Battery pack equalization method for improving performance and prolonging service life

Abstract

The invention relates to the technical field of batteries, in particular to a battery pack equalization method for improving the performance and prolonging the service life. Comprising the following steps: predicting the real capacity of a battery cell by adopting a neural network algorithm, fitting SOC and OCV function mapping according to open-circuit voltage method data on the basis of outputting the real capacity of the battery cell, mapping the SOC of each battery cell, and dynamically adjusting and balancing in a full life cycle; compared with the prior art that direct prediction of the SOC value lacks detailed understanding of the internal state of the battery, the method for predicting the SOC value based on the real capacity firstly estimates the current capacity of the battery and then calculates the SOC according to the capacity, so that the accuracy of SOC value measurement is improved; moreover, accumulative errors are corrected more accurately, capacity estimation is adjusted dynamically, the accumulative errors are reduced, and aging of the battery can be better adapted by dynamically estimating the current capacity of the battery.

Description

Technical Field

[0001] The present invention relates to the field of battery technology, and in particular to a battery pack balancing method for improving performance and life. Background Art

[0002] The