In the context of increasing development and utilization of renewable energy sources, microgrid power supply systems have found widespread application in settings such as industrial parks, remote areas, and border outposts. To ensure that energy storage cells operate in an optimal state, a Battery Management System (BMS) is essential for monitoring and managing their conditions, thereby enhancing energy utilization efficiency and ensuring safe and stable operation. In microgrid energy storage applications, energy storage cells often cannot be fully charged or discharged, and the system demands high energy utilization rates and stringent reliability standards. Consequently, BMS solutions designed for electric vehicles are not suitable for microgrid energy storage due to differing operational characteristics. Current energy storage BMS implementations often face shortcomings, including insufficient accuracy in estimating the State of Charge (SOC) of energy storage cells, which reduces energy storage utilization, and inefficient energy management based on passive balancing, leading to energy waste and low balancing efficiency.
To address these issues, I have designed an energy storage battery management system based on active balancing. This system adopts a master-slave architecture and a modular design approach to achieve efficient management and energy utilization for microgrid energy storage cells. The core functions of this BMS include data acquisition, SOC estimation, balancing management, safety protection, and communication. By real-time online monitoring of the state parameters of energy storage cells, the master control module performs real-time SOC estimation, comprehensively analyzes and handles faults such as overvoltage, overcurrent, overtemperature, and energy imbalance, and interfaces with the microgrid Energy Management System (EMS), Power Conversion System (PCS), and upper-level monitoring systems to provide a basis for dispatching and managing energy storage devices in the microgrid system.
The overall design of the system is illustrated in Figure 1. The slave control module is responsible for real-time monitoring and acquisition of parameters such as voltage, temperature, and current of the energy storage cell pack. After filtering and analog-to-digital conversion, the data is transmitted to the master control module’s Digital Signal Processor (DSP) via Serial Peripheral Interface (SPI) communication. Upon receiving the sampled data, the master control module conducts battery parameter identification and SOC estimation based on the Unscented Kalman Filter (UKF) algorithm. It analyzes the operating state of the energy storage cell pack; if conditions such as overcharge, over-discharge, or overtemperature are detected, the DSP proactively issues protection commands and sends alarm information. When inconsistencies among individual energy storage cells arise, the master control module initiates balancing strategies and algorithms, utilizing a PWM wave generator to control the balancing module for consistency equalization. Additionally, the master control module communicates battery state parameters to monitoring industrial computers, EMS, PCS, and other devices via Controller Area Network (CAN) bus, providing critical information for microgrid system management.
Hardware Design of the System
The hardware system forms the foundation for the stable operation of the BMS. The designed energy storage cell management system hardware primarily includes the master control module, data acquisition module, balancing module, power supply module, and communication module.
The master control module chip serves as the brain of the BMS, handling tasks such as data reception, computation, and management command issuance, which involve substantial computational loads and complex logical control. I selected the TMS320F2812 chip, which operates at a frequency of 150 MHz with an instruction cycle of approximately 6.67 ns. It features 128 kB of built-in FLASH memory, 18 kB of RAM, supports external memory expansion, includes 16 channels of 12-bit ADC conversion, and offers various communication interface modules such as SCI, SPI, and CAN. This chip is well-suited to the complex environments in which microgrid energy storage systems operate and meets the requirements for a BMS controller. Based on the chip datasheet, I designed a minimized system for the master control chip, including reset circuits, crystal oscillator circuits, and filtering circuits, constituting the core control part of the energy storage cell management system.
Voltage reflects the working state of energy storage cells and provides essential data support for SOC estimation and safety management. I employed the LTC6811 chip to accomplish battery voltage monitoring and active balancing management tasks. This chip offers high voltage acquisition accuracy and strong anti-interference performance, capable of measuring up to 12 series-connected individual energy storage cell voltages with a measurement range of 0 to 5 V. Its built-in analog-to-digital converter facilitates the transmission of measurement results without requiring additional ADC modules. Through the SPI communication interface, it can communicate with the master control module, and multiple LTC6811 chips can be configured in a daisy-chain arrangement using digital isolation chips like Si8441 to monitor more energy storage cells, with only the bottom chip connected to the master control module via an isoSPI port.
The voltage measurement circuit using the LTC6811 chip is shown in Figure 2. The chip’s voltage acquisition terminals measure the voltages of 12 energy storage cells. RC filter circuits with resistors of 100 Ω and capacitors of 10 μF are introduced in the measurement branches to mitigate high-frequency noise interference and reduce voltage ripple. Zener diodes with values greater than twice the working voltage of the energy storage cells are connected in parallel across the cell terminals to enhance the chip’s surge resistance capability.
