Design of battery management system for pure electric vehicles based on LIN bus

Abstract: An online monitoring and management system for lithium iron phosphate battery pack based on LIN bus is proposed. The system adopts a distributed network control structure. Through the design of Dspic30f4012 chip as the core underlying hardware, it realizes the accurate monitoring of lithium iron phosphate battery parameters, realizes data transmission through LIN bus technology, and estimates the battery state of charge (SOC) based on a more accurate battery model by using the extended Kalman algorithm, which improves the estimation accuracy. The experimental results show that the system can well monitor and effectively protect the battery pack in real time, and provides application value for the development of intelligent battery management system for electric vehicles.

As a new type of electric vehicle power battery, lithium iron phosphate battery has the advantages of large capacity, high safety, high temperature resistance, and especially long cycle life. Its cycle life is at least 4 times higher than that of ordinary lead-acid batteries, and it has great application potential in the automotive power battery market. At present, without a fundamental breakthrough in the capacity of power batteries, the application of battery management system (BMS) in electric vehicles will be extremely important. It can detect the voltage, current, and temperature of the power battery in real time, and estimate the battery state of charge (SOC) through these parameters to provide drivers with a reference for vehicle driving range; in addition, BMS can alarm and protect the battery from overcharge and overdischarge, and effectively protect the battery pack and single battery, thereby improving battery performance and battery life. LIN bus is a low-cost automotive Class A bus, which is very suitable for data transmission with low real-time requirements such as temperature and current. The bus transmission of data through LIN bus further reduces costs.

1 Overall structure and function of the system

In this design, the battery management system is divided into two parts: signal detection module and communication and information processing module. In the signal detection module, each single cell corresponds to an underlying ECU (DSPic30f4012), which can realize single cell voltage acquisition, current detection, and temperature sampling; it can also detect the voltage, current, and ambient temperature of the entire battery pack, which is used for detection and protection of the battery during general charging and balanced charging, as shown in Figure 1.

The bottom ECU encapsulates the detected variables such as voltage, current, temperature, etc. into the LIN bus frame format, and then communicates with the upper ECU through the LIN bus. The information processing module can realize real-time estimation of the state of charge of the power battery and fault analysis, and display information such as temperature, voltage, current, etc.

2 Battery Management System Design

2.1 Basic Hardware Design of Battery Management System

Due to the large number of cells in the battery pack, this system adopts a distributed structure, which can effectively reduce the sampling line passing through the battery, reduce the complexity of installation and debugging, and also reduce safety hazards. The underlying ECU uses the Dspic30f4012 chip, which can work in the temperature range of -40~125℃ and is an automotive-grade chip; it has rich analog and digital I/O interfaces, 10-bit A/D conversion functions, and SCI communication functions.

2.1.1 Signal acquisition module design

Dspic30f4012 has a wide operating voltage range of 2.5~5.5 V, so it can be directly powered by a single lithium iron phosphate battery. Only a 0.1 μF filter capacitor is needed to make the chip work, and the power supply circuit is greatly simplified. Since the F4012 chip does not provide an internal reference voltage for A/D conversion, an external A/D conversion reference voltage is required when performing voltage detection. This article uses the low-power, low-voltage error LM385 to provide a 2.5 V external reference voltage, as shown in Figure 2.

The characteristic of the voltage detection module in this design is that each detection module detects the voltage on its own single cell separately, instead of realizing it through the traditional multi-way switch time-sharing selection method, so that a purely distributed battery management structure is fully realized. The voltage of the lithium iron phosphate battery is directly drawn from the two ends of the single cell, and then divided by two high-precision resistors. The voltage obtained by the voltage division is introduced into the A/D analog signal conversion channel inside the Dspic30f4012 chip for voltage detection. The A/D converter in the Dspic30f4012 chip has a 10-bit accuracy and a reference voltage of 2.5 V, so the voltage detection module can detect a voltage range of 0~5 V, which is greater than the maximum voltage of the single cell of 3.65 V. The total voltage of the battery pack is detected, and the battery pack voltage is collected by connecting to the A/D conversion channel in the Dspic30f4012 chip through the signal attenuation circuit and the anti-common mode voltage circuit.

