Addressing the inherent limitations of conventional renewable energy generation systems, which suffer from high energy losses and redundant power devices due to multi-stage cascaded topologies, this thesis proposes an improved integrated LLC three-port converter topology. This design synergistically combines the advantages of a full-bridge LLC resonant converter and an interleaved Buck/Boost converter, enabling inter-port power flow control and wide-range soft-switching operation. The proposed system significantly enhances power conversion efficiency and power density for the energy storage system. A coordinated control strategy is developed based on the functional characteristics of the three ports. The primary contributions of this work are systematically presented, covering topology construction, operating characteristics, control strategy design, and comprehensive experimental validation.
1. Introduction and Background
The rapid development of the global economy and technology has led to a continuous increase in the demand for fossil fuels such as petroleum, coal, and natural gas. The extensive use and over-exploitation of these non-renewable resources have resulted in their gradual depletion. Data from the International Energy Agency indicates that the remaining extractable years for global oil, coal, and natural gas are only 50, 132, and 52 years, respectively. Concurrently, the widespread use of fossil fuels has triggered severe environmental issues, including greenhouse gas emissions, air pollution, water contamination, and ecological degradation.
To address these challenges and reduce dependence on fossil fuels, the development of clean and renewable energy sources has become a global consensus. Solar and wind energy, characterized by their cleanliness and sustainable utilization, have emerged as crucial alternatives. However, renewable energy sources are profoundly influenced by natural conditions, rendering their power generation uncontrollable and unstable. This intermittency and volatility make it difficult for photovoltaic (PV) generation systems to meet real-time load demands.
To solve the intermittent power supply problem in PV systems and enhance the reliability and stability of power supply, researchers have proposed incorporating energy storage units such as batteries into PV systems to balance active power. When the PV system’s generation is less than the load demand, the battery provides a portion or even all of the required energy. Conversely, when generation exceeds load demand, surplus energy is stored in the battery. The integration of the energy storage system is now a vital component in applications such as renewable energy microgrids, spacecraft power systems, distributed generation, and electric vehicles.
2. Topology Construction of the Three-port Converter
2.1 Analysis of the Full-bridge LLC Resonant Converter
The full-bridge LLC resonant converter is an ideal topology choice for PV power generation systems due to its advantages of low voltage stress and soft-switching characteristics across the full power range. The circuit topology comprises a primary-side full-bridge switching network, a resonant network, and a secondary-side rectifier network. The primary-side four power switches S1-S4 form the switching network. The resonant network consists of a series resonant inductor Lr, a resonant capacitor Cr, a parallel magnetizing inductor Lm, and a high-frequency isolation transformer T. The secondary-side full-bridge rectifier network is composed of four diodes D1-D4 and a filter capacitor Co.
Based on the fundamental harmonic approximation (FHA) method, the key parameters can be derived. The normalized voltage gain of the full-bridge LLC resonant converter is expressed as:
$$M(f_n) = \frac{1}{\sqrt{\left(1 + \frac{1}{k} – \frac{1}{k f_n^2}\right)^2 + Q^2\left(f_n – \frac{1}{f_n}\right)^2}}$$
where \( f_n = f_s / f_r \) is the normalized switching frequency, \( k = L_m / L_r \) is the inductance ratio, and Q is the quality factor. The gain characteristics were analyzed under various Q and k values, revealing that the design typically selects k between 4-6, Q between 0.3-0.5, and \( f_n \) in the range of 0.6-1.4 to ensure soft-switching conditions and cover a wide load variation range.
2.2 Analysis of the Interleaved Buck/Boost Converter
The interleaved Buck/Boost (IBB) converter is a bidirectional DC-DC topology widely used in energy storage systems. It offers structure simplicity, control flexibility, high efficiency, and fast dynamic response. The output voltage \( U_b \) and input voltage \( U_{in} \) are related by the duty cycle D in Buck mode, and by 1/(1-D) in Boost mode. Figure presents the current waveform of inductors Lb1 and Lb2 under different duty cycles.
