead to body diode conduction, which causes reverse recovery losses and increased conduction losses due to the diode’s relatively high forward voltage drop. Therefore, optimizing the dead-time is crucial for maintaining high efficiency in practical applications.

2.5. Design of Leakage Inductance

The proposed topology relies on a single inductor, i.e., the leakage inductance, to perform high-frequency power conversion. Therefore, the core of the circuit parameter design lies in the leakage inductance design. According to (7), when the input voltage and the switching frequency are determined, the leakage inductance determines the maximum output power of the circuit. Given that the voltage variation range of 400 V-level electric vehicles is 300–450 V, and the corresponding SOC is 20–100%, the inductance value should be designed based on the lowest DC voltage, as this represents the worst working condition and the maximum current stress.

$$L_{k1}\le {\displaystyle \frac{n_1{V}_{in1}{\beta }_{1,\max }{V}_g{\alpha }_{1m}}{\sqrt{2}{P}_{o,\max }{f}_s}}$$

(12)

where P_o,max is the maximum output power. The amplitude of the inner phase shift angle is α_1m = 0.35. Substituting parameter values, including n₁ = 1, V_in1 = 400 V, β_1,max = 0.05, V_g = 220, P_o,max = 6600, and f_s = 50 kHz, into (12), the leakage inductance is designed as 3.29 μH.

3. Simulation Results

This paper develops a simulation model of a three-port single-stage DC–AC converter using the PLECS version 4.8.9 platform. The simulation model is developed and analyzed using the PLECS platform, which is well-suited for power electronic simulations due to its fast simulation speed and robust handling of switched systems.

To improve the accuracy of the model and better reflect real-world performance, practical parasitic parameters are included in the device models. The MOSFETs are modeled with a specified on-resistance to represent conduction losses, and parasitic inductances are added at the leads to account for switching transients and voltage ringing. Similarly, the diodes are modeled with their on-resistance and forward voltage drop.

This simulation platform supports the import of the actual parameters of power devices into the model. For instance, we imported the parameters of Infineon’s 650 V MOSFET, which are usually obtained from the data sheets provided by the manufacturer. This enables the simulation platform to conduct loss simulation and thermal simulation. These functions ensure that the simulation results more closely match the behavior of an actual hardware prototype.

The circuit parameters are listed in

Table 1. Circuit parameters of the proposed topology.

Parameters	Symbols	Value
DC port voltage	Vin1, Vin2	400 V
Grid voltage	vg	220 Vrms
Grid frequency	fg	50 Hz
Switching frequency	fs	50 kHz
Rated power	P1, P2	6.6 kW
Leakage inductance	Lk1, Lk2	3.29 μH
Transformer turns ratio	n1, n2	1:1
Peak phase-shift angle	α1m, α2m	0.35
Filter inductor	Lf	1 mH
Filter capacitor	Cf	30 μF

. To evaluate its performance in EV grid-connection applications, the two DC ports are assumed to be connected to two EVs. Under this assumption, four representative operating conditions are defined: (1) both DC Ports 1 and 2 deliver power to the utility grid, (2) both DC ports operate in charging mode and draw power from the grid, (3) DC Port 1 delivers power to the grid while DC Port 2 operates in charging mode, (4) both DC ports operate at rated power in the discharging mode and deliver power to the grid.

Correspondingly, four simulation cases are established and verified. Case 1 represents the simultaneous discharging of two EVs into the grid, with the output power set to 6.6 kW at DC Port 1 and 3.3 kW at DC Port 2, as shown in Figure 8. Case 2 corresponds to the simultaneous charging of both EVs from the grid, where the charging power is 6.6 kW at DC Port 1 and 3.3 kW at DC Port 2, as shown in Figure 9. Case 3 illustrates a mixed operating condition in which one EV discharges to the grid while the other is in the charging mode. In this case, DC Port 1 delivers 6.6 kW to the grid, whereas DC Port 2 draws 3.3 kW, as shown in Figure 10. Case 4 corresponds to the simultaneous charging of both EVs from the grid, where the charging power is 6.6 kW at both DC ports, as shown in Figure 11.

The summary of the total harmonic distortion (THD) of the grid-connected current for the proposed topology is shown in

Table 2. Total harmonic distortion of the grid-connected current.