The detection of single battery current is realized by Hall sensor, which can output voltage signal up to 3 V, and can be directly connected to the A/D sampling channel in Dspic30f4012 chip; the detection of battery temperature is realized by TJ1047 temperature detection chip, which outputs 0.5 V and 1.75 V at -40 ℃ and 125 ℃ respectively, and has temperature-voltage proportional characteristic of 10 mV/℃ and error of ±0.5 ℃. Therefore, the voltage output from TJ1047 chip can be directly connected to A/D conversion channel in Dspic30f4012 chip, so as to complete the acquisition of battery temperature and ambient temperature.

2.1.2 LIN Communication Interface Design

In modern automobiles, bus technology is increasingly being used, and CAN/LIN networks have become the mainstream development direction of distributed-based on-board electronic networks. As a high-speed transmission bus, CAN bus has the outstanding characteristics of high speed, high bandwidth, and multiple functions, but its cost is relatively expensive; LIN bus is a low-end bus, but it has outstanding advantages in reducing costs and is suitable for data transmission that does not require high network speed and real-time performance. Therefore, in situations where the bandwidth and speed of CAN bus are not required, LIN bus supplements the existing bus technology of the automotive multiplexing network guided by CAN bus. Battery temperature, current, and voltage detection do not require extremely high real-time performance and bus speed, so LIN bus can well meet the requirements of battery management system.

The Dspic30f4012 chip does not have a LIN bus interface, but has an SCI communication interface. This article uses the TPIC1021 chip as the chip for SCI and LIN bus conversion, as shown in Figure 3. After the SCI communication pins U1RX and U1TX are electrically isolated by the magnetic coupling isolation device, they are connected to the LIN driver's LIN_RXD and LIN_TXD respectively. After conversion, the LIN bus signal is finally output on the LIN pin. A magnetic coupling isolation device ADUM1201ARZ is added between the bottom controller Dspic30f4012 and the LIN transceiver TPIC1021 to improve the anti-interference ability of the battery pack detection system communication and solve the problem of short circuit caused by "common ground" in distributed detection, effectively isolating the electrical connections of each detection unit, and also isolating the bottom voltage from the upper LIN bus. When the LIN transceiver is used as a host node, the J3 jumper in Figure 3 needs to be shorted with a jumper pin. When used as a slave node, do not short the jumper pin.

2.2 Battery Management System Software Design

2.2.1 Software design and overall structure of battery management system

The software design in ECU includes the software design of the bottom ECU and the upper ECU. The software design of the bottom ECU mainly includes the acquisition program of voltage, current and temperature and the calculation program of the acquisition results, data communication program, interrupt program, etc.; the software design of the upper ECU mainly includes the SOC estimation program, LIN bus communication program, fault analysis and alarm program, display program of voltage, current, temperature and charge state, clock program, interrupt program, etc. The whole program design is realized by structured and modular programming method. The main program flow chart of the upper ECU is shown in Figure 4.

Among them, the battery voltage detection includes the detection of single cell voltage and the detection of battery pack voltage. When the single cell voltage exceeds the limit, the system can determine the number of the single cell that exceeds the limit, determine whether the single cell voltage exceeds the limit or the high voltage exceeds the limit, and display it on the display and sound an alarm. When the battery pack voltage exceeds the limit, the program can analyze the reason for the limit and enter the protection program. The battery temperature detection includes the detection of single cell temperature and the detection of ambient temperature. When the temperature exceeds the limit, the system can analyze the reason for the temperature exceeding the limit through the detected data and enter the protection program. The battery state of charge exceeding the limit mainly means that the remaining battery power is too low, and continued discharge may affect the battery life.

2.2.2 Implementation of LIN Communication

The LIN protocol is an open bus protocol. A complete message frame consists of a message header and a response. Each data transmission starts from the host node, marking the beginning of a data communication process message frame [3].

Figure 5 shows the LIN bus identifier field of the No. 5 single-cell lithium iron phosphate battery, which is used as an example to illustrate the setting of the LIN bus identifier field. The ID bit of the No. 5 single-cell battery is 0101, so the ID of this single-cell battery is 0x5, and ID4 and ID5 are set to 01, that is, the data field bytes sent are set to 4 bytes, and the parity check values ​​obtained through the previous parity check are 0 and 1, as shown in Figure 5.

