In the context of the global energy transition and the “dual carbon” strategy, the penetration of renewable energy sources such as wind and photovoltaic power in the national energy structure has been steadily increasing. However, the inherent intermittency and uncertainty of environmental factors like wind speed and solar irradiance pose significant challenges to the stable operation of power grids. To mitigate these disturbances, energy storage battery systems are indispensable, playing a critical role in maintaining voltage and frequency stability of the power system. In practical applications, multiple energy storage battery units are typically connected in parallel to a DC bus in a distributed manner to enhance overall system capacity. Nevertheless, due to factors such as unreasonable load power distribution, differences in initial states, and inconsistencies in control precision, the state of charge (SOC) among multiple energy storage battery units often diverges during actual operation. This SOC inconsistency can lead to overcharging or over-discharging of some units, causing them to exit operation prematurely, thereby compromising the stability and reliability of the entire system. Therefore, investigating effective methods to balance the SOC among clusters of energy storage converters is of great significance for ensuring the stable operation of power systems.
Addressing the SOC balancing problem, researchers have primarily focused on two levels: the battery management system (BMS) and the power conversion system (PCS). At the BMS level, some studies have proposed cascaded H-bridge converters to indirectly regulate voltage and achieve SOC balancing by monitoring output power and adjusting switching operations. Other approaches have designed multi-winding transformers to control current flow between batteries, enabling uniform SOC distribution. Some research has employed a combination of solid-state relays and DC-DC converters to manage SOC balancing by controlling the on/off states of relays. However, these SOC balancing methods based on BMS often involve complex hardware structures, leading to significantly increased costs, and multi-stage energy conversion inevitably results in efficiency losses. Consequently, at the PCS level, improved strategies based on droop control have become a research hotspot. For instance, some researchers have adjusted the frequency under no-load conditions in droop control to achieve differentiated frequency management for battery groups with different SOC levels. Others have set dynamic adjustment factors based on the SOC difference; when the difference is large, the adjustment factor accelerates the response speed, and when the difference approaches zero, it amplifies the disparity to speed up balancing. Several studies have introduced the hyperbolic tangent function to achieve adaptive adjustment of the droop coefficient through SOC, utilizing the characteristics of the hyperbolic tangent function and the adjustment of correction coefficients to confine the droop coefficient within a certain range, thereby realizing rapid SOC balancing. Some researchers have introduced power factors and power functions to achieve grid-connected voltage stability through the offset of the output voltage of dual converters. Other approaches have adopted distributed communication means to achieve system SOC balancing by comparing the actual SOC values of each station with the average value. Some studies have introduced an acceleration factor to enable the energy storage battery to release maximum power with an optimized droop coefficient during the initial discharge phase. Nevertheless, methods relying on communication means for regulation often lack sufficient adaptive capability when facing complex and variable system dynamics.
Despite the progress made, existing SOC balancing methods, whether BMS-based schemes or PCS-based improved droop control strategies, generally face common challenges when pursuing fast, efficient, and low-cost balancing. These challenges include stability and dynamic performance trade-offs, as well as difficulties in parameter optimization design. To address these issues, this paper proposes a segmented dynamic droop control strategy. By establishing a mathematical model of the energy storage battery system and introducing an SOC deviation threshold mechanism, the balancing process is divided into a fast balancing region and a precise adjustment region, taking into account both balancing speed and accuracy. This research provides a new approach to solving the SOC balancing problem of energy storage battery systems in new energy power grids, holding significant engineering importance for improving system operation efficiency and extending battery life.
Analysis of Energy Storage Converter Control Methodology
To effectively address the SOC balancing challenge, I first establish a comprehensive control framework for the energy storage battery system. The overall control block diagram of the energy storage battery system comprises several key components: the energy storage battery itself, a DC-AC converter, filtering inductors and capacitors, and the control loops for power calculation, droop control, voltage regulation, and current regulation. The energy storage battery voltage is denoted as \(U_{bat}\), the current flowing through the energy storage inductor \(L_c\) is \(i_{Lc}\), the DC side voltage of the energy storage converter is \(U_{dc}\), the current through the filter inductor \(L\) is \(i_L\), the AC side output voltage of the energy storage converter is \(u_o\), and the AC side output current is \(i_o\). The active power is represented by \(P\), the reactive power by \(Q\), the reference voltage by \(u_{ref}\), and the reference current by \(i_{ref}\).