Output Power (W)	THD (%)
6600	2.91
4950	4.61
3300	3.17
1650	9.73

. The THD of the grid-connected current at the rated output power is 2.91%, which is less than 3%. The simulation results show that the proposed three-port single-stage DC–AC converter can achieve precise power control and bidirectional power transmission.

Figure 12 presents the dynamic response curve of the grid-connected current. Figure 12a shows the response curve of the output power when it steps from 0 to 6.6 kW, and Figure 12b shows the response curve when the output power decreases from 6.6 to 0 kW. The proposed topology can enter a new steady state within half of a power frequency cycle, demonstrating excellent dynamic response speed.

4. Discussion

4.1. Analysis of Power Loss

The loss of the proposed converter includes conduction loss, switching loss, and core loss. The conduction loss can be divided into the conduction loss of the unfolding bridge P_{c_ufd}, and the conduction loss of the high-frequency switches P_{c_dab}. P_{c_ufd} can be expressed as

$$P_{c\_ufd}=I_g^2\cdot 2R_{ds\_ufd}$$

(13)

where I_g is the rms grid current, and R_{ds_ufd} is the on-resistance of the unfolding bridge switches.

$$P_{\mathrm{c}\_\mathrm{dab}}=(1+n_1^2)I_{s1}^2\cdot 2R_{on}+(1+n_2^2)I_{s2}^2\cdot 2R_{on}$$

(14)

where I_s1 and I_s2 are the rms values of currents i_s1 and i_s2, n₁ and n₂ are the transformer turns ratio, and Ron is the on-resistance of the high-frequency switches.

The switching loss consists of turn-off loss P_{s_off} and turn-on loss P_{s_on}. The turn-on loss expression of one switch is expressed as

$${\left.P_{s\_on}\right|}_{S_x}={\displaystyle \frac{C_{oss}{U}_{ds}^2}{2T_s}}$$

(15)

where U_ds is the drain–source voltage. The turn-off loss expression of one switch is evaluated as

$${\left.P_{s\_off}\right|}_{S_x}={\displaystyle \frac{U_{ds}{I}_d}{2T_s}}\cdot {\displaystyle \frac{Q_{gd}+Q_{gs}-Q_{th}}{U_{\mathrm{plat}}/R_{G\mathrm{o}\mathrm{f}\mathrm{f}}}}$$

(16)

where U_plat is the voltage at the Miller platform. R_G is the gate resistance. Q_gd, Q_gs, and Qth are the charges of the different stages, which can be obtained from the datasheet of the switch.

The core loss of the transformer is calculated by the improved generalized Steinmetz equation.

$$\begin{array}{l}{P}_{\mathrm{Core}}=V_e{\displaystyle \frac{1}{T_s}}{{\displaystyle \int }}_0^{T_s}{k}_i|{\displaystyle \frac{dB}{dt}}{|}^{\alpha }{(\Delta B)}^{\beta -\alpha }dt\\ k_i={\displaystyle \frac{k}{{\left(2\pi \right)}^{\alpha -1}{{\displaystyle \int }}_0^{2\pi }{\left|\cos \theta \right|}^{\alpha }{2}^{\beta -\alpha }d\theta }}\end{array}$$

(17)

dB/dt is the gradient of the magnetic flux density. ΔB is the peak-to-peak flux density. V_e is the volume of the core. k, α, and β are the parameters of the Steinmetz equation. The copper loss of the transformer can be expressed as

$$P_{\mathrm{cu}}=I_{s1}^2\cdot R_{cu1,\mathrm{s}}+n_1^2{I}_{s1}^2\cdot R_{cu1,\mathrm{p}}+I_{s2}^2\cdot R_{cu2,\mathrm{s}}+n_2^2{I}_{s2}^2\cdot R_{cu2,\mathrm{p}}$$

(18)

The average value of the total power losses, P_loss, can be obtained by integrating these loss components over half a line cycle.

$$P_{loss}={\displaystyle \frac{1}{\pi }}{{\displaystyle \int }}_0^{\pi }(P_{c\_ufd}+P_{c\_\mathrm{dab}}+4P_{s\_on}+4P_{s\_off}+P_{core}+P_{\mathrm{cu}})d\theta$$

(19)

Based on (19), the calculated conversion efficiencies under different output powers are summarized in

Table 3. The calculated conversion efficiencies under different output powers.

Output Power (W)	Efficiency (%)
6600	96.61
4950	97.09
3300	96.88
1650	95.66
825	91.07

. The detailed distribution of losses is shown in

Table 4. Distribution of power losses.

