This paper presents a comprehensive study on the design of a battery management system (BMS) for a pure electric bus application, focusing on the monitoring of lithium iron phosphate (LiFePO₄) power battery modules and the estimation of their state of charge (SOC). The work covers the performance analysis of individual cells and modules, the hardware and software architecture of the BMS, the design of a CAN-based communication network following SAE J1939 and TTCAN protocols, the development of an upper-computer graphical interface for debugging and fault analysis, and a fuzzy-logic-based SOC estimation algorithm using the Mamdani inference method. The proposed SOC estimator is validated through two independent approaches: direct residual capacity measurement of a battery module and vehicle road-test data analysis. The results show that the SOC estimation accuracy reaches within 5%, satisfying the practical requirements of pure electric buses. The BMS demonstrates high reliability, low hardware cost, and a short development cycle, making it suitable for large-scale deployment in electric public transportation.
1. Introduction
With the rapid development of new energy vehicles, pure electric buses have entered a critical phase of industrialization. Cities such as Beijing, Hefei, and Shanghai have already launched demonstration bus routes. The widespread adoption of pure electric buses has created both opportunities and challenges for lithium iron phosphate battery technology. Among the most difficult issues are the cell-to-cell consistency within battery modules and the accurate estimation of the state of charge (SOC). In this context, I designed a battery management system to monitor the battery state and used a fuzzy logic algorithm to estimate the SOC of LiFePO₄ power battery packs. Two verification methods were adopted to evaluate the SOC estimation error, contributing to the promotion of pure electric buses and the reduction of energy consumption and environmental pressure.
1.1 Development Status of Pure Electric Buses
In 2009, the Chinese government issued strong policies to support new energy vehicles. The Ministry of Finance, the National Development and Reform Commission, the Ministry of Science and Technology, and the Ministry of Industry and Information Technology were jointly formulating policies to promote the development of electric vehicles. The State Council attached great importance to electric vehicle projects. Many pure electric vehicles had already been put into trial operation in several cities. According to the “Automotive Adjustment and Revitalization Plan”, the sales share of new energy vehicles was initially set at 5% of passenger cars. Based on the production and sales volume of buses in 2008 (160,000 units), the minimum number of new energy buses in 2009 was at least 10,000 units. With an average annual growth rate of 10%, the number of new energy buses could reach 13,000 by 2012. Additionally, the “Ten Cities, Thousand Vehicles” program and the efforts of various provinces to improve urban environments and public transport capacity have driven the establishment of pure electric demonstration routes. Therefore, from 2010 to 2012, the number of new energy buses was expected to reach at least 25,000 units by 2012, with pure electric buses accounting for more than 40% of the total. By 2020, the market demand for new energy buses is projected to be 45,000 units, of which pure electric buses will account for more than 60%. The bus industry will predominantly adopt pure electric vehicles.
1.2 Application Status of LiFePO₄ Power Batteries
LiFePO₄ batteries are considered the safest type of lithium-ion battery available today. They have excellent thermal stability up to 400–500 °C, which ensures the high safety of battery modules. They do not explode or burn under overcharging, over-discharging, high temperature, short circuit, or impact. Due to these advantages, LiFePO₄ batteries are widely used in large electric vehicles, light electric vehicles, solar and wind energy storage systems, and various portable devices. The following table compares several rechargeable battery technologies:
| Parameter | LiCoO₂ | LiMn₂O₄ | LiFePO₄ |
|---|---|---|---|
| Tap density (g/cm³) | 2.8–3.0 | 2.2–2.4 | 1.0–1.4 |
| Specific surface area (m²/g) | 0.4–0.6 | 0.4–0.8 | 12–20 |
| Specific capacity (mAh/g) | 135–140 | 100–115 | 130–140 |
| Voltage plateau (V) | 3.6 | 3.7 | 3.2 |
| Cycle life | ≥300 | ≥500 | ≥2000 |
| Transition metal | Poor | Rich | Very rich |
| Raw material cost | High | Low | Low |
| Environmental impact | Contains cobalt | Nontoxic | Nontoxic |
| Safety | Poor | Good | Excellent |
| Typical applications | Small batteries (phones) | Power batteries (bicycles) | Power batteries / large-format energy storage |
1.3 BMS Technology Status
In the application of power batteries for pure electric buses, the cycle life of a single cell cannot fully represent the life of the entire battery pack. The long life of LiFePO₄ cells is often hindered by the lack of consistency among cells when connected in series or parallel. Even if a single cell can achieve 2000 cycles, a pack of more than a dozen cells often cannot reach 1000 cycles. The BMS must operate over a wide temperature range, handle high-rate charging/discharging, and manage the heat dissipation of densely packed cells. The fundamental responsibilities of a BMS are to effectively monitor the battery state, optimize energy usage, extend battery life, and communicate with external devices such as chargers and electronic control units. Current BMS products often suffer from insufficient accuracy due to the complex nonlinear behavior of batteries, the difficulty of modeling internal electrochemical processes, and the lack of deep integration among battery manufacturers, BMS developers, charger suppliers, and motor controller suppliers.