When multiple energy storage battery systems are connected in parallel, the structure becomes more complex. The output complex power of the i-th energy storage converter can be expressed as:
$$S = U_o I_o^* = U \angle \varphi \left( \frac{U \angle \varphi – U_{pcc} \angle 0}{Z_L \angle \theta} \right)^*$$
Expanding this expression, I obtain the active and reactive power components:
$$S = \frac{U [R_L (U – U_{pcc} \cos \varphi) + X_L U_{pcc} \sin \varphi]}{R_L^2 + X_L^2} – j \frac{U [X_L (U – U_{pcc} \cos \varphi) – R_L U_{pcc} \sin \varphi]}{R_L^2 + X_L^2}$$
In low-voltage microgrids, the line resistance \(R_L\) is significantly larger than the line reactance \(X_L\), i.e., \(R_L \gg X_L\). Under this condition, I can approximate \(\sin \varphi \approx \varphi\) and \(\cos \varphi \approx 1\). This simplification leads to a linear relationship between active power \(P\) and voltage amplitude \(U\), as well as between reactive power \(Q\) and phase angle \(\varphi\). Furthermore, by differentiating the phase angle, I can establish a linear relationship between reactive power \(Q\) and frequency \(f\). The resulting droop control equations are:
$$U = U^* – k_p P$$
$$f = f^* + k_q Q$$
Here, \(U^*\) and \(f^*\) represent the reference values for the output voltage amplitude and frequency, respectively. The parameter \(k_p\) is the active droop coefficient, and \(k_q\) is the reactive droop coefficient. These equations form the foundation of the P-U and Q-f droop control characteristics. The droop control method offers a distributed control approach that does not require complex communication networks. Each energy storage battery unit makes control decisions based on its local information, enhancing the robustness of the entire system. Even if a control component of one energy storage battery unit fails, it does not directly impact the control of other units.
Segmented Droop Control Considering SOC
Analysis of SOC Imbalance under Traditional Droop Control
The performance of an energy storage battery is influenced by various factors including materials, manufacturing processes, temperature, and environmental conditions. To analyze the SOC imbalance issue, I adopt the internal resistance model of the energy storage battery, which consists of a voltage source and a resistor. This model is simple in structure and convenient for parameter design, facilitating the analysis. I employ the ampere-hour integration method for SOC estimation:
$$SOC_i = SOC_{0-i} – \frac{\int I_{b-i} dt}{C_{e-i}}$$
where \(SOC_{0-i}\) is the initial SOC of the i-th energy storage battery, \(C_{e-i}\) is the capacity of the i-th energy storage battery, and \(I_{b-i}\) is the current of the i-th energy storage battery.
Assuming ideal operating conditions where the power consumed by the energy storage converter equals the power generated by the energy storage battery, I have:
$$SOC_i = SOC_{0-i} – \frac{\int P_i dt}{C_{e-i} U_{b-i}}$$
Taking the derivative of SOC with respect to time yields:
$$SOC_i’ = -\frac{P_i}{C_{e-i} U_{b-i}}$$
From this equation, I observe that the rate of change of SOC is closely related to the output power of the energy storage converter, the voltage of the energy storage battery, and the battery capacity. During steady-state operation, the output voltages of all parallel energy storage battery units remain consistent. Based on the power distribution principle of the droop control strategy, assuming all energy storage battery units have the same capacity, I can derive:
$$P_1 : P_2 : \ldots : P_n = \frac{1}{k_{p1}} : \frac{1}{k_{p2}} : \ldots : \frac{1}{k_{pn}} = SOC_1′ : SOC_2′ : \ldots : SOC_n’$$
The SOC of an energy storage battery depends on its initial value and the power \(P\) supplied by the energy storage converter. Specifically, SOC increases linearly with output power and is inversely proportional to the active droop coefficient \(k_p\). In the traditional droop control strategy, when the line impedances of the parallel system are consistent, the active power \(P\) is only related to \(k_p\), and \(k_p\) remains constant. However, when the initial SOC values are not uniform, each energy storage battery unit outputs a stable active power \(P_n\), resulting in the same rate of change of SOC for each system. This maintains the initial SOC difference state.