The current ripple \( \Delta I \) of a single branch inductor can be expressed as:
$$\Delta I = \frac{U_b (1-D)}{L f_s} = \frac{U_b}{L f_s}\left(1 – \frac{U_b}{U_{in}}\right)$$
The critical inductance for boundary conduction mode is:
$$L = \frac{U_b}{2 P f_s}\left(1 – \frac{U_b}{U_{in}}\right)^2$$
The relationship between branch inductance and output voltage at different power levels is illustrated, indicating that inductance should be selected by balancing ripple, efficiency, and cost. Under high transmission power, larger inductance values are required to reduce ripple, while considering heat dissipation and efficiency.
2.3 Proposed Integrated Three-port Converter Topology
Combining the advantages of the two aforementioned circuit topologies, this thesis proposes an improved integrated PV energy storage three-port converter topology, as shown in the following figure.
In this topology, switches S1, S2 and inductor Lb1, along with switches S3, S4 and inductor Lb2, form the interleaved bidirectional Buck/Boost circuit. Switches S1-S4, diodes D1-D4, resonant inductor Lr, resonant capacitor Cr, and magnetizing inductor Lm together constitute the LLC full-bridge resonant converter. The bidirectional interleaved Buck/Boost converter and the full-bridge LLC resonant converter constitute an integrated three-port converter through bridge-arm复用. The PV array is connected to Port 1, the battery to Port 2, and the load to Port 3.
3. Operating Mode Analysis
The proposed three-port converter adopts a hybrid modulation strategy combining PWM, PFM, and Phase Shift Modulation (PSM). The output voltage \( u_o \) is jointly determined by the duty cycle D and switching frequency \( f_s \). The resonant tank input voltage \( u_{tank} \) is a three-level rectangular wave with positive-negative symmetry, with amplitude approximately equal to the PV port voltage. The converter exhibits two resonant frequencies: the LC series resonant frequency \( f_r = 1/(2\pi\sqrt{L_r C_r}) \) and the LLC resonant frequency \( f_m = 1/(2\pi\sqrt{(L_r + L_m) C_r}) \), where \( f_m < f_r \).
The boundary conditions for different operating modes are summarized in Table 1.
| Operating Mode | Half-cycle Resonance State | Boundary Condition | Specific Boundary |
|---|---|---|---|
| Mode I (D≤0.5) | LC-P, LC-0, LLC-0 | Dtank Ts ≤ 0.5 Tr | D ≤ 0.5 fn |
| Mode I (D＞0.5) | LC-P, LC-0, LLC-0 | Dtank Ts ≤ 0.5 Tr | D ＞ 1-0.5 fn |
| Mode II (D≤0.5) | LC-P, LLC-P, LLC-0 | Dtank Ts ＞ 0.5 Tr | D ≥ 0.5 fn |
| Mode II (D＞0.5) | LC-P, LLC-P, LLC-0 | Dtank Ts ＞ 0.5 Tr | D ＜ 1-0.5 fn |
A detailed modal analysis was conducted for Mode I (D≤0.5). In the interval [t0, t1], switch S1 turns on, the resonant tank voltage equals the PV voltage, and \( i_{Lr} > i_{Lm} \). The secondary-side rectifier diodes D1,4 conduct, transferring energy to the load. This state is defined as the LC-P resonance mode. During [t1, t2], the dead-time interval, the junction capacitors of S1 and S2 charge/discharge, enabling ZVS for S2. In [t2, t3], the resonant tank voltage is zero (LC-0 mode), and the Lr and Cr resonate to deliver energy. At t3, \( i_{Lr} \) equals \( i_{Lm} \), the secondary-side current reaches zero, and diodes D1,4 achieve Zero Current Switching (ZCS). This mode is defined as LLC-0, where Lm, Lr, and Cr resonate together. In [t4, t5], the junction capacitors of S3 and S4 charge/discharge, enabling ZVS for S3. The time-domain expressions for the resonant elements are presented for each interval, confirming the theoretical analysis.