Due to the different ranges of each signal, the number of data bits used for voltage, current, and temperature signals is also different. The voltage range is within 0~5 V, the current is within 0~20 A, and the temperature is within -40~125 ℃. Therefore, this paper uses the first byte and the lower two bits of the fourth byte in the data field, a total of 10 bits to represent the voltage; the second byte and the middle 4 bits of the fourth byte, a total of 12 bits to represent the current; the third byte and the upper two bits of the fourth byte, a total of 10 bits to represent the temperature. Since the voltage, current, and temperature are accurate to the decimal point, it is more complicated to represent decimals in the data field. This paper uses 10 or 100 times the actual parameter value in the data frame, as shown in Figure 6.

Table 1 is the ID resource allocation table of the LIN bus node corresponding to each single battery.

The upper-layer ECU acts as the host node of the LIN bus. When the LIN host node requests data from the single-cell slave node, data will be transmitted from the slave node to the host node on the LIN bus. At this time, the LIN host node sends a message frame header to the bus. After receiving the message frame header, the LIN slave node on the bus determines whether it matches its own ID. If it matches, it sends a message frame response. The LIN host node receives the message frame response and completes the data request of the host node.

2.2.3 Battery SOC estimation and operation control strategy

When estimating SOC, an accurate and appropriate model is very necessary. For the Kalman filter algorithm, accurate SOC estimation is based on an accurate battery model. The Thevenin model is currently a relatively accurate model. The model describes the external characteristics of the battery by connecting the battery electromotive force, a pure resistor and a capacitive resistor circuit in series. The mathematical relationship of its electrical model is as follows:

In formula (1), k is the k moment, E(k) is the battery terminal voltage, V(k) is the battery electromotive force, R1 is the battery's ohmic internal resistance, R2 is the battery's polarization internal resistance, Uc is the battery's polarization voltage, and the capacitor R2C loop is used to simulate the dynamic characteristics of the battery during polarization. Considering the influence of temperature, the battery's electromotive force and state of charge have the relationship (3):

Where: F [Soc (k)] is the functional relationship between the battery and the electromotive force, Soc (k) represents the change in the electromotive force of the battery at different temperatures relative to the reference conditions. Through the above formula, the state space equation is obtained after discretization as follows.

The state space equations accurately give the system-related coefficient matrices A(K), B(K), C(K), D(K) and constant matrices W(K), V(K). Based on the above equations and related matrices, the extended Kalman filter estimation formula can be obtained.

The extended Kalman filter algorithm consists of two parts: filter calculation and filter gain calculation. The filter calculation is completed by equations (6) to (8). At time k, the predicted value of SOC is obtained by equation (7) using the filtering result at time (k-1). Then, the predicted value of the state variable V (K) at time k is obtained according to the state space equation (6). The prediction error is obtained by comparing it with the actual measured value. Then, the predicted value of the state variable is corrected according to equation (8) to obtain a new filtering result. The filter gain calculation is completed by equations (9) to (11), where Q and R are the variance matrices of the noise W (k) and V (k), respectively.

3 Experimental results analysis

The circuit board of the bottom ECU of this design is shown in Figure 7. A circuit board of the bottom ECU is fixed on each single battery.

The working condition of the battery management system is tested under different charging strategies. The detection accuracy of the system is demonstrated by detecting the charging and discharging voltage, current, temperature, SOC and other parameters of each single battery in the battery pack and comparing them with the actual values, as shown in Figure 8. The data is recorded once per minute, and the horizontal axis is time min.

This design sets the upper voltage limit of 3.65 V and the lower voltage limit of 2.95 V during charging and discharging. The upper limit of temperature alarm is 80 ℃. The experiment charges the battery, and the final charging voltage is 3.53~3.62 V. The maximum deviation of the charging process is 50 mV, and the battery voltage error is less than 1%; in addition, the temperature measurement error meets the 1% requirement, the current measurement accuracy is 1%, and the SOC error is within 8%. When the single cell is artificially over-voltage, the system can alarm and display in time. The experiment shows that this battery management system can achieve the expected goal of battery parameter detection and can meet the accuracy requirements.

4 Conclusion

This paper designs and develops a lithium iron phosphate battery management system. Based on the distributed method to detect the parameters of each single battery, the LIN bus technology is introduced to further reduce the cost of the system. This system realizes the functions of real-time battery monitoring and protection, SOC estimation, LIN bus communication, etc. The system has a simple structure, high measurement accuracy, and can effectively protect the battery pack. The LIN bus replaces the commonly used CAN or RS232 communication, which provides an important basis for the design of a new electric vehicle battery management system.