Throughout my master’s research, I have focused on the operational challenges of battery energy storage systems (BESS) when they participate in microgrid power regulation. The core issue is that BESS units are often modeled as single large cells, while the internal inconsistencies among individual cells inside each energy storage unit are ignored. This simplification severely reduces the actual usable capacity of the storage unit and weakens the overall regulation capability of the BESS. My work aims to reveal how lithium-ion battery characteristics influence internal inconsistency and to propose power control strategies that integrate cell balancing management with system-level power distribution. In this article, I present my research findings in a structured manner, emphasizing the role of the energy storage battery state estimation, equalization techniques, and adaptive optimization algorithms.
1. Working Characteristics and State Estimation of Lithium Iron Phosphate Batteries
The precise knowledge of battery operating characteristics and state parameters is essential to guarantee the reliable and safe operation of an energy storage system. Among all parameters, the state of charge (SOC) and state of health (SOH) of the energy storage battery are the most critical for evaluating the working state and regulating capability of a BESS. In my research, I selected lithium iron phosphate (LiFePO4) batteries as the study object because they are the most promising technology for large-scale stationary storage applications.
1.1 Working Principle and Equivalent Circuit Model
The positive electrode material of a LiFePO4 battery is lithium iron phosphate, while the negative electrode is graphite. During charging and discharging, lithium ions shuttle between the two electrodes through a polymer separator. The separator permits only positive lithium ions to pass while blocking electrons. The charging reaction can be expressed as:
$$
\mathrm{LiFePO_4 – yLi^+ – ye^- \rightarrow yFePO_4 + (1-y)LiFePO_4}
$$
during the positive electrode reaction, and
$$
\mathrm{yC_6 + yLi^+ + ye^- \rightarrow yLiC_6}
$$
for the negative electrode. The total reaction is:
$$
\mathrm{6yC + LiFePO_4 \rightleftharpoons Li_yC_6 + Li_{1-y}FePO_4}
$$
To simulate the dynamic behavior of the energy storage battery, I compared several equivalent circuit models including the Rint model, the PNGV model, and the Thevenin model. After careful consideration, I adopted the second-order Thevenin model due to its balance between accuracy and simplicity. The model is depicted by a voltage source representing the open-circuit voltage (OCV), an ohmic internal resistance R0, and two parallel resistor-capacitor (RC) networks representing the activation polarization and concentration polarization effects. The state space equations are given as:
$$
\begin{bmatrix} \dot{U}_1 \\ \dot{U}_2 \end{bmatrix} =
\begin{bmatrix} -\frac{1}{R_1 C_1} & 0 \\ 0 & -\frac{1}{R_2 C_2} \end{bmatrix}
\begin{bmatrix} U_1 \\ U_2 \end{bmatrix} +
\begin{bmatrix} \frac{1}{C_1} \\ \frac{1}{C_2} \end{bmatrix} I(t)
$$
$$
U_t(t) = U_{OC}(SOC) – I(t)R_0 – U_1(t) – U_2(t)
$$
This model accurately reflects the terminal voltage response under various load conditions and provides a foundation for SOC and SOH estimation.
1.2 SOH Estimation Based on Weighted Discharge Accumulation
SOH represents the attenuation degree of battery parameters compared with the factory state. I have considered multiple definitions based on capacity, internal resistance, and cycle number. The capacity-based definition is the most widely used:
$$
SOH = \frac{E_{now}}{E_N}
$$
where E_now is the current maximum available capacity and E_N is the rated capacity. When SOH drops below 0.8, the energy storage battery is usually retired.
For SOH estimation, I employed a model-free method based on the weighted accumulation of discharge quantities. The key is to account for the random depth-of-discharge (DOD) cycles experienced by batteries in a BESS when smoothing renewable power fluctuations. I used the rain-flow counting method to extract the number of cycles and their corresponding DOD values. The relationship between cycle life and DOD for a 60 Ah LiFePO4 battery is fitted by the following equation:
$$
Cnt_A = 170188 – 264167D_A + 158680D_A^2 – 4.23 \times 10^8 D_A^3 + 22569 e^{D_A} + 46733
$$
For a given DOD D_A, the total discharge capacity over its cycle life is:
$$
C_{total-A} = D_A \cdot C_N \cdot Cnt_A
$$
where C_N is the rated capacity. The ratio K_A between the maximum total discharge capacity and that at DOD D_A is defined as:
$$
K_A = \frac{C_{total-max}}{C_{total-A}}
$$
Thus, the discharge quantity at any DOD can be converted to the equivalent discharge quantity at the maximum DOD. The SOH is then obtained by:
$$
SOH = \frac{\sum C_{dis-A} \cdot K_A}{C_{total-max}}
$$
This method avoids dependence on model accuracy and improves the estimation robustness, which is essential for long-term monitoring of the energy storage battery in a BESS.
