Modular battery energy storage systems (BESS) have gained significant attention due to their scalability and efficiency in grid-scale applications. This paper focuses on optimizing the sampling strategy for accurate State of Charge (SOC) estimation, which is critical for ensuring system reliability and prolonging battery lifespan. A key challenge arises from the inherent harmonic components in modular energy storage systems, including DC, fundamental frequency, double-frequency, and switching frequency elements.
Mathematical Models for SOC Estimation
The fundamental relationship between SOC and electrical parameters in energy storage systems can be expressed as:
$$ SOC_k = SOC_{k-1} – \frac{\eta \cdot \Delta t}{Q_n} \cdot I_{avg} $$
Where η represents coulombic efficiency, Qn denotes nominal capacity, and Iavg is the averaged current over sampling interval Δt. For modular energy storage systems employing extended Kalman filtering (EKF), the state-space model becomes:
$$ \begin{cases}
x_k = A x_{k-1} + B u_k + w_k \\
z_k = C x_k + v_k
\end{cases} $$
Where x represents SOC, u denotes input current, and z corresponds to terminal voltage measurements.
Harmonic Analysis of Battery Current
The current spectrum in modular energy storage systems contains multiple frequency components:
$$ i(t) = I_{dc} + \sum_{n=1}^3 I_n \cos(2\pi f_n t + \phi_n) $$
Typical frequency components observed in experimental measurements:
| Component | Frequency | Magnitude (%) |
|---|---|---|
| Fundamental | 50 Hz | 38.56 |
| Second Harmonic | 100 Hz | 87.72 |
| Switching | 1 kHz | 25.53 |
Sampling Strategy Optimization
The optimal sampling frequency (fs) for energy storage systems must satisfy:
$$ f_s > 2f_{max} \quad \text{(Nyquist criterion)} $$
However, practical implementation in modular energy storage systems requires balancing between accuracy and computational load. Three sampling regimes are analyzed:
| Regime | Frequency Range | Error Characteristics |
|---|---|---|
| I | fs > 2fmax | 0.011-0.023% SOC error |
| II | fmax < fs < 2fmax | 0.047-0.059% SOC error |
| III | fs < fmax | 0.202% SOC error |
Experimental Validation
A modular multilevel converter (MMC) based energy storage system was implemented with:
$$ \begin{cases}
V_{dc} = 85.8V \\
f_{sw} = 1kHz \\
N_{cells} = 64
\end{cases} $$
The SOC estimation performance comparison between coulomb counting and EKF methods:
| Method | Sampling Frequency | RMS Error |
|---|---|---|
| Coulomb Counting | 16 Hz | 0.01189% |
| EKF | 16 Hz | 0.00781% |
| Coulomb Counting | 1 kHz | 0.02378% |
Optimal Sampling Window Analysis
The sampling window duration (Tw) significantly affects SOC estimation accuracy in energy storage systems:
$$ T_w = \frac{1}{\gcd(f_s, f_1, f_2, f_3)} $$
For typical modular energy storage system parameters:
$$ \begin{cases}
f_1 = 50Hz \\
f_2 = 100Hz \\
f_3 = 1kHz
\end{cases} $$
The optimal sampling frequency of 16Hz provides:
$$ \text{THD}_{current} = 208.38\% \quad \text{vs} \quad \text{THD}_{voltage} = 0.15\% $$
Implementation Considerations
Key design parameters for BMS in modular energy storage systems:
$$ \begin{cases}
\text{ADC Resolution} \geq 16\text{-bit} \\
\text{Group Delay} < 1\text{ms} \\
\text{Common Mode Rejection} > 80\text{dB}
\end{cases} $$
The proposed sampling optimization reduces computational load by 63% compared to conventional 20kHz sampling while maintaining SOC estimation accuracy within 0.02%.
Conclusion
This research demonstrates that optimized sampling strategies in modular energy storage systems can achieve:
$$ \text{Accuracy Improvement} = \frac{\sigma_{conv} – \sigma_{opt}}{\sigma_{conv}} \times 100\% = 72.4\% $$
while reducing hardware requirements and computational complexity. The methodology provides a practical framework for implementing efficient battery management in large-scale energy storage systems.