Leveraging the LTC6811 chip, I designed an active balancing circuit based on an improved Buck-Boost topology to manage consistency among energy storage cells. The circuit is depicted in Figure 3. The chip’s interfaces control the switching transistors of the balancing circuit. I selected IRF2807 N-channel MOSFETs, which can handle currents up to 20 A, have a maximum VDS of 75 V, short turn-on times, and low impedance. Diodes are IN5822 Schottky diodes with low forward voltage drop to improve balancing efficiency. A 1 kΩ resistor is connected in parallel between the source and gate of the MOSFET to prevent gate voltage overshoot and enhance circuit EMC performance. Considering the long wiring from the LTC6811 to the energy storage cells, high-frequency currents during balancing can cause energy loss along the wires; therefore, 22 μF ceramic capacitors are connected in parallel at the energy storage cell terminals to improve balancing energy efficiency.
The LTC6811 chip’s integrated ADC module and General-Purpose Input/Output (GPIO) interfaces provide a convenient means for current and temperature measurement, converting external sensor measurements into the voltage sampling system. I used the DS18B20 one-wire digital temperature sensor, which offers higher accuracy and ease of use compared to thermistors. Temperature sensors can be connected in parallel on a single data bus for data communication, enabling multi-point temperature acquisition. The circuit design is shown in Figure 4, where the DS18B20’s VCC pin is connected to a +3.3 V power supply, the GND pin is grounded, and the signal pin DQ is connected to an I/O port of the LTC6811 through a pull-up resistor.
Hall current sensors provide high sensitivity and accuracy in current detection. I chose the CHB-200SF closed-loop compensation Hall current sensor. The current acquisition circuit is illustrated in Figure 5. The Hall current sensor detects the primary current and, through magnetic induction, produces a secondary current. This current signal is first converted into a voltage signal via resistor R16, then processed through noise reduction, inversion, and adder circuits before being input to an I/O interface of the LTC6811. Using the internal ADC conversion module, synchronous acquisition and conversion of voltage and current data are achieved.
The master control and acquisition modules require different voltage levels to function normally and efficiently, necessitating various DC/DC power supply modules for voltage conversion. In this design, each module is powered by the energy storage cell pack, eliminating the need for an additional power source. The required voltage levels in the circuit include 12 V, 5 V, and 3.3 V. Power is derived from the 12-series energy storage cell pack monitored by the voltage measurement chip using the LT3990 voltage regulator chip, which converts 15–62 V to 12 V. An isolated step-down power chip, WRB1205S, then converts 12 V to 5 V, and finally, the XC6209 chip converts 5 V to 3.3 V. The circuit diagram is presented in Figure 6.
Software Design of the System
In the Code Warrior development environment, I developed software for the minimized system of the master control module TMS320F2812 to implement the core control functions of the energy storage cell management system. Key programs include hardware system initialization, data acquisition, SOC estimation, active balancing management, safety protection, and communication configuration. The software program flow is depicted in Figure 7.
The system first performs initialization settings, configuring and assigning parameters for clock, I/O ports, communication, etc. It then runs the parameter acquisition subroutine to sample voltage, current, and temperature. The results are converted via the ADC module and uploaded to the DSP through SPI serial communication. The SOC estimation subroutine based on the Unscented Kalman Filter algorithm and the safety protection subroutine run concurrently to monitor the working state of the energy storage cells and ensure system safety. The active balancing subroutine, using SOC as the balancing variable, determines whether balancing needs to be initiated based on the SOC of the energy storage cells, thereby achieving consistency management in series-connected energy storage cell packs. Finally, information and control commands are exchanged with peripheral devices such as the upper-level monitor, EMS, and PCS via CAN bus communication.
Experimental Validation
I conducted experiments using a pack of 12 energy storage cells connected in series as the test subject for the battery management system, primarily validating measurement accuracy and balancing effectiveness.
Measurement Accuracy Test
Accurate state detection of energy storage cells is fundamental to the normal operation of the battery management system, and its precision directly impacts BMS management effectiveness. I used a Fluke 175C digital multimeter to measure the voltages of the series-connected energy storage cells and compared the results with the voltage detection results from the BMS, as shown in Table 1. The Fluke 175C digital multimeter has a DC voltage measurement resolution of 0.1 mV, providing extremely high measurement accuracy, and can be considered the true value for the experiment. From the data in Table 1, the difference between the BMS sampled values and the multimeter measured values is within ±5 mV, indicating high measurement accuracy for individual energy storage cell voltages, meeting design requirements.