Since \(R_L \gg X_L\) in low-voltage microgrids, combining the relevant equations, I can express the active power output of each parallel energy storage battery unit under traditional P-U droop control as:
$$P_i = \frac{(U^* – U_{pcc}) U_{pcc}}{R_{Li} + k_p U_{pcc}}$$
From this relationship, when the line impedances are not equal, the conditions for SOC balancing among the parallel energy storage battery units cannot be satisfied, making it impossible to achieve SOC balance under traditional droop control.
Proposed SOC Segmented Droop Control Method
In the traditional droop control strategy, the droop coefficient is constant, making it difficult to adapt to changes in SOC state and resulting in imprecise balancing management of energy storage battery units. To solve this problem, I propose an improved control strategy that introduces a deviation threshold mechanism, dividing the SOC balancing process into a fast balancing region and a precise adjustment region. In the fast balancing region, when the SOC deviation exceeds the threshold, the system switches to fast balancing mode. A fixed droop coefficient is used to enable the energy storage battery group with high SOC to output power quickly, while the group with low SOC slows down its output, achieving dynamic balance. Once the SOC deviation decreases below the threshold, the system enters the precise adjustment region, where a proportional-integral (PI) controller adjusts the droop coefficient to gradually eliminate the SOC deviation, ensuring SOC consistency among energy storage battery groups and achieving precise balancing management.
The P-U droop control equation with dynamic adjustment is:
$$U = U^* – mP$$
The expression for the dynamically adjusted droop coefficient \(m\) under charging and discharging states is:
$$m = \begin{cases}
m_0 [1 – k (SOC_i – SOC_{ave})] & \text{discharging} \\
m_0 [1 + k (SOC_i – SOC_{ave})] & \text{charging}
\end{cases}$$
where \(m_0\) is the initial droop coefficient, \(k\) is the adjustment factor, and \(SOC_{ave}\) is the average SOC of all energy storage battery units, representing the overall assessment of the energy stored in each energy storage battery:
$$SOC_{ave} = \frac{\sum_{i=1}^{n} SOC_i}{n}$$
By combining the initial droop coefficient \(m_0\) with the deviation information of the actual storage state of each energy storage battery unit, the system can adaptively adjust the droop coefficient to better respond to changing load demands and environmental conditions. This strategy improves the adaptability and reliability of the energy storage battery system, helping to balance battery energy distribution, maintain battery health while meeting user power demands, reduce maintenance costs, and extend battery life.
The adjustment factor \(k\) is expressed as:
$$k = \begin{cases}
\frac{k_c (SOC_{0-i} – SOC_{0-ave})}{SOC_i – SOC_{ave}} & |\Delta SOC_i| > T_{SOC} \\
k_{mp} + \frac{k_{mi}}{s} & |\Delta SOC_i| \leq T_{SOC}
\end{cases}$$
Here, \(k_c\) is the initial adjustment coefficient, \(k_{mp}\) is the proportional coefficient of the PI controller, \(k_{mi}\) is the integral coefficient of the PI controller, \(SOC_{0-ave}\) is the initial average SOC of each energy storage battery, and \(T_{SOC}\) is the SOC deviation threshold. The threshold \(T_{SOC}\) is a key parameter that divides the fast balancing stage and the precise adjustment stage in the segmented droop control strategy. If the threshold is set too large, the fast balancing stage covers too wide a range, causing the system to enter the precise adjustment stage prematurely, and the remaining deviation takes longer to eliminate. If the threshold is too small, measurement errors may misjudge the balancing state, triggering oscillations in the control mode. In my simulation setup, the ampere-hour integration method with a 10 kHz sampling frequency has an error of less than 0.01%, almost approaching perfect integration, so I set \(T_{SOC} = 0.2\%\). The SOC deviation \(\Delta SOC_i\) is defined as:
$$\Delta SOC_i = SOC_i – SOC_{ave}$$
To illustrate the SOC balancing process under the proposed control strategy, I consider two parallel energy storage battery systems as an example. When the energy storage battery systems are in discharging mode and \(SOC_1\) is higher than \(SOC_2\), the control strategy enters the fast balancing region. This phase continues until \(\Delta SOC_i\) reaches the preset threshold \(T_{SOC}\). During this period, the droop coefficient of the first energy storage converter is fixed at \(m_1\), and that of the second converter is fixed at \(m_2\). These coefficients are finely adjusted constant parameters based on the initial SOC values and the deviation \(k_c\). During this phase, since \(P_1 > P_2\), the SOC of the first energy storage battery decreases at a relatively faster rate. When \(\Delta SOC_i\) reaches the \(T_{SOC}\) threshold, the control strategy transitions to the precise adjustment region. At this point, the SOC deviation is minimal, and the droop coefficient \(m\) is dynamically adjusted according to real-time SOC data, gradually approaching the initial value \(m_0\). When the SOC values of both parallel energy storage battery systems simultaneously reach the average value \(SOC_{ave}\), the SOC balancing state of the entire system is achieved, returning to the standard P-U droop control mode.