4. Operating Characteristics of the Three-port Converter
4.1 DC Voltage Gain Characteristics
To analyze the system, the three-port converter is simplified into equivalent two-port networks. The DC voltage gains are defined as:
PV-to-battery gain: \( M_1 = \frac{U_{bat}}{U_{pv}} \)
Battery-to-load gain: \( M_{bat\_o} = \frac{nV_o}{V_{bat}} \)
PV-to-load gain: \( M_2 = \frac{nV_o}{V_{pv}} \)
These gains are related by: \( M_2 = M_{bat\_o} \cdot M_1 \). The resonant tank’s three-level input voltage renders the traditional FHA method inaccurate. Therefore, a time-domain analysis method is employed to establish the steady-state gain equations. Since solving these complex equations directly is challenging, a numerical fitting approach was adopted. By substituting specific parameter values into the equations, gain data were calculated and curve-fitted. The resulting fitting expressions for the PV-to-load DC gain under different conditions are derived.
For example, when considering the influence of duty cycle D and normalized frequency \( f_n \), the fitted gain expression is:
$$M_2(D, f_n) = 2.2 – 4.96 |\Delta f_n| + 3.75 \Delta D + \frac{10}{3} \Delta D^2$$
The gain data for different values of Q and k are summarized in Tables 2 and 3.
| Q | ƒn=0.65 | ƒn=0.70 | ƒn=0.75 | ƒn=0.80 | ƒn=0.85 | ƒn=0.90 | ƒn=0.95 | ƒn=1.0 |
|---|---|---|---|---|---|---|---|---|
| Q=0.1 | 1.486 | 1.326 | 1.205 | 1.125 | 1.058 | 1.010 | 0.973 | 0.940 |
| Q=0.2 | 1.484 | 1.324 | 1.197 | 1.108 | 1.044 | 0.987 | 0.944 | 0.910 |
| Q=0.5 | 1.482 | 1.322 | 1.191 | 1.092 | 1.010 | 0.946 | 0.894 | 0.850 |
| Q=1.0 | 1.480 | 1.320 | 1.188 | 1.076 | 0.986 | 0.912 | 0.852 | 0.800 |
| k | ƒn=0.65 | ƒn=0.70 | ƒn=0.75 | ƒn=0.80 | ƒn=0.85 | ƒn=0.90 | ƒn=0.95 | ƒn=1.0 |
|---|---|---|---|---|---|---|---|---|
| k=3 | 1.78 | 1.49 | 1.31 | 1.18 | 1.03 | 1.01 | 0.96 | 0.9 |
| k=4 | 1.49 | 1.32 | 1.20 | 1.11 | 1.05 | 1.00 | 0.95 | 0.9 |
| k=5 | 1.35 | 1.24 | 1.15 | 1.08 | 1.02 | 0.99 | 0.95 | 0.9 |
| k=6 | 1.28 | 1.19 | 1.12 | 1.06 | 1.01 | 0.98 | 0.94 | 0.9 |
| k=7 | 1.23 | 1.16 | 1.09 | 1.04 | 1.00 | 0.97 | 0.94 | 0.9 |
4.2 Soft-switching Characteristics
The integrated LLC three-port converter can achieve ZVS for the primary-side switches by utilizing the body diodes and junction capacitances of the MOSFETs. By analyzing the inductor currents at switching moments, the ZVS conditions can be derived. The simplified unique ZVS condition is given as:
$$i_{ZVS}(t_{S12}) = i_{Lr}(t_{S12}) – i_{Lb1}(t_{S12}) > 0$$
Different boundary conditions apply for Modes I and II. Under no-load conditions, the circuit may not achieve ZVS. The ZVS boundary conditions for the different operating modes are summarized below, considering both output load and duty cycle:
(1) Mode I, D ≤ 0.5: \( P < \frac{U_{pv}D}{L_{b1,2} f_r f_m n} + \frac{nU_o}{L_m f_r} \cdot \frac{D}{f_n} \)
(2) Mode I, D > 0.5: \( P < \frac{U_{pv}D(1-D)}{2 L_{b1,2} f_r f_m} + \frac{nU_o}{L_m f_r} \cdot \frac{1}{2} \)
4.3 Current Ripple Analysis
The current ripple of a single branch inductor in the interleaved Buck/Boost circuit is \( \Delta i_{Lb} = \frac{V_{pv} \cdot D(1-D)}{L_b f_s} \). The total current ripple after interleaving is significantly reduced. The ripple ratio λ depends on the duty cycle D and the phase-shift angle φ between the bridge arms. For different ranges of D and φ, the ripple ratio expressions were derived, showing that as φ approaches π and D approaches 0.5, the total current ripple can be minimized, even reaching zero in ideal conditions.