1.4 SOC Estimation Research Status
Most existing BMS products define SOC based on a single cell, which is not suitable for a module or pack. The capacity of a battery pack is closely related to the capacity and SOC of every individual cell. If one cell becomes fully charged or fully discharged first, the pack capacity becomes smaller than that of the weakest cell. Common SOC estimation methods include ampere-hour counting, open-circuit voltage (OCV) measurement, neural networks, Kalman filtering, and electrochemical impedance spectroscopy. Ampere-hour counting is an open-loop method that serves as a baseline but suffers from accumulated errors. OCV methods require long rest periods and are unsuitable for real-time vehicle operation. Neural networks require large amounts of training data and are sensitive to training quality. Kalman filters are well suited for dynamic current profiles but require high computational power. My proposed fuzzy logic approach offers a practical alternative that balances accuracy and computational efficiency, especially for the highly nonlinear characteristics of LiFePO₄ batteries.
2. Performance Analysis of LiFePO₄ Power Battery
This chapter presents the working principle and performance of LiFePO₄ cells and modules. The SOC estimation of a battery pack is closely related to the cell performance, consistency, and module structure. Therefore, a series of charge/discharge tests on cells and modules were conducted to provide the basis for the BMS hardware design and SOC estimation.
2.1 Working Principle and Performance Indicators
The charge and discharge processes of a LiFePO₄ battery involve the insertion and extraction of lithium ions between the positive and negative electrodes. The reaction equations are as follows:
$$ \text{Charge: } \mathrm{LiFePO_4 – xLi^+ – xe^- \rightarrow xFePO_4 + (1-x)LiFePO_4} \tag{2-1} $$
$$ \text{Discharge: } \mathrm{FePO_4 + xLi^+ + xe^- \rightarrow xLiFePO_4 + (1-x)FePO_4} \tag{2-2} $$
The internal structure consists of an olivine-structured LiFePO₄ positive electrode connected to aluminum foil, a polymer separator that allows lithium ions to pass while blocking electrons, a carbon (graphite) negative electrode connected to copper foil, and a metal casing hermetically sealed. When charging, lithium ions migrate from the positive to the negative electrode through the separator; when discharging, the migration direction is reversed. LiFePO₄ batteries have significant advantages: high safety, no explosion or combustion, good performance at elevated temperatures (65 °C ambient, internal temperature up to 95 °C, still safe at 160 °C), discharge to zero volts without damage, environmental friendliness, high rate capability (2–5 C standard discharge, up to 10 C continuous, 20 C pulse), fast charging, low self-discharge, no memory effect, and excellent cycle life (more than 95% capacity after 2000 cycles at 1C).
2.2 Single Cell Charge/Discharge Tests
For each batch of cells, a sampling and detailed testing procedure was conducted to determine the true capacity and external characteristics. The equipment included an Arbin battery test system, a HIOKI internal resistance meter, a temperature chamber, and 26650E cylindrical cells. The test procedure involved:
- Randomly selecting several cells and performing a standard 1C constant-current charge to 3.7 V, followed by constant-voltage charge until the current dropped to 100 mA.