Parameter Design and Analysis
Although the droop control system for parallel energy storage battery converters has advantages, it may cause a certain degree of deviation in the grid voltage amplitude. Therefore, when setting the droop coefficient, it is crucial to ensure that the grid voltage remains within an acceptable deviation range. Considering the discharging mode, the design principle follows a clear criterion:
$$0 < m_0 [1 – k (SOC_i – SOC_{ave})] < \frac{\Delta U_{max}}{P_{max}}$$
This criterion revolves around two key parameters: the maximum possible amplitude deviation of the grid voltage \(\Delta U_{max}\), which should be limited to within ±5% of the nominal grid voltage, and the maximum output power of the energy storage converter \(P_{max}\). By precisely adjusting these two parameters, I can ensure the stable operation of the system and maintain the normal voltage level of the grid.
The relationship between the output power of the energy storage battery converter and the adaptive droop coefficient is:
$$P_1 : P_2 = \frac{1}{m_0 [1 – k (SOC_1 – SOC_{ave})]} : \frac{1}{m_0 [1 – k (SOC_2 – SOC_{ave})]}$$
The power of the common load \(P_{load}\) equals the sum of the output powers of the two energy storage battery systems connected to the grid. The output power deviation \(P_{12}\) between the two energy storage battery converters can be further derived as:
$$P_{12} = P_1 – P_2 = P_{load} \frac{k (SOC_1 – SOC_2)}{2}$$
The output power of each energy storage battery converter is limited within a certain range. When \(k = k_c (SOC_{0-i} – SOC_{0-ave}) / (SOC_i – SOC_{ave})\), the output power of the converter connected to the energy storage battery with higher SOC is:
$$P_1 = P_{load} \frac{[1 – k_c (SOC_{0-1} – SOC_{0-ave})]}{2} < P_{max}$$
The output power deviation between the two converters is:
$$P_{12} = P_1 – P_2 = P_{load} \frac{k_c (SOC_{0-1} – SOC_{0-2})}{2}$$
When \(k = k_{mp} + k_{mi}/s\), the output power of the converter connected to the energy storage battery with higher SOC is:
$$P_1 = P_{load} \frac{[1 – (k_{mp} + k_{mi}/s) (SOC_1 – SOC_{ave})]}{2} < P_{max}$$
The output power deviation between the two converters in this case is:
$$P_{12} = P_1 – P_2 = P_{load} \frac{(k_{mp} + k_{mi}/s) (SOC_1 – SOC_2)}{2}$$
In the fast balancing region, the setting of \(k_c\) has a significant impact on the power limit of the energy storage converter. In the precise adjustment region, the value of \(k_{mp}\) directly affects the power output of the energy storage converter. To analyze the influence of parameter changes on the converter output power, I set the initial SOC values of two energy storage battery units to 80% and 70% respectively, with an average SOC of 75%, \(P_{max} = 1000\) W, \(P_{load} = 1200\) W, and \(T_{SOC} = 0.2\%\).
When \(|\Delta SOC_i| > T_{SOC}\), the output power deviation remains constant. As the parameter \(k_c\) increases, the power deviation of the energy storage converter correspondingly increases, prompting the rate of change of SOC of the energy storage battery to accelerate. When \(|\Delta SOC_i| \leq T_{SOC}\), if the SOC deviation remains unchanged, an increase in \(k_{mp}\) and \(k_{mi}\) directly leads to an increase in the output power deviation, similarly driving an increase in the rate of change of SOC of the energy storage battery.
