In the context of global efforts toward carbon neutrality, the large‑scale integration of renewable energy sources introduces significant challenges to power system stability. To mitigate these issues, battery energy storage systems have become indispensable. Among various topologies, the cascaded H‑bridge (CHB) converter stands out as a highly modular, scalable solution suitable for medium‑ and high‑voltage grid connections. However, the performance of such battery energy storage systems heavily depends on two critical aspects: the state‑of‑charge (SoC) equalization among distinct battery modules and the suppression of transient voltage surges at the DC side. In this paper, we propose a unified control strategy that simultaneously addresses SoC balancing and temporary voltage overshoot, thereby enhancing both the energy utilization and the operational reliability of cascaded H‑bridge battery energy storage systems.
The typical configuration of a single‑phase CHB converter in a battery energy storage system is illustrated in the schematic below. Each submodule consists of a battery pack connected to a DC/DC converter, which feeds a full‑bridge (FB) cell. The FB cells are cascaded and connected to the AC grid through a filter inductor. To prevent the double‑line‑frequency ripple from penetrating into the battery, a two‑stage topology is adopted. Table 1 lists the key circuit parameters used in our simulations and experimental validations.
| Parameter | Value |
|---|---|
| Grid phase voltage (RMS) | 5.77 kV (simulation) / 6 kV (experiment) |
| Rated power | 1 MW |
| Number of submodules | 10 (simulation) / 4 (experiment) |
| Filter inductance | 5 mH |
| Submodule DC‑link capacitance | 5 mF |
| Rated DC‑link voltage | 1 kV (simulation) / 2.5 kV (experiment) |
| Battery rated capacity | 15 Ah (simulation) / 1500 Ah (experiment) |
| Battery rated voltage | 0.6 kV (simulation) / 1.2 kV (experiment) |
| Boost inductance | 5 mH |
SoC Equalization Control Strategy
The instantaneous SoC of each battery module is related to its initial value, capacity, and output power. For a discharging scenario, a higher power leads to a faster decrease in SoC; for charging, the opposite holds. To balance the SoC among modules, we propose an adaptive power‑sharing coefficient ki that adjusts the power reference of each DC/DC converter. The total active power of the system is kept unchanged by ensuring that the sum of the individual adjustments is zero.
The power reference for the i‑th module is expressed as:
$$P_{\text{ref},i} = k_i \cdot P_{\text{ref}}$$
where Pref is the total active power command. The coefficient ki is defined as:
$$k_i = \frac{1}{N} + K_{\text{SoC}} \cdot \text{sign}(P_{\text{ref}}) \cdot \Delta\text{SoC}_i$$
with ΔSoCi = SoCi – SoCave and KSoC being an adaptive gain. The gain is designed to be small (Kmin) when the maximum deviation ΔSoCmax exceeds a threshold, and then increases according to a power‑law function as the deviation shrinks, until it saturates at Kmax. This technique ensures a fast equalization speed without causing excessive power imbalance. The computational procedure for KSoC is summarised in Table 2.
| Condition | Expression for KSoC |
|---|---|
| ΔSoCmax > ΔSoCthreshold | KSoC = Kmin (e.g., 0.05) |
| ΔSoCthreshold > ΔSoCmax > ΔSoCmin | KSoC = a1 + a2·(ΔSoCmax)a3 |
| ΔSoCmax ≤ ΔSoCmin | KSoC = Kmax (upper saturation) |
Our design guarantees that each module’s power stays within 0.8 to 1.2 times the rated value, thus preserving system stability. The SoC evolution for all batteries converges monotonically, as confirmed by both simulation and experiments.