- Recording the constant-current charge capacity, constant-voltage charge capacity, and total charge capacity.
- Discharging at 1C to 2.0 V and recording the total discharge capacity.
- Performing external characteristic tests: charge curves at 0.5C and 1C, discharge curves at 0.2C, 1C, and 3C at 20 °C, discharge curves at different temperatures, and cycle life tests.
The charge curves at normal temperature for 0.5C and 1C are shown below.
The discharge curves at different temperatures (0.2C) show that the capacity is strongly affected by temperature, especially below 8 °C.
The discharge curves at normal temperature for different rates indicate the voltage plateau typical of LiFePO₄ batteries.
The cycle life test results demonstrate that after 1000 cycles the capacity retention remains above 95%.
2.3 Module Battery Structure and Tests
The battery module (designated CBE24300) consists of multiple cells connected in series and parallel using metal tabs and laser-welded copper plates. The module specifications are given below.
| Parameter | Value |
|---|---|
| Nominal voltage | 25.6 V |
| Nominal capacity | 300 Ah |
| Dimensions (L×W×H) | 700 mm × 242 mm × 330 mm |
| Weight | 75 kg |
| Specific energy | 100 W/kg |
| Specific power | 250 W/kg |
| Charging cutoff voltage | 28.8 V |
| Nominal charge current (20 °C) | 300 A |
| Continuous discharge current | 300 A |
| Peak current (10 s) | 900 A |
| Discharge cutoff voltage | 20.0 V |
| Charging temperature range | 0–40 °C |
| Discharging temperature range | -20–60 °C |
| Storage temperature | -40–70 °C |
| Features | Dynamic self-balancing, embedded cooling tubes, 8-cell voltage acquisition, embedded BMU, 8 temperature sensors |
The module structure has the following features: eight cells in series form one string, twenty-five strings are connected in parallel to form one layer, four layers are paralleled to obtain 300 Ah capacity, with silver-plated cooling tubes inside and copper sheets laser-welded for connection. A BMS is embedded within the module.
The module charge/discharge tests were performed using a ZM8960 power battery test system. The true capacity of the module was determined by charging at 1C to 28.8 V then constant-voltage until the current fell to 3 A, with individual cell voltage limits as the stop condition. Discharge was performed at 1C to 20 V, again with cell voltage limits. External characteristic tests included 1C charge/discharge curves, internal temperature curves at 8 points during 1C cycling, discharge curves at 0.2C, 1C, and 3C at 20 °C, and an accelerated pulse cycle test. The results show that the module behaves differently from a single cell due to the series-parallel configuration and spatial temperature gradients.
2.4 SOC-Related Parameter Analysis
The cell-to-cell differences become apparent at high SOC, low SOC, and especially at SOC levels around 9–10%. These differences are caused by variations in DC internal resistance, polarization voltage, capacity, and SOC. While DC resistance and capacity differences cannot be corrected by balancing, SOC differences can be compensated. However, the nonlinear relationship between voltage and SOC makes balancing decisions difficult. I proposed a comprehensive evaluation method based on current, internal resistance, polarization voltage, capacity, and SOC. Ground simulation data for 20 modules in a vehicle were collected, and the relationship between open-circuit voltage and SOC was derived from experiments.
| SOC (%) | V_cell (V) | V_module (V) |
|---|---|---|
| 100 | 3.414 | 27.312 |
| 90 | 3.394 | 27.152 |
| 80 | 3.342 | 26.736 |
| 70 | 3.329 | 26.632 |
| 60 | 3.306 | 26.448 |
| 50 | 3.303 | 26.416 |
| 40 | 3.302 | 26.410 |
| 30 | 3.301 | 26.408 |
| 20 | 3.285 | 26.281 |
| 10 | 3.253 | 26.024 |
| 5 | 3.102 | 24.816 |
| 0 | 2.621 | 20.968 |
3. Design of the Battery Management System
The BMS is responsible for monitoring the state of each battery module, estimating SOC, controlling the thermal management system, managing the breaker, and communicating with the vehicle controller and other ECUs. The overall architecture consists of two main units: the Battery Management Unit (BMU) embedded in each module, and the Battery Electronic Control Unit (BECU) installed at the vehicle rear.