Therefore, for parameter design in the fast balancing region, the selection of \(k_c\) should be as large as possible without exceeding the power limit. For parameter design in the precise adjustment region, the setting of \(k_{mp}\) should ensure that it matches the peak output power of the energy storage converter before the threshold. The selection of \(k_{mi}\) must be made while ensuring system stability.
| Parameter | Value | Parameter | Value |
|---|---|---|---|
| Energy storage battery voltage \(U_{bat}\) | 200 V | Switching frequency \(f_s\) | 10 kHz |
| Filter inductance \(L\) | 4.5 mH | Initial active droop coefficient \(m_0\) | 0.005 |
| Filter capacitance \(C\) | 4.7 μF | Reactive droop coefficient \(k_q\) | 0.006 |
| Line impedance \(Z_{L1}\) | 0.0642 + j0.005 Ω | Proportional coefficient \(k_{mp}\) | 310 |
| Line impedance \(Z_{L2}\) | 0.2842 + j0.005 Ω | Integral coefficient \(k_{mi}\) | 0.05 |
| Converter maximum power \(P_{max}\) | 1000 W | Energy storage battery capacity | 0.5 Ah |
| Operating Condition | Operating State | SOC Initial Values | Line Impedance |
|---|---|---|---|
| Condition 1 | Discharging | SOC0-1 = 80%, SOC0-2 = 70% | \(Z_{L1} = Z_{L2} = 0.0642 + j0.005\ \Omega\) |
| Charging | SOC0-1 = 40%, SOC0-2 = 30% | \(Z_{L1} = Z_{L2} = 0.0642 + j0.005\ \Omega\) | |
| Condition 2 | Discharging | SOC0-1 = SOC0-2 = 80% | \(Z_{L1} = Z_{L2} = 0.0642 + j0.005\ \Omega\) |
| Charging | SOC0-1 = SOC0-2 = 30% | \(Z_{L1} = Z_{L2} = 0.2842 + j0.005\ \Omega\) | |
| Condition 3 | Discharging | SOC0-1 = 80%, SOC0-2 = 70% | \(Z_{L1} = 0.0642 + j0.005\ \Omega, Z_{L2} = 0.2842 + j0.005\ \Omega\) |
| Charging | SOC0-1 = 40%, SOC0-2 = 30% | \(Z_{L1} = 0.2842 + j0.005\ \Omega, Z_{L2} = 0.0642 + j0.005\ \Omega\) |
Simulation Results and Analysis
To validate the effectiveness of the proposed dynamic droop control method for energy storage battery converter clusters, I built a simulation model in MATLAB/Simulink using two lithium energy storage battery converters as examples. Lithium energy storage battery converters are characterized by high energy conversion efficiency and fast response speed, making them core equipment in energy storage battery systems. They are widely used in various applications: facilitating renewable energy integration on the generation side, providing key auxiliary services on the grid side, achieving peak-valley arbitrage and improving power supply reliability on the user side, and supporting the stable operation of microgrids. As a key technology for building new power systems and enhancing grid flexibility and resilience, I conducted research on the cluster coordinated control of energy storage battery systems. I designed three different operating conditions to verify the effectiveness of the proposed method.
Traditional Droop Control Simulation Results
To more clearly demonstrate the balancing effect of the proposed SOC balancing control method, I conducted simulation analysis under the three operating conditions using the traditional droop control strategy. The simulation results show that under traditional droop control, none of the three operating conditions can achieve SOC balancing. When the initial SOC values differ, the traditional method maintains the initial SOC difference state. Even when the initial SOC values are identical, the system cannot maintain balance if the line impedances are mismatched. This confirms the inherent limitation of traditional droop control in dealing with SOC imbalance issues in energy storage battery systems.
Segmented Droop SOC Balancing Control Strategy Simulation Results
To verify the correctness of the proposed SOC segmented balancing control strategy, I conducted simulation validation under the three operating conditions and compared the results with those obtained under traditional droop control. The simulation results under Condition 1 demonstrate the effectiveness of the proposed strategy.