Transient Voltage Surge Suppression
In battery energy storage systems employing the CHB topology, sudden reversals of the active‑power command cause a mismatch between the response speeds of the DC/DC converter (which actively controls battery power) and the DC/AC stage (which passively balances the power). This mismatch leads to an instantaneous accumulation or release of energy in the DC‑link capacitors, resulting in a severe voltage overshoot. To analyze this phenomenon, we derive a small‑signal model of the DC/DC converter. Assuming the current inner loop is sufficiently fast, the closed‑loop transfer function from the power reference disturbance to the capacitor voltage disturbance yields the characteristic equation that determines the damping ratio ζ:
$$\zeta \approx \frac{K_{P,p} + \lambda\alpha}{2\sqrt{\alpha\beta + \lambda\beta\alpha}}$$
where KP,p is the proportional gain of the power outer loop, and α, β, λ are constants depending on operating points and circuit parameters. The damping ratio strongly influences the overshoot magnitude. To mitigate the voltage surge, we propose a capacitor‑voltage feedforward strategy. When the deviation between the average capacitor voltage and its reference exceeds a predetermined threshold (e.g., 5 % of the steady‑state value), the PI gains of the power loop are temporarily reduced, thereby increasing the damping ratio and limiting the overshoot. The gains are then restored using a sigmoid‑like function to quickly track the new power command once the voltage settles. Table 3 compares the peak overshoot under different control schemes.
| Control Strategy | Overshoot at discharge→charge (%) | Overshoot at charge→discharge (%) |
|---|---|---|
| Conventional (no SoC balancing) | 19 | 17 |
| With SoC balancing (no voltage suppression) | 12 | 11 |
| Proposed compound strategy | 9.7 | 9.5 |
Compared with the conventional method, the proposed voltage‑surge suppression reduces the peak overshoot by approximately 50 %, which is crucial for protecting capacitors and prolonging the system lifetime in high‑power battery energy storage systems.
Simulation and Experimental Validation
We built a detailed simulation model in MATLAB/Simulink for a 10‑module single‑phase CHB‑BESS, and a hardware‑in‑the‑loop (HIL) platform using RTDS with a DSP+FPGA controller for a 4‑module system. The initial SoC values for the simulation were deliberately spread around 70 % to test the equalization algorithm. Figure (a) shows the four‑quadrant operation capability of the system; the active and reactive power commands are accurately tracked. In the absence of SoC balancing, all batteries output identical power and their SoC curves remain parallel. After enabling the proposed adaptive balancing at 0.5 s, the power references diverge according to each battery’s SoC, leading to a converging SoC trajectory. The system reaches equilibrium in about 8 s, which is much faster than conventional fixed‑gain methods without inducing instability.
The HIL experiments further confirm the effectiveness. The initial SoC of the four modules were set to 69.8 %, 70.2 %, 70.4 %, and 69.7 %. After operating the system for half an hour under the proposed control, the SoC values became nearly identical (maximum deviation below 0.1 %). The capacitor voltage overshoot during power reversal was measured at 10.4 % and 8.8 % without the surge suppression; with the feedforward control, it dropped to 4.8 % and 4.0 %, respectively. These experimental results align closely with the simulation trends and demonstrate the practical viability of our approach.
Table 4 summarizes the experimental platform parameters used for the HIL tests.
| Parameter | Value |
|---|---|
| Grid phase voltage (RMS) | 6 kV |
| Number of submodules | 4 |
| Rated DC‑link voltage | 2.5 kV |
| Battery rated voltage | 1.2 kV |
| Battery rated capacity | 1500 Ah |
Conclusion
In this work, we have developed a comprehensive control framework for cascaded H‑bridge battery energy storage systems that integrates SoC balancing and transient voltage surge mitigation. The adaptive power‑sharing coefficient yields a fast and stable equalization across modules without exceeding safe power limits. The voltage feedforward mechanism effectively dampens the DC‑link overshoot by dynamically adjusting the power‑loop gains, reducing the peak by approximately 50 %. Both simulation and hardware‑in‑the‑loop experiments validate the superior performance of the proposed strategy. Our solution can be directly extended to three‑phase systems and is well suited for large‑scale grid‑connected battery energy storage systems where reliability and energy efficiency are of paramount importance.