3.1 Hardware Design and Functions
The BMU hardware includes a TMS320LF2407 DSP, voltage acquisition circuits with three-stage filtering for cell voltage measurement, eight DS18B20 digital temperature sensors, a non-volatile memory for storing module ID, a CAN transceiver PCA82C250, and an equalization circuit (EQU). The voltage sampling circuit is designed to suppress common-mode and series-mode interference. The BECU hardware includes an MPC5510 microcontroller, dual-range LEM current sensors (-20 A to 20 A and -500 A to 500 A) for accurate current measurement, two CAN controllers (CAN1 for internal BMU communication and CAN2 for vehicle external communication), communication interfaces for chargers, thermal management control for air compressor, breaker control for safety, non-volatile memory for storing module IDs and calibration coefficients, and an SCI module for PC communication. The overall BMS architecture integrates all these functions and includes additional features such as power management, thermal management, safety management, and data communication.
3.2 Signal Acquisition System Design
The key to reliable BMS operation is accurate acquisition of voltage, current, and temperature. The cell voltage acquisition circuit is designed to handle the high potential differences between series-connected cells. The total voltage and current sensing system uses a voltage-type sampling circuit with high requirements on anti-interference, zero drift, temperature drift, and linearity. The temperature measurement system uses DS18B20 digital sensors for their simplicity, low cost, and good noise immunity.
3.3 Software Design
Software design is divided into BMU software and BECU software. Both use the μC/OS-II real-time operating system. The porting of μC/OS-II to the TMS320LF2407 and MPC5510 was accomplished by modifying the OS_CPU.H, OS_CFG.H, OS_CPU.ASM, and OS_CPU.C files. The application software is designed as a set of priority-based tasks.
The BMU tasks are listed in the following table:
| Priority | Task Name | Function |
|---|---|---|
| 2 | BMUIDTask() | Vehicle networking |
| 3 | BMUADCTask() | Digital filtering of ADC data and data exchange |
| 4 | BMUECANRXDTask() | CAN bus data reception |
| 5 | BMUECANTXDTask() | CAN bus data transmission |
| 7 | BMUTEMPTask() | Acquisition of 8 temperature points |
| 9 | BMUEQUTask() | Cell equalization |
| 13 | BMUSCIRXDTask() | SCI data reception |
| 14 | BMUSCITXDTask() | SCI data transmission |
The interrupt service routines include BMUADCISR, BMUT1fISR, BMUOSTickISR, BMUCANISR, and BMUSCIISR. The system tick is set to 2 ms, the ADC and CAN transmission tasks execute every 10 ms, the temperature task every 50 ms, and the SCI transmission task every 20 ms. Task communication uses message mailboxes to minimize overhead.
The BECU tasks are shown below:
| Priority | Task Name | Function |
|---|---|---|
| 2 | BECUIDTask() | Vehicle networking |
| 3 | BECUADCTask() | ADC sampling and filtering |
| 4 | BECUCAN1RXDTask() | CAN1 internal bus reception |
| 5 | BECUCAN1TXDTask() | CAN1 internal bus transmission |
| 6 | BECUCAN2RXDTask() | CAN2 external bus reception |
| 7 | BECUCAN2TXDTask() | CAN2 external bus transmission |
| 8 | BECUTempManageTask() | Air compressor control |
| 9 | BECUSafeManageTask() | Breaker control |
| 10 | BECUSOCTask() | SOC estimation |
| 11 | BECUMoniTask() | System monitoring |
| 13 | BECUSCIRXDTask() | SCI data reception |
| 14 | BECUSCITXDTask() | SCI data transmission |
3.4 Vehicle Communication Network Design
The pure electric bus uses a CAN bus following the SAE J1939 protocol. The network consists of eight nodes: vehicle control unit, motor controller, AMT, ABS, EPS, battery management system (BMS), and instrument cluster. The bus speed is 250 kbps. The source addresses are assigned as follows:
| Node | Source Address |
|---|---|
| Vehicle controller | 208 |
| Motor controller | 239 |
| AMT | 3 |
| ABS | 11 |
| EPS | 228 |
| Battery management system (BECU) | 243 |
| Instrument cluster | 40 |
The BMS uses two CAN channels: CAN1 for internal communication with all BMUs (source addresses 1 to 20), and CAN2 for external communication with the vehicle network. The BECU sends periodic messages to the instrument cluster. The following tables show some of the message definitions.