Under Condition 1, in the discharging state, the initial SOC of system 1 is 80% and that of system 2 is 70%. Since the SOC deviation \(\Delta SOC_i\) is greater than the threshold \(T_{SOC} = 0.2\%\), the system operates in the fast balancing region. During this stage, the system fixes the droop coefficient at a relatively large value, causing system 1, with higher SOC, to output larger power and its SOC to decrease at a significantly faster rate from 80%. Meanwhile, the SOC of system 2 decreases slowly from 70%. The SOC difference between the two systems rapidly diminishes. When \(\Delta SOC_i\) drops below \(T_{SOC}\), the system enters the precise adjustment region. At this point, the droop coefficient is adjusted in real-time according to the SOC changes, gradually decreasing and approaching the initial droop coefficient. Simultaneously, the output powers of the two energy storage converters gradually converge, and the SOC change rates also approach each other, achieving precise balancing. In the charging state under Condition 1, the initial SOC of system 1 is 40% and that of system 2 is 30%. Since \(\Delta SOC_i\) is greater than \(T_{SOC}\), the system operates in the fast balancing region throughout the entire simulation process. During this stage, system 2 with lower SOC accelerates charging, and its SOC rises at a significantly faster rate. The simulation results show that after 300 seconds, the SOC deviation between the two systems decreases from the initial 10% to 5.698%, indicating the effectiveness of the system in the fast balancing region.
Under Condition 2, the initial SOC values of the two systems are identical (80% for discharging or 30% for charging). Since \(\Delta SOC_i \leq T_{SOC}\), the system operates in the precise adjustment region throughout the entire simulation process. During this stage, the droop coefficient is adjusted in real-time according to SOC changes, making it equal to the initial droop coefficient, ensuring consistent output power. The SOC of both systems remains consistent throughout the process, and the output powers also tend to be consistent, demonstrating that the system can maintain the balancing state when the initial SOC values are the same.
Under Condition 3, in the discharging state, the initial SOC of system 1 is 80% and that of system 2 is 70%. Since \(\Delta SOC_i > T_{SOC}\), the system enters the fast balancing region. During this stage, system 1 with higher SOC outputs larger power, and its SOC decreases at a significantly faster rate from 80%, while the SOC of system 2 decreases more slowly. When \(\Delta SOC_i \leq T_{SOC}\), the system enters the precise adjustment region. At this point, the droop coefficient is adjusted in real-time, gradually decreasing and approaching the initial droop coefficient. The output powers of the two energy storage converters gradually converge, and the SOC change rates also approach each other, eventually tending to consistency and achieving precise balancing. In the charging state under Condition 3, the initial SOC of system 1 is 40% and that of system 2 is 30%. Since \(\Delta SOC_i > T_{SOC}\), the system operates in the fast balancing region throughout the simulation process. During this stage, system 2 with lower SOC accelerates charging, and its SOC rises at a significantly faster rate. After 300 seconds, the SOC deviation decreases from the initial 10% to 6.132%, confirming the effectiveness of the system in the fast balancing region.
| Operating Condition | State | Initial SOC Difference | Final SOC Difference (300s) | Balancing Region |
|---|---|---|---|---|
| Condition 1 | Discharging | 10% | ≤ 0.2% (converged) | Fast + Precise |
| Condition 1 | Charging | 10% | 5.698% (converging) | Fast |
| Condition 2 | Discharging | 0% | 0% (maintained) | Precise |
| Condition 2 | Charging | 0% | 0% (maintained) | Precise |
| Condition 3 | Discharging | 10% | ≤ 0.2% (converged) | Fast + Precise |
| Condition 3 | Charging | 10% | 6.132% (converging) | Fast |
| Operating Condition | State | Power Distribution Characteristic | Current Output Characteristic |
|---|---|---|---|
| Condition 1 | Discharging | Initially P1 > P2, gradually converging to equal | Initially Io1 > Io2, gradually converging |
| Condition 1 | Charging | Initially |P1| < |P2|, gradually converging | Initially |Io1| < |Io2|, gradually converging |
| Condition 2 | Discharging | P1 = P2 maintained throughout | Io1 = Io2 maintained |
| Condition 2 | Charging | |P1| = |P2| maintained throughout | |Io1| = |Io2| maintained |
| Condition 3 | Discharging | Initially P1 > P2, gradually converging | Initially Io1 > Io2, gradually converging |
| Condition 3 | Charging | Initially |P1| < |P2|, gradually converging | Initially |Io1| < |Io2|, gradually converging |
Discussion and Implications
The simulation results across all three operating conditions consistently demonstrate the superior performance of the proposed segmented dynamic droop control strategy for energy storage battery converter clusters. Compared with traditional droop control, which fails to achieve SOC balancing under any of the tested conditions, the proposed method successfully achieves SOC convergence in all scenarios. This represents a significant improvement in the coordinated control capability of energy storage battery systems.