1.3 SOC Estimation Combining Open-Circuit Voltage and Ampere-Hour Integration
SOC is defined as the ratio of the remaining capacity to the maximum available capacity under the same conditions:
$$
SOC = \frac{E_{now}}{E_{max}}
$$
Both the open-circuit voltage (OCV) method and ampere-hour integration method have limitations. The OCV method is accurate at high and low SOC ranges, but very inaccurate in the plateau region where the OCV-SOC curve is almost flat. The ampere-hour integration method is simple and real-time, but it suffers from accumulated current measurement errors and requires an accurate initial SOC value.
To overcome these drawbacks, I proposed a hybrid SOC estimation method. When the energy storage battery is operating in the high or low SOC region, the OCV method is used to recalibrate the initial SOC value for the ampere-hour integration. In the middle SOC range, the ampere-hour integration method is used to avoid the OCV plateau inaccuracy. The SOC update formula is:
$$
S(t) = S_0 + \frac{1}{E_N} \int_0^t \eta \, i(\tau) \, d\tau
$$
where S_0 is the corrected initial SOC, eta is the Coulombic efficiency, and i(tau) is the current. This hybrid approach improves the accuracy and reliability of SOC determination, which is a prerequisite for internal inconsistency analysis.
2. Inconsistency Analysis of Energy Storage Batteries and Suppression Methods
In practice, a large-scale BESS is composed of many energy storage battery cells connected in series and parallel to achieve high voltage and large capacity. Even cells from the same production batch exhibit inherent differences due to manufacturing variations such as coating thickness, impurity content, and electrode porosity. These small differences accumulate during operation, leading to severe inconsistency among cells inside a storage unit. I have systematically analyzed the causes, manifestations, and types of inconsistency.
2.1 Root Causes of Inconsistency
The main causes of inconsistency can be categorized as follows:
- Temperature differences: Internal resistance variations cause uneven heat generation. Cells in different locations also experience different heat dissipation conditions.
- Depth-of-discharge differences: A weaker cell will reach its cut-off voltage earlier than other cells, causing it to experience a deeper DOD and thus faster degradation.
- SOC cycling range differences: Cells that are forced to operate at high or low SOC windows due to initial capacity differences suffer more severe degradation. The following table summarizes the capacity fade rates for different SOC cycling windows:
| Design Scheme | Initial SOC (%) | Terminal SOC (%) | Capacity Fade Rate (%) | Equivalent Fade Rate (%) |
|---|---|---|---|---|
| Scheme 1 | 100 | 80 | 2.0 | 8.0 |
| Scheme 2 | 60 | 40 | 0.6 | 2.4 |
| Scheme 3 | 20 | 0 | 1.9 | 7.6 |
| Scheme 4 | 100 | 20 | 7.0 | 7.0 |
| Scheme 5 | 90 | 10 | 4.9 | 4.9 |
| Scheme 6 | 80 | 0 | 6.1 | 6.1 |
The table clearly demonstrates that cycling near 50% SOC is beneficial for extending the lifetime of the energy storage battery.
2.2 Manifestations and Types of Inconsistency
Inconsistency manifests in two major categories:
(1) SOH inconsistency: This represents the difference in health status among cells. A long-term cycling test on 48 identical cells showed that after 1000 cycles, the worst cell had a SOH 25% lower than the best cell. SOH inconsistency appears as resistance dispersion and capacity dispersion.
(2) Voltage/SOC inconsistency: This directly reflects the operating state divergence. In a test on 24 series-connected LiFePO4 cells, the voltage dispersion exhibited a clear “broom” effect after 300 cycles, with some cells deviating far from the average voltage.
From the system perspective, I classified the imbalance states into three types: high-end imbalance, low-end imbalance, and combined high-low imbalance. In a high-end imbalance, the weakest cell has too high a SOC, limiting the charging capacity of the entire unit. In low-end imbalance, the weakest cell has too low a SOC, limiting the discharging capacity. The combined state limits both charging and discharging capabilities, directly reducing the usable capacity of the energy storage battery unit.