	6600 W		4950 W
	Loss	Percentage	Loss	Percentage
Ps_on + Ps_off	46.75 W	20.9%	36.73 W	25.5%
Pc_ufd + Pc_dab	90.62 W	40.5%	45.81 W	31.8%
Pcu	46.32 W	20.7%	28.95 W	20.1%
Pcore	40.05 W	17.9%	32.56 W	22.6%
Ploss	223.74 W	100%	144.05 W	100%

. Under heavy-load conditions, the main loss comes from conduction loss.

4.2. Grid Current’s THD

Based on the PLECS simulation platform, we can obtain the total harmonic distortion of the grid-connected current at the rated power as well as the amplitudes of each harmonic component. These data are obtained through the FET tool and shown in Figure 13. By comparing with the IEEE std 1547, it can be seen that the harmonics injected by the proposed converter meet the standard requirements and have a large margin.

4.3. Comparison Studies

This section compares the proposed scheme with those of the four existing schemes in terms of the high-frequency switches, inductors, capacitors, high voltage DC-link, and control strategies, as shown in

Table 5. Comparison of existing schemes in multi-source applications.

Reference	[29]	[27]	[20]	[22]	This Work
Topology	Individual Two-stage DC–AC converters	Individual single-stage DC–AC converters	Two DC–DC converters with a DC–AC inverter	Three-port DC–DC converter with a DC–AC inverter	Three-port single-stage DC–AC converter
High-Frequency Switches	24	16	20	16	12
Transformer	2	2	2	1 (Three-winding)	2
Inductors	Grid-side inductors: 2; Transformer-side inductors: 2	Grid-side inductors: 4; Transformer-side inductors: 2	DC-side inductors: 2; Transformer-side inductors: 2	0 (Leakage Inductance)	0 (Leakage Inductance)
Capacitors	DC Bus Capacitors: 4700 μF/450 V × 4	Capacitors: 470 μF/450 V × 2	DC Bus Capacitors: 4700 μF/450 V × 4	DC Bus Capacitors: 4700 μF/450 V × 4;	Capacitors: 470 μF/450 V × 2
High-voltage DC-link	Yes	No	Yes	Yes	No
DC-side Power Coupling	No	No	No	Yes	No
Integration	Low	Middle	Middle	High	High

. Compared with the existing solutions, the proposed topology has a significant advantage in terms of the number of high-frequency switches, reduced by up to 50%. Moreover, additional boost inductors and bulky, large-capacity capacitors are eliminated. This makes the proposed scheme a promising solution in terms of cost and power density.

4.4. Application Prospects

The proposed three-port single-stage DC–AC converter exhibits promising potential for further extension to both single-phase DC and three-phase AC applications. The modular design and flexible power flow control inherent in the topology allow for straightforward transplantation.

Regarding extension to single-phase DC systems, such as DC microgrids or DC-coupled EV charging stations, the proposed topology can be adapted by eliminating the grid-side inverter stage and directly interfacing the DC ports with a common DC bus. This would simplify the control structure and reduce conversion stages, potentially improving overall efficiency. In addition, further research is also required to address the control complexity, component sizing, and system-level integration for extended topologies.

For three-phase AC grid integration, the proposed single-phase structure can be extended to form a three-phase system, enabling higher power transfer capability, such as active and reactive power control. For instance, the secondary circuit can be replaced by a T-type converter and a Swiss converter to interface the three-phase utility grid. The extended circuit topology for three-phase grid-connected applications is shown in Figure 14. Being similar to the unfolding bridge, the Swiss converter operates at low frequency. However, this extension introduces challenges in terms of increased component count and more complex control algorithms.

5. Conclusions

This paper presents a three-port single-stage DC–AC converter for photovoltaic energy storage–charging systems. Compared with conventional two-stage multiport topologies, the number of high-frequency switches is reduced by 50%, leading to lower circuit complexity and cost. By coupling multiple DC sources through sub-transformers, the proposed topology achieves single-stage power conversion, decoupling control, and high expansion flexibility. To satisfy the flexible bidirectional power interaction requirements between the grid and multiple sources, an EPS-based modulation strategy is proposed, thereby enabling independent and precise power control of each port. Simulation results under multiple operating conditions confirm that the proposed topology achieves precise current control and bidirectional power transmission, demonstrating its suitability for integrated multiport power conversion applications.