Under Condition 1, where the initial SOC values differ but the line impedances are identical, the proposed method reduces the SOC deviation from 10% to within 0.2% in the discharging state and to 5.698% in the charging state within 300 seconds. The faster convergence in the discharging state can be attributed to the contribution of both the fast balancing and precise adjustment regions, while the charging state primarily relies on the fast balancing region within the simulation timeframe. This indicates that the proposed method can effectively handle the most common scenario in practical energy storage battery systems: SOC imbalance caused by different initial states.
Under Condition 2, where the initial SOC values are identical but the line impedances are mismatched, the proposed method maintains the SOC balance throughout the operation. This is particularly important for practical applications where line impedance variations are inevitable due to differences in cable lengths, connector resistances, and other factors. The ability to maintain balance under impedance mismatch conditions enhances the robustness and reliability of the energy storage battery system.
Under Condition 3, which combines both initial SOC differences and line impedance mismatch, the proposed method again successfully achieves SOC convergence. The discharging state reduces the SOC deviation from 10% to within 0.2%, while the charging state reduces it to 6.132% within 300 seconds. This demonstrates the robustness of the proposed method under the most challenging operating conditions, where multiple factors contributing to SOC imbalance coexist.
The successful implementation of the proposed method can be attributed to the two key innovations: the introduction of the SOC deviation threshold to divide the balancing process into two distinct regions, and the dynamic adjustment of the droop coefficient based on real-time SOC information. The fast balancing region enables rapid reduction of large SOC deviations, while the precise adjustment region ensures accurate convergence to the balanced state. This two-stage approach effectively addresses the conflicting requirements of balancing speed and accuracy, which is a common challenge in existing SOC balancing methods.
From a practical perspective, the proposed method offers several advantages for real-world energy storage battery systems. First, it eliminates the need for complex communication networks, relying only on local SOC measurements for control decisions. This reduces system complexity and cost while improving reliability. Second, the method is compatible with existing energy storage battery system architectures, requiring only software modifications to the converter control algorithm. Third, the method is scalable to systems with any number of parallel energy storage battery units, making it suitable for large-scale energy storage applications.
The parameter design guidelines derived from the analysis provide practical rules for implementing the proposed method in real systems. The selection of the initial droop coefficient \(m_0\) ensures that the grid voltage deviation remains within acceptable limits. The choice of the adjustment factor \(k_c\) in the fast balancing region and the PI controller parameters \(k_{mp}\) and \(k_{mi}\) in the precise adjustment region can be optimized based on the specific requirements of the application, such as the desired balancing speed and the power rating of the converters.
| Performance Metric | Traditional Droop Control | Proposed Method |
|---|---|---|
| SOC balancing under initial SOC difference | Not achieved | Achieved (deviation reduced to ≤0.2%) |
| SOC balancing under line impedance mismatch | Not achieved | Maintained (deviation = 0%) |
| SOC balancing under combined conditions | Not achieved | Achieved (deviation reduced to ≤0.2%) |
| Balancing speed | Slow or non-convergent | Fast (within 300s for 10% initial deviation) |
| Control complexity | Low (fixed coefficient) | Moderate (adaptive adjustment) |
| Communication requirement | None | None (local measurement only) |
| Grid voltage deviation | Within acceptable range | Within acceptable range |
| Adaptability to system changes | Low (fixed parameters) | High (real-time adaptation) |
Conclusion and Future Work
In this paper, I have addressed the critical challenge of SOC imbalance in energy storage battery converter clusters operating in parallel. The problem of SOC imbalance, arising from differences in initial SOC values and mismatched line impedances, can lead to reduced capacity utilization, overcharging or over-discharging of individual energy storage battery units, and compromised system reliability and battery lifespan. To solve this problem, I proposed a dynamic droop control strategy that effectively balances the SOC among multiple parallel energy storage battery systems.
The key contributions and findings of this research can be summarized as follows. First, I developed a dynamic droop control strategy that adjusts the droop coefficient in real-time based on the SOC deviation of each energy storage battery unit. This approach overcomes the fundamental limitation of traditional droop control, where the fixed droop coefficient cannot adapt to changing SOC states, resulting in persistent SOC imbalance. By continuously modifying the droop coefficient according to the SOC information, the proposed method enables the energy storage battery system to dynamically redistribute power among parallel units, actively driving the SOC values towards convergence.