2.3 Active Equalization Using an Auxiliary Power Architecture
To suppress the internal inconsistency, I adopted an active balancing architecture based on an auxiliary power source, referred to as Cell-to-Auxiliary-Power (C2AP). In this architecture, a main battery management unit (BMU) communicates with slave BMUs via CAN bus. Each slave BMU controls a switch array and a bidirectional flyback converter to transfer energy between the selected cell and the auxiliary power source. The key advantage of this architecture is that only the cell being balanced is involved in the equalization process, which speeds up balancing and improves efficiency.
The switch array consists of polarity switching switches and cell selection switches. For a battery string containing N cells, the number of cell-selection switches is N+1 because adjacent cells share one switch. The following table gives the switch states for selecting different cells in a 6-cell module:
| Selected Cell | PS1 | PS2 | PS3 | PS4 | CS0 | CS1 | CS2 | CS3 | CS4 | CS5 | CS6 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| C1 | 1 | 0 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 |
| C2 | 0 | 1 | 1 | 0 | 0 | 1 | 1 | 0 | 0 | 0 | 0 |
| C3 | 1 | 0 | 0 | 1 | 0 | 0 | 1 | 1 | 0 | 0 | 0 |
| C4 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
| C5 | 1 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 0 |
| C6 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 1 |
The bidirectional flyback converter is used for energy transfer. When energy flows from the battery side to the auxiliary side, the primary-side MOSFET Q1 is turned on to store energy in the transformer core, then Q1 is turned off and Q2 is turned on to release the stored energy to the auxiliary side. The process is reversed for energy transfer in the opposite direction.
2.4 Small-Period Equalization Control Strategy
Traditional equalization strategies select the cell with the largest SOC deviation from the average and balance it until the deviation falls below a threshold. This approach suffers from low efficiency when a single equalizer is used, because cells whose SOC deviation becomes severe during the balancing process must wait until the current balancing task finishes. To overcome this problem, I proposed a small-period equalization strategy. The entire balancing process is divided into multiple short intervals of length T. At the end of each interval, the system re-evaluates the SOC of all cells and selects the cell with the largest SOC deviation for balancing in the next interval.
The SOC deviation of the i-th cell is defined as:
$$
\Delta SOC_i = SOC_i – \frac{1}{n} \sum_{j=1}^{n} SOC_j
$$
The balancing process is initiated when the SOC range exceeds a threshold α:
$$
r = SOC_{max} – SOC_{min} > \alpha
$$
and stops when all deviations satisfy:
$$
|\Delta SOC_i| \leq \beta
$$
where β is the stopping threshold. I also used the SOC standard deviation as an indicator of balance quality:
$$
\varepsilon = \sqrt{\frac{\sum_{i=1}^{n} (SOC_i – \overline{SOC})^2}{n-1}}
$$
I simulated a 6-cell lithium-ion battery string with initial SOC values of 51%, 52.7%, 48.9%, 47.4%, 50%, and 50%. The rated capacity was 5 Ah and the nominal voltage was 2.2 V. The equalization thresholds were α=0.3% and β=0.03%, with a small period T=1 s. The simulation was performed in MATLAB/Simulink under a 0.2C charging condition.
The results showed that my proposed strategy completed the balancing in 1966.87 seconds, while the traditional strategy required 2329.66 seconds. After balancing, the SOC range was reduced to about 0.06% and the standard deviation to about 0.027%. The traditional strategy caused over-equalization and unnecessary balancing actions on some cells, whereas my strategy only balanced the four cells that actually deviated from the average, thus reducing stress on the energy storage battery and improving its lifetime.
3. BESS Power Distribution Strategy Considering Internal State Consistency
When a BESS is used to smooth power fluctuations in a microgrid, the irregular charging and discharging actions lead to SOH and SOC inconsistency among different energy storage units. I studied the coupling relationship between SOH difference and SOC consistency inside each unit. Based on this, I developed a power allocation strategy that combines a priority ranking function with an adaptive mutation particle swarm optimization (AMPSO) algorithm. The objective is to minimize the degradation of the weakest energy storage battery cells and to improve the overall efficiency and regulating capability of the BESS.