4.4 Power Transmission Modes
The three-port converter operates in four distinct power transmission modes based on power flow:
Mode A (PV alone supplies load): When PV power is sufficient to meet load demand and the battery is not charging, \( P_{pv} = P_o \).
Mode B (Battery alone supplies load): During nighttime or low irradiance, the battery supplies the load, \( P_{bat} = P_o \).
Mode C (PV supplies load and battery): When PV power exceeds load demand, the surplus charges the battery, \( P_{pv} = P_{bat} + P_o \).
Mode D (PV and battery co-supply load): When PV power is insufficient for load demand, both sources supply the load, \( P_{pv} + P_{bat} = P_o \).
Each power transmission path contains an independent controllable switch, enabling decoupled power flow control between any two ports.
5. Control Strategy Design for the Stand-alone PV Energy Storage System
5.1 Improved MPPT Control Strategy
To overcome the low tracking efficiency of conventional Perturb and Observe (P&O) algorithms under rapidly changing irradiance, an adaptive variable-step MPPT algorithm is proposed. The algorithm combines the advantages of the Constant Voltage Tracking (CVT) method and the variable-step P&O method. The control flow dynamically adjusts the perturbation step size based on the power change ratio. When the PV power is below a threshold (P < Pmin), reflecting low irradiance conditions, the system switches to CVT mode to maintain a stable reference voltage. In contrast, under high irradiance, it transitions to the variable-step P&O method. This hybrid strategy ensures both high tracking speed and precision while improving the system’s adaptability to environmental changes.
5.2 Battery Charging and Discharging Control
To ensure safe battery operation, a charge/discharge control strategy with SOC state feedback was designed. The strategy primarily determines the function based on battery voltage (ub), current (ib), and SOC. When the battery voltage exceeds the reference value or SOC exceeds the maximum limit (overcharge protection), the system exits MPPT mode and switches to Constant Voltage (CV) charging mode, managed by the Battery Voltage Regulator (BVR). When the battery operates within the safe range, normal MPPT mode is maintained. If the discharge current exceeds the limit or SOC falls below the minimum, Constant Current (CC) discharge protection is triggered, managed by the Battery Current Regulator (BCR), to reduce the output power and protect the battery. This is crucial for the longevity and safety of the energy storage system.
5.3 Hybrid Control Strategy
To address the conflicts between PV MPPT and battery safety control, a minimum-value competition mechanism is introduced. This mechanism selects the minimal output from the four control loops (IVR, BVR, BCR, and OVR) to determine the system’s operating mode. When the battery voltage or current is within safe limits, the MPPT controller’s output dominates. As the battery approaches its limits, the battery controller’s output decreases and, when lower than the MPPT output, takes precedence, forcing the PV array to exit MPPT state. This ensures automatic and seamless mode switching within the energy storage system.
A hybrid control strategy combining PWM, PFM, and PSM is proposed. The duty cycle D controls the power balance between PV and the battery. The operating frequency fs controls the output voltage regulation. The phase-shift angle φ between the bridge arms provides an additional degree of freedom for output voltage control. This approach significantly widens the output voltage range and enhances the dynamic adaptability and steady-state accuracy of the system. The overall control block diagram integrates these three variables, managed by the mixed modulation module, to generate the driving signals for the switching devices. The Min-selection control strategy diagram is implemented with PI controllers for each port.
5.4 System Simulation Parameters
A simulation model of the stand-alone PV energy storage three-port converter system was built in MATLAB/Simulink. The key simulation parameters are summarized in Table 4.
| Parameter | Value | Parameter | Value |
|---|---|---|---|
| PV port voltage Vpv | 90-100 V | PV input power | 300-1200 W |
| Battery voltage Vbat | 48 V | Battery power | 0-500 W |
| Load port voltage Vo | 45-50 V | Load port power | 500 W |
| Switching frequency fs | 70-120 kHz | Transformer turns ratio | 2:1 |
| Resonant inductor Lr | 13.62 uH | Resonant capacitor Cr | 186.3 nF |
| Magnetizing inductor Lm | 55 uH | Filter inductor Lb1/Lb2 | 50 uH |
6. Simulation and Experimental Verification
6.1 Simulation Results
The simulation results demonstrate the dynamic performance of the proposed control strategies under varying irradiance conditions. When the irradiance stepped from 1000 W/m² to different levels, the proposed MPPT algorithm achieved faster tracking and smaller steady-state oscillations compared to the conventional P&O and CVT methods. The voltage ripples remained within ±2%, and the load output voltage was maintained at 48V.