| Message 1 | ID: 0xCF0FF20 (65312) | Period: 100 ms |
|---|---|---|
| Byte 1 | Battery total voltage low byte | 1 V/bit, offset 0, range 0–650 V |
| Byte 2 | Battery total voltage high byte | |
| Byte 3 | Battery total current low byte | 1 A/bit, offset -400, range -400–400 A |
| Byte 4 | Battery total current high byte | |
| Byte 5 | SOC | 0.4%/bit, offset 0, range 0–100% |
| Byte 6 | SOH | 0.4%/bit, offset 0, range 0–100% |
| Byte 7 | Remaining energy / estimated range | 1 kWh/bit, offset 0, range 0–250 kWh |
| Byte 8 | Battery status | See Table 3-8 for bit definitions |
Additionally, messages for maximum/minimum cell voltage and temperature along with the corresponding module numbers are transmitted every 100 ms. Messages 3 through 5 transmit the temperature of all 20 modules (500 ms period), and messages 6 through 25 transmit the individual cell voltages of all 20 modules (500 ms period).
3.5 Network Robustness Design
The pure electric bus requires a real-time control network with high reliability. Traditional event-triggered CAN is not sufficient; therefore, a time-triggered CAN (TTCAN) protocol was adopted. In TTCAN, a master node sends reference frames containing the global time, and all other nodes synchronize to this time. A system matrix predefines the transmission time of all messages, reducing response time. The communication delay of TTCAN can be divided into generation delay, queue delay, transmission delay, and reception delay. The queue delay for periodic messages is zero because they are transmitted in exclusive time windows. For event-triggered messages, the delay depends on the position and length of arbitrary time windows. The expected queue delay for an event-type message \(m\) is given by:
$$ E\left[ t_m \right] = t_{\text{arbi}} + t_{\text{nonarbi}} + G_m $$
where \(G_m\) is the expected delay caused by exclusive and free time windows:
$$ G_m = \sum_{i=1}^{Z} \frac{(W_i + C_m)^2}{2 Q T} $$
Here, \(Z\) is the number of exclusive and free windows, \(W_i\) is the length of window \(i\), \(C_m\) is the transmission time of message \(m\), \(Q\) is the number of basic cycles, and \(T\) is the basic cycle period.
To further enhance robustness, several hardware anti-interference measures were implemented: bypass capacitors between motor controller/drive motor and chassis to suppress common-mode interference, DC/DC isolation power supplies for each BMU, and optocouplers (6N137) for CAN, RS232, and SPI communication interfaces to block electromagnetic coupling.
3.6 Upper-Computer Software Design
For debugging and maintenance, I developed an upper-computer graphical interface using Visual C++ with the MSComm control for SCI serial communication and the Kvaser CANlib SDK for the Kvaser USBcan II device. The SCI interface allows the user to graphically display driving current, voltage, and temperatures, and to set BECU and BMU IDs. The Kvaser-based software is designed for engineering analysis: it monitors CAN bus errors, counts lost frames, and logs data. The interface is shown in the development environment screenshot. The software supports both 11-bit and 29-bit identifiers and a listen-only mode for analyzing bus traffic.
4. Fuzzy Logic Algorithm for SOC Estimation
SOC estimation is a key and difficult issue in BMS design. Due to the highly nonlinear behavior of LiFePO₄ battery packs under varying load conditions, I adopted a fuzzy logic approach that mimics the reasoning of experienced engineers. The fuzzy logic algorithm is computationally efficient and can be implemented on the low-cost microcontroller without the heavy load of Kalman filtering.