3.1 Coupling Mechanism between SOH and SOC Inconsistency
An initial inferior cell has lower capacity and higher internal resistance. During repeated charge/discharge cycles, this cell will become polarized more quickly, causing its SOC to deviate from the average. The larger SOC deviation leads to a deeper DOD for that cell, which accelerates its SOH decay. As SOH decays, the capacity decreases further, making the SOC deviation even worse. This creates a vicious circle, as illustrated conceptually in the following relationship:
From a control point of view, I can express the update of the cell SOC difference inside unit i as:
$$
\Delta S_{OC,i}^{P}(t) =
\begin{cases}
\Delta S_{OC,i}(t-1) + \frac{P_{B,i}(t) \cdot \eta_{P,i} \cdot \Delta t}{E_{N,i} \cdot S_{OH,i}} & , P_{B,i}(t) \geq 0 \\
\Delta S_{OC,i}(t-1) + \frac{P_{B,i}(t) \cdot \Delta t}{E_{N,i} \cdot S_{OH,i}} & , P_{B,i}(t) < 0
\end{cases}
$$
where P_{B,i} is the power of unit i, E_{N,i} is its rated capacity, and S_{OH,i} is the minimum SOH inside the unit.
3.2 System Structure and Power Balancing
I considered a common DC bus centralized microgrid system containing wind power, photovoltaic generation, conventional loads, electric vehicle charging stations, and a BESS composed of multiple storage units. The power flow is normalized with the direction from the sources through the DC bus to the loads as positive. The unbalanced power at the point of common coupling is:
$$
P(t) = P_W(t) + P_P(t) – P_L(t) – P_E(t)
$$
where P_W(t) is wind power, P_P(t) is photovoltaic power, P_L(t) is conventional load power, and P_E(t) is EV charging power. The BESS must absorb or release power to keep the grid-connected power within the required fluctuation range. The grid connection standard requires that the 30-minute power fluctuation rate does not exceed 7% of the total installed renewable capacity. I used a moving average method to determine the target grid-connected power P*(t):
$$
P^*(t) = \frac{1}{N} \sum_{\tau = t – \frac{N-1}{2}}^{t + \frac{N-1}{2}} P(\tau)
$$
where N is an odd sliding coefficient. The required BESS power is:
$$
P_B(t) = P(t) – P^*(t)
$$
To determine the appropriate BESS capacity, I used a probability-based statistics method. The rated power and capacity of the BESS are chosen so that the power fluctuation is suppressed with a given confidence level β:
$$
P_{BN} = F^{-1}(\beta)
$$
$$
C_{BN} = a \cdot P_{BN}
$$
where a is a coefficient typically between 4 and 6. In my case study, the maximum fluctuation amplitude was 16.37 kW, and only 1.03% of the fluctuations exceeded 16 kW. Therefore, I configured the BESS with a rated power of 16 kW and a rated energy of 64 kWh, divided into four identical storage units each rated at 4 kW and 16 kWh.
3.3 Charging/Discharging Priority Ranking Function
To determine which storage units should participate in each dispatch cycle, I defined a priority function F1 that combines two sub-functions: f1 considers the SOC of the energy storage battery unit, and f2 considers the internal SOH and SOC inconsistency.
The first sub-function f1 is based on the SOC level. It is designed to give higher priority to units whose SOC is close to 0.5, where the energy storage battery has the best cycling capability and minimal lifetime degradation. I used the inverse hyperbolic tangent function to penalize extreme SOC values:
$$
f_1 =
\begin{cases}
0.15 \ln\left( \frac{S_{max} + S_{min} – 2(S(t) – S)}{S_{max} + S_{min} – 2S(t)} \right) & , P_B < 0 \\
0.15 \ln\left( \frac{S_{max} – S_{min} – 2(S(t) – S)}{S_{max} – S_{min} – 2(S(t) – S)} \right) & , P_B \geq 0
\end{cases}
$$
where S(t) is the current SOC of the unit.
The second sub-function f2 is derived from the SOH level, SOH range inside the unit, and the SOC deviation among cells. It is defined as:
$$
f_2 =
\begin{cases}
\frac{S_{OH,i}}{\Delta S_{OH,i}} \cdot z & , \Delta S_{OC,i} \cdot P_B < 0, \Delta S_{OC,i} > z \\
\frac{S_{OH,i}}{\Delta S_{OH,i}} & , \Delta S_{OC,i} \cdot P_B \geq 0 \\
0 & , \Delta S_{OC,i} \cdot P_B < 0, \Delta S_{OC,i} \leq z
\end{cases}
$$
where S_{OH,i} is the lowest SOH in unit i, ΔS_{OH,i} is the SOH range inside unit i, ΔS_{OC,i} is the SOC difference between the worst cell and the best cell, and z is a preset imbalance threshold. This function ensures that units with low SOH and large internal dispersion are not selected for operation, especially when the operation would further increase the internal SOC inconsistency.