| Cell Number | Sampled Value (V) | Measured Value (V) | Error (mV) |
|---|---|---|---|
| 1 | 3.245 | 3.249 | 4 |
| 2 | 3.252 | 3.254 | 2 |
| 3 | 3.201 | 3.200 | -1 |
| 4 | 3.250 | 3.247 | -3 |
| 5 | 3.184 | 3.186 | 2 |
| 6 | 3.215 | 3.218 | 3 |
| 7 | 3.226 | 3.223 | -3 |
| 8 | 3.237 | 3.241 | 4 |
| 9 | 3.268 | 3.266 | -2 |
| 10 | 3.216 | 3.218 | 2 |
| 11 | 3.221 | 3.218 | -3 |
| 12 | 3.239 | 3.244 | 5 |
To validate the temperature acquisition accuracy of the system, I placed the energy storage cell pack in a temperature chamber. The chamber temperature was set in the range of -10°C to +50°C, with experiments conducted every 10°C. The temperature sampling results are presented in Table 2. As seen, the system temperature sampling error is within ±1°C, demonstrating high measurement accuracy.
| Temperature Chamber (°C) | System Sampled Temperature (°C) | Error (°C) |
|---|---|---|
| -10 | -10.5 | -0.5 |
| 0 | -0.2 | -0.2 |
| 10 | 9.7 | -0.3 |
| 20 | 20.1 | 0.1 |
| 30 | 29.8 | -0.2 |
| 40 | 40.6 | 0.6 |
| 50 | 50.8 | 0.8 |
Balancing Experiment Test
I tested two groups of series-connected energy storage cells. Each group had identical initial SOC values, with a maximum SOC difference of 6% among individual energy storage cells within the group. The first group was connected to the balancing circuit, while the second group was not. A constant current discharge was applied to the entire energy storage cell pack using an electronic load. SOC data displayed on the upper-level monitor was recorded every 3 minutes over a 30-minute test period. The results are shown in Table 3. The energy storage cell pack with the balancing system achieved a final SOC error among individual energy storage cells within 1%, whereas the unbalanced pack showed a maximum SOC error of 6.09%, with inconsistency tending to increase.
| Cell Number | Initial SOC (%) | Group 1 (Balanced) Final SOC (%) | Group 2 (Unbalanced) Final SOC (%) |
|---|---|---|---|
| 1 | 74 | 68.85 | 70.52 |
| 2 | 76 | 68.92 | 71.23 |
| 3 | 73 | 68.57 | 68.85 |
| 4 | 75 | 68.75 | 70.21 |
| 5 | 72 | 68.34 | 67.95 |
| 6 | 70 | 68.05 | 65.14 |
The SOC estimation for energy storage cells is a critical function of the BMS. I employed the Unscented Kalman Filter (UKF) algorithm for SOC estimation due to its ability to handle nonlinear systems without linearization, providing higher accuracy. The state-space model for the energy storage cell can be represented as:
$$x_k = f(x_{k-1}, u_k) + w_k$$
$$y_k = h(x_k, u_k) + v_k$$
where \(x_k\) is the state vector (including SOC), \(u_k\) is the input (current), \(y_k\) is the output (voltage), \(w_k\) and \(v_k\) are process and measurement noises, respectively. The UKF algorithm uses sigma points to approximate the distribution of the state. The SOC is defined as:
$$SOC(t) = SOC(0) – \frac{1}{Q} \int_0^t \eta i(\tau) d\tau$$
where \(Q\) is the capacity of the energy storage cell, \(\eta\) is the coulombic efficiency, and \(i(\tau)\) is the current. The UKF update equations involve prediction and correction steps. For example, the prediction step computes:
$$\chi_{k-1} = [\hat{x}_{k-1}, \hat{x}_{k-1} + \sqrt{(n+\lambda)P_{k-1}}, \hat{x}_{k-1} – \sqrt{(n+\lambda)P_{k-1}}]$$
where \(\chi\) are sigma points, \(n\) is the state dimension, \(\lambda\) is a scaling parameter, and \(P\) is the covariance matrix. The active balancing control is based on SOC differences. The balancing current \(I_{bal}\) for an energy storage cell is determined by:
$$I_{bal,i} = K_p (SOC_{avg} – SOC_i) + K_i \int (SOC_{avg} – SOC_i) dt$$
where \(SOC_{avg}\) is the average SOC of the pack, \(SOC_i\) is the SOC of the i-th energy storage cell, and \(K_p\) and \(K_i\) are proportional and integral gains. This ensures that energy is transferred from higher-SOC energy storage cells to lower-SOC ones, improving consistency.
Conclusion
I have designed an energy storage cell management system for microgrid applications using a master-slave architecture. Based on the TMS320F2812 and LTC6811 chips, with modular circuit design, the system accomplishes functions such as energy storage cell pack state detection, SOC estimation, balancing management, and safety protection, enabling efficient management of series-connected energy storage cell packs. Through physical system validation, this design achieves voltage measurement accuracy within 5 mV and temperature measurement accuracy within 1°C, demonstrating high detection precision. The SOC-based active balancing management maintains energy storage cell pack consistency within 1%, showing effective balancing performance. This design offers valuable reference for practical applications, enhancing the monitoring and management capabilities of energy storage cells in microgrid systems, ultimately contributing to improved energy utilization efficiency and system reliability.