4.1 Definition of SOC for Module Battery
For a single cell, SOC is defined as the ratio of remaining capacity to rated capacity:
$$ SOC_{\text{cell}} = \frac{Q_c}{C_i} $$
where \(Q_c\) is the remaining capacity and \(C_i\) is the capacity at a constant current \(i\) (usually 1C). For a module, the SOC is similarly defined as \(SOC_{\text{module}} = Q_c / C_i\), where \(C_i\) is the module capacity at 1C discharge. However, this definition has limitations under variable discharge currents and temperature effects. To improve accuracy, I used the voltage correction method described in the next section.
4.2 Fuzzy Logic Scheme
The concept is to correct the measured terminal voltage \(U_I\) of the module to a reference voltage \(U_{60}\) corresponding to an average discharge current of 60 A. The simplified physical model of the LiFePO₄ module is represented by an ideal voltage source \(E_0\), an ohmic resistance \(R_0\), and a polarization resistance \(R_r\) with a parallel capacitance \(C_r\). At steady state, the terminal voltage can be expressed as:
$$ U_I = E_0 – I (R_0 + R_r(I)) $$
For \(I=60\) A:
$$ U_{60} = E_0 – 60 (R_0 + R_r(60)) $$
Subtracting the two equations gives:
$$ U_{60} = U_I + (I – 60) \cdot r $$
where \(r\) is an integrated internal resistance:
$$ r = \frac{U_I – U_{60}}{I – 60} $$
In practice, the integrated resistance can be identified from a set of discharge curves at different currents. This correction allows the fuzzy algorithm to use a voltage value that is largely independent of the discharge current.
4.3 Membership Functions
The corrected voltage \(U_{60}\) ranges from 20 V to 28 V. It is partitioned into seven fuzzy subsets: {VVH, VH, H, M, S, VS, VVS} (Very Very High, Very High, High, Medium, Small, Very Small, Very Very Small). The membership functions are asymmetric because the sensitivity of voltage to SOC changes with the discharge region. The temperature \(\theta\) ranges from -20 °C to 40 °C and is divided into four subsets: {Vcold, Cold, Warm, Hot}. The output SOC ranges from 0 to 100% and is divided into seven subsets: {VLow, Low, ML, Medium, MH, High, VHigh}.
4.4 Fuzzy Rules
Fuzzy rules are derived from experimental data and experienced engineers’ knowledge. Examples:
| Rule | Antecedent | Consequent |
|---|---|---|
| 1 | Voltage is VVS | SOC is VLow |
| 2 | Voltage is VS and temperature is Hot | SOC is VLow |
| 3 | Voltage is S and temperature is Hot | SOC is Low |
| 4 | Voltage is M and temperature is Hot | SOC is ML |
| 5 | Voltage is H and temperature is Hot | SOC is Medium |
| 6 | Voltage is H and temperature is Hot | SOC is MH |
| 7 | Voltage is VH and temperature is Hot | SOC is High |
| 8 | Voltage is VVH | SOC is VHigh |
The complete rule table is a 7×4 matrix. During vehicle testing, the weights of the rule consequents are adjusted to obtain the desired SOC accuracy.
4.5 Defuzzification Using Mamdani Min-Max Centroid Method
The Mamdani inference method is applied to combine the fuzzy outputs. For each rule \(j\), the output membership function is clipped by the minimum of the input membership degrees. Then the combined output fuzzy set is the maximum of all clipped sets. Finally, the centroid of the combined set yields the crisp SOC value:
$$ SOC^* = \frac{\sum_{i=1}^{n} y_i \mu_C(y_i)}{\sum_{i=1}^{n} \mu_C(y_i)} $$
where \(y_i\) are the discrete output values and \(\mu_C(y_i)\) is the final membership degree. The figure below illustrates the MIN-MAX-centroid process for a specific input voltage and temperature. The shaded area represents the centroid of the fuzzy output, which corresponds to the estimated SOC.
5. Vehicle Road Test and SOC Estimation Verification
A 12-meter pure electric bus was instrumented with the proposed BMS. The vehicle was tested according to the technical requirements for pure electric buses. Several performance metrics were evaluated.