Combining both sub-functions with weight coefficients b and c:
$$
F_1 = b \cdot f_1 + c \cdot f_2
$$
The larger the value of F1, the lower the priority of that unit for participating in the power dispatch.
3.4 Power Distribution Optimization Using AMPSO
After selecting the group of storage units to participate, the power is distributed among them using an optimization algorithm. The objective is to minimize the future internal SOC inconsistency of the selected units and to maximize their efficiency. I formulated the objective function F2 as:
$$
F_2 = \sum_{i=1}^{n} d_i \cdot
\begin{cases}
1 + \frac{\Delta S_{OC,i}^{P,max}(t)}{\Delta S_{OC,i}^{max}} \cdot 0.15 \ln\left( 1 + \frac{\Delta S_{OC,i}^{P}(t)}{\Delta S_{OC,i}^{P,max}} \right) & \text{if charged} \\
1 + \frac{\Delta S_{OC,i}^{P,max}(t)}{\Delta S_{OC,i}^{max}} \cdot 0.15 \ln\left( 1 – \frac{\Delta S_{OC,i}^{P}(t)}{\Delta S_{OC,i}^{P,max}} \right) & \text{if discharged}
\end{cases}
$$
where d_i is 1 if unit i is selected and 0 otherwise, and ΔS_{OC,i}^{P,max} is the maximum predicted internal SOC deviation for unit i within the dispatch horizon.
The AMPSO algorithm is an improved version of the standard particle swarm optimization (PSO) that incorporates adaptive mutation inspired by genetic algorithms. In each iteration, some particles are re-initialized with a certain probability to maintain population diversity and to avoid premature convergence. The algorithm parameters are set as follows: population size = 50, maximum iterations = 200, inertia weight = 0.8, cognitive acceleration factor = 0.5, social acceleration factor = 0.5, velocity range = [-0.2, 0.2], and mutation probability = 0.15. The iteration curve shows that the algorithm converges at iteration 164, and the computation time for 200 iterations is less than 1 second, which satisfies the requirement for real-time dispatch.
3.5 Case Study and Simulation Results
I used the measured data of a common DC bus microgrid demonstration platform, including wind power, photovoltaic power, conventional load, and EV charging load, with a 24-hour operation horizon divided into 96 sampling points of 15 minutes each. The rated wind capacity was 60 kW, and the rated photovoltaic capacity was 35 kW. The BESS parameters are listed in the following table:
| Unit No. | SOH of weakest cell | SOH of best cell | Smax | Smin | Initial SOC | ΔSOC,max | z |
|---|---|---|---|---|---|---|---|
| 1 | 95% | 97% | 0.8 | 0.2 | 0.5 | 3.2% | 0.2% |
| 2 | 93% | 96% | 0.8 | 0.2 | 0.5 | 3.2% | 0.2% |
| 3 | 89% | 95% | 0.8 | 0.2 | 0.5 | 3.2% | 0.2% |
| 4 | 84% | 94% | 0.8 | 0.2 | 0.5 | 3.2% | 0.2% |
I compared my proposed strategy with two traditional strategies: the proportional SOC-based power allocation strategy and the maximum power allocation strategy. The comparison was based on the number of operating actions, the average internal SOC inconsistency ΔS_OC, the SOC range, and the operational efficiency.
3.5.1 Number of Operating Actions
My proposed strategy significantly reduced the number of charge and discharge actions for all storage units. Specifically, the 3rd and 4th units, which have lower SOH and larger SOH dispersion, were dispatched much less frequently than in the traditional strategies. The total number of actions for all four units was reduced by approximately 59% compared with the proportional allocation strategy. This reduction slows down the SOH degradation of the weakest energy storage battery units and extends the overall BESS lifetime.
3.5.2 Internal SOC Inconsistency Reduction
I defined the average internal SOC inconsistency ΔS_OC,B as the mean value of the internal SOC deviations of all four units over the 24-hour period. The results are:
| Strategy | ΔSOC,max (%) | ΔSOC,B (%) |
|---|---|---|
| Proposed AMPSO strategy | 0.97 | 0.25 |
| Proportional SOC allocation | 1.58 | 0.37 |
| Maximum power allocation | 1.75 | 0.34 |
Compared with the proportional strategy and the maximum power strategy, my proposed control strategy reduced ΔS_OC,B by 32.59% and 26.37%, respectively. The reduction effect was observed throughout the entire 24-hour period, except in a few intervals where the strategy deliberately avoided the worse units to protect their SOH, which slightly increased the internal inconsistency in the short term but was beneficial in the long term.