Second, I introduced an SOC deviation threshold mechanism that divides the balancing process into two distinct operating regions: the fast balancing region and the precise adjustment region. In the fast balancing region, activated when the SOC deviation exceeds the threshold, a fixed droop coefficient is used to achieve rapid SOC convergence. This is particularly effective when large SOC differences exist, as it allows the energy storage battery units with higher SOC to discharge more power or units with lower SOC to charge more power, quickly reducing the imbalance. In the precise adjustment region, activated when the SOC deviation falls below the threshold, a PI controller dynamically adjusts the droop coefficient to gradually eliminate the remaining small SOC deviation, ensuring precise balancing. This two-stage approach strikes an optimal balance between balancing speed and accuracy, addressing a common trade-off in existing SOC balancing methods.
Third, I conducted a comprehensive parameter design analysis to guide the practical implementation of the proposed method. I derived the relationships between the key parameters, including the initial droop coefficient \(m_0\), the adjustment factor \(k_c\), the PI controller parameters \(k_{mp}\) and \(k_{mi}\), and the system performance metrics such as power output limits and voltage deviation constraints. The analysis showed that in the fast balancing region, \(k_c\) should be selected as large as possible without exceeding the power limit to maximize balancing speed. In the precise adjustment region, \(k_{mp}\) should match the peak output power of the converter before the threshold, and \(k_{mi}\) should be chosen to ensure system stability. These design rules provide practical guidance for engineers implementing the proposed method in real energy storage battery systems.
Fourth, I validated the effectiveness of the proposed method through extensive simulations under three different operating conditions. The results demonstrated that the proposed method consistently achieves SOC balancing, reducing the SOC deviation from 10% to within 0.2% in the discharging state under Conditions 1 and 3. In the charging state, the SOC deviation was reduced from 10% to 5.698% and 6.132% under Conditions 1 and 3, respectively, within 300 seconds. Under Condition 2, where the initial SOC values were identical, the method maintained the balanced state throughout the operation, even with mismatched line impedances. These results confirm the robustness and adaptability of the proposed method across a wide range of operating scenarios.
Fifth, the proposed method offers significant practical advantages for real-world energy storage battery systems. It operates based on local SOC measurements, eliminating the need for complex communication networks and reducing system cost and complexity. It is compatible with existing energy storage battery system architectures, requiring only software updates to the converter control algorithm. The method is scalable to systems with any number of parallel energy storage battery units, making it suitable for a wide range of applications, from small-scale residential storage to large-scale utility-level energy storage systems.
Despite the promising results, this research has some limitations that warrant further investigation. First, the simulation study was conducted under idealized conditions with a simplified energy storage battery model. Real energy storage battery systems exhibit more complex behaviors, including nonlinear internal resistance, capacity degradation over cycles, and temperature-dependent characteristics. Future work should incorporate more accurate energy storage battery models to validate the proposed method under more realistic conditions.
Second, while the proposed method effectively balances SOC among parallel energy storage battery units, it does not explicitly consider other important factors such as battery aging, thermal management, and state of health (SOH). In practical applications, the SOH of individual energy storage battery units may differ significantly, affecting their usable capacity and internal resistance. Future research should explore the integration of SOH information into the control strategy to achieve more comprehensive battery management.
Third, the current study focuses on the steady-state performance of the proposed method. The transient response during load changes, grid disturbances, and system startup or shutdown conditions should be investigated in future work. The interaction between the proposed SOC balancing control and other control loops in the system, such as voltage regulation and current limiting, should be analyzed to ensure overall system stability.
Fourth, experimental validation using hardware-in-the-loop or physical prototype testing would provide stronger evidence for the practical feasibility of the proposed method. The computational requirements and real-time implementation challenges should be assessed to ensure that the method can be executed within the control cycle of typical energy storage converters.
In conclusion, the proposed dynamic droop control method represents a significant advancement in SOC balancing technology for energy storage battery converter clusters. By effectively addressing the limitations of traditional droop control and existing balancing methods, it provides a practical and effective solution for enhancing the performance, reliability, and lifespan of energy storage battery systems in modern power grids. As the penetration of renewable energy and the deployment of energy storage battery systems continue to grow, the importance of effective SOC balancing will only increase. The proposed method, with its balance of simplicity, effectiveness, and adaptability, is well-positioned to meet this growing need and contribute to the development of more resilient and efficient power systems.