The three-port converter was shown to maintain stable output voltage under varying irradiance and load conditions. The simulation waveforms also confirmed the ZVS operation of the primary-side switches and ZCS operation of the secondary-side diodes, validating the soft-switching analysis. The interleaved Buck/Boost output current ripple was significantly reduced, from 3.2 A for a single inductor to only 0.5 A for the combined output current, aligning with the theoretical analysis. The modulation waveforms for the hybrid control strategy, including the generation of dead time and phase-shift angles, were also successfully verified.
The simulation of the minimum-value competition mechanism showed smooth transitions between MPPT, Constant Current (CC) charging, and Constant Voltage (CV) charging modes under different conditions, effectively protecting the battery in the energy storage system.
6.2 Experimental Verification
To validate the feasibility and engineering applicability of the proposed topology and control strategy, a 500W experimental prototype platform was constructed. The system employs a TMS320F28335 DSP for centralized control. A PV simulator was used in place of actual PV panels, and four 12V gel batteries connected in series (48V) were used as the energy storage unit.
Steady-state Characteristics: The experimental results showed that the port voltages were stable at approximately 100V (PV input), 50V (battery), and 48V (output load). The measured ZVS waveform for the primary-side switches and ZCS waveform for the secondary-side diodes were consistent with the theoretical and simulation analysis.
Energy Management: Experiments were conducted to verify system operation in different modes. When the PV power was 400W, below the rated load of 500W, the battery was in discharge mode (DISO mode) providing the power difference. When the PV power was increased to 800W, the battery transitioned to charging mode (SIDO mode) to absorb the surplus energy. Throughout the transitions, the output voltage remained stable at 48V. A scenario simulating a sudden drop in PV power (from 400W to 0W) showed a rapid response from the battery, with the discharge current increasing from 2.2A to 10.5A, while the output voltage stayed stable, demonstrating excellent dynamic response and energy compensation capability.
MPPT Dynamics: By gradually increasing the PV port voltage, the system’s ability to track the maximum power point was demonstrated. The battery automatically switched between charging and discharging states to maintain the power balance, with the output voltage consistently regulated at 48V.
Load Step Change: When the load current stepped between 10.5A (full load) and 5.2A (half load), the battery seamlessly balanced the power difference. The voltage deviation was controlled within ±1.5%, and the energy switching response time was approximately 15ms, demonstrating excellent load regulation capability. The full-load efficiency reached 96.8%.
7. Conclusion and Future Work
This thesis has systematically investigated the design, analysis, and control of a three-port converter based stand-alone photovoltaic energy storage system. The main contributions and conclusions are as follows:
An improved integrated three-port converter topology was constructed by combining a full-bridge LLC resonant converter and an interleaved Buck/Boost converter. This topology achieves high power density, low current ripple, and wide-range soft-switching operation for the energy storage system. Time-domain analysis and numerical fitting methods were employed to derive the accurate voltage gain expressions. The soft-switching boundary conditions and current ripple characteristics were thoroughly analyzed, providing a solid foundation for design and optimization.
An adaptive variable-step MPPT control algorithm was proposed to improve tracking efficiency under varying irradiance conditions. A minimum-value competition mechanism was designed for battery charge/discharge control, effectively resolving conflicts between MPPT and battery safety. A hybrid control strategy combining PWM, PFM, and PSM was implemented, which significantly widens the output voltage range and enhances the system’s dynamic adaptability and steady-state accuracy for the energy storage system.
Both simulation and experimental results demonstrated stable voltage output and efficient power transfer under various conditions. The system effectively managed energy flow between the PV, battery, and load ports, validating the correctness and engineering feasibility of the proposed topology and control strategy. Future work could explore methods to simplify the design process by deriving a more direct analytical expression for the system gain. The control strategies could also be extended to more complex applications, such as grid-connected scenarios and higher-power systems, to further advance the development and large-scale application of renewable energy systems.