3.5.3 SOC Range and Regulation Capability
To evaluate the bidirectional regulating capability of the BESS, I used the index R(t):
$$
R(t) = \frac{S(t) – \frac{S_{min} + S_{max}}{2}}{0.5(S_{max} – S_{min})}
$$
R(t) ranges from -1 to 1, with values closer to 0 indicating stronger ability to charge and discharge. The values for the three strategies were:
- Proposed strategy: R(t) in [-0.28, 0.48]
- Maximum power allocation: R(t) in [-0.42, 0.48]
- Proportional allocation: R(t) in [-0.33, 0.49]
My strategy kept the SOC of the poorer units close to 0.5, improving the ability of the BESS to handle bidirectional power fluctuations. The cycling intervals of each storage unit are shown in the following table:
| Unit No. | Proposed strategy | Proportional allocation | Maximum power allocation |
|---|---|---|---|
| Unit 1 | 69.71% – 37.11% | 63.84% – 38.70% | 65.31% – 41.15% |
| Unit 2 | 69.66% – 38.21% | 64.02% – 38.49% | 63.66% – 38.29% |
| Unit 3 | 65.37% – 42.09% | 64.40% – 38.07% | 65.40% – 40.90% |
| Unit 4 | 58.96% – 46.28% | 64.89% – 37.53% | 66.45% – 41.47% |
The cycling range of Unit 4, which has the worst SOH, was narrowed to 58.96% – 46.28% in my strategy, whereas the traditional strategies forced it to cycle from about 64% down to 37%. This clearly demonstrates that my strategy protects the weakest energy storage battery by keeping it in a favorable SOC window, thereby reducing capacity fade and prolonging service life.
3.5.4 Operational Efficiency Improvement
Operational efficiency is a key indicator for BESS performance. It is affected by both the power converter losses and the internal resistance of the energy storage battery. I modeled the efficiency of each unit as a function of its operating power:
$$
\eta_{P,i}(P_{B,i}) = \left( 1 – \frac{22.03}{1.5293 \cdot \frac{P_{B,i}}{P_{BN,i}} + 20} \cdot 0.15615 \right) \times 100\%
$$
This expression accounts for the fact that efficiency is low at very low power levels, increases with power, and then slightly decreases near the rated power due to higher internal losses. My proposed strategy avoids multiple units operating simultaneously at very low partial loads. Instead, it selects fewer units and operates them closer to their rated power. The average charge and discharge efficiency values were:
- Proposed strategy: 84.44%
- Proportional SOC allocation: 84.43%
- Maximum power allocation: 73.98%
Thus, my strategy improved the average efficiency by 10.46% compared with the maximum power allocation strategy and slightly improved compared to proportional allocation. This improvement reduces energy losses and increases the economic benefit of the BESS.
4. Summary and Outlook
In this research, I have systematically addressed the issues of internal inconsistency in lithium-ion energy storage battery systems and proposed effective control strategies to mitigate them. The main conclusions are as follows:
- The hybrid SOC estimation method based on OCV and ampere-hour integration provides accurate SOC values for the energy storage battery, which is essential for system monitoring and control.
- The small-period equalization strategy combined with the auxiliary-power active balancing architecture significantly shortens the balancing time and reduces unnecessary balancing actions, thereby improving the consistency of the energy storage battery unit.
- The proposed BESS power distribution strategy, which integrates priority ranking and AMPSO optimization, reduces the average internal SOC inconsistency by more than 26% compared with traditional strategies, improves operational efficiency by about 10%, and reduces the number of operating actions of degraded units by 59%.
- The strategy successfully protects the weakest energy storage battery cells by narrowing their SOC cycling window and keeping their SOC near the optimal 50% level, which extends the overall BESS lifetime and enhances its bidirectional regulation capability.
For future work, I plan to consider the impact of frequent switching of the equalization circuit on its own lifetime and reliability. Additionally, I will refine the control strategy to include the effects of internal resistance increase and temperature rise along with capacity decay in the SOH model, so that the power allocation can be even more precise and robust.