5.1 Vehicle Test Items and Performance Indicators
| Category | Item | Target |
|---|---|---|
| Dynamic performance | Maximum speed (full load) | ≥80 km/h |
| Acceleration 0–50 km/h | ≤15 s | |
| Acceleration 0–30 km/h | ≤8 s | |
| Maximum gradeability | ≥25% | |
| Braking | Braking distance at 30 km/h (full load) | ≤10 m |
| Maximum parking grade | ≥20% | |
| Coasting | Coasting distance from 50 km/h | ≥800 m |
| Noise | Interior noise at 50 km/h | ≤78 dB(A) |
| Exterior acceleration noise | ≤82 dB(A) | |
| Ride comfort | Comfort limit TCD | ≥1.0 h |
| Electrical insulation | Insulation resistance of traction battery | >33.6 kΩ |
| Safety | Creepage distance | >8 mm |
The BMS functional tests included power consumption (BMU max 1.5 W, min 0.012 W; BECU max 2 W, min 0.02 W), equalization current of 500 mA, thermal protection at 65 °C, overvoltage protection (3.7 V max, 2.5 V min), overcurrent protection (900 A max), and EMC compliance with relevant standards.
5.2 SOC Estimation Accuracy Verification
I used two independent methods to verify the SOC estimation accuracy of the fuzzy logic algorithm.
Method 1: Residual capacity measurement. Two arbitrary modules were randomly selected from the bus. They were discharged at 300 A and 20 °C to a cutoff voltage of 20 V. The discharged capacity was found to be 158 Ah. Then the modules were subjected to one full 1C cycle, giving a capacity of 290 Ah. This indicates that after three months of operation, the real available capacity was 158 Ah when the BMS indicated a certain SOC point. By comparing the residual capacity with the indicated SOC, the error was calculated.
Method 2: Trend curves of vehicle operating parameters. The vehicle was driven under real road conditions. The SOC, vehicle speed, distance, motor output power, and accelerator pedal position were recorded using the Kvaser data acquisition system. The relationship curves between SOC and these parameters were analyzed. The fuzzy-logic estimated SOC was compared with the integrated ampere-hour reading and the open-circuit voltage method after rest. The two methods complement each other: the first gives an accurate but offline measurement, while the second provides a real-time tendency check.
5.3 Results and Parameter Correction
The verification results showed that the SOC estimation error was within 5% of the true value. Based on the verification data, I adjusted the membership functions and rule weights of the fuzzy algorithm. For instance, the integrated internal resistance parameter was fine-tuned according to the module temperature and aging status. After the correction, the SOC estimation remained accurate over the full operating range, from 0% to 100%, under varying load conditions and temperature.
Conclusion
In this study, I designed a battery management system specifically for LiFePO₄ power battery packs used in pure electric buses. The work included battery cell and module testing, hardware and software design, CAN network design, upper-computer software development, and a fuzzy-logic SOC estimation algorithm. The following conclusions were drawn:
- The designed battery module experiments satisfy the vehicle test requirements and provide a reliable basis for SOC verification.
- The BMS provides a stable hardware and software platform for SOC estimation. Its reliable operation during road tests confirms the rationality of the design.
- The fuzzy-logic-based SOC estimation achieves an accuracy of 5%, which meets the practical application requirements.
This research provides a practical solution for battery monitoring and SOC estimation in pure electric buses. Future work will focus on developing a more universal BMS that can accommodate different battery chemistries, improving the SOC estimation algorithm with adaptive parameters, and exploring optical fiber communication for the vehicle CAN network to further enhance immunity to electromagnetic interference.
Through this project, I have gained deep insight into the challenges of battery management for electric vehicles. The combination of fuzzy logic with carefully designed hardware and communication protocols demonstrates that a low-cost, reliable, and accurate BMS is achievable for the rapidly growing electric bus market. The ‘EV battery pack’ remains at the heart of this achievement, and the methods presented here contribute directly to maximizing its usable capacity, extending its lifetime, and ensuring safe operation. The ‘EV battery pack’ technology is evolving rapidly, and the BMS must evolve together with it, incorporating advanced algorithms and more robust communication interfaces to meet future demands. In summary, the successful design and validation of this BMS proves that a well-engineered system can significantly enhance the performance and safety of the ‘EV battery pack’ in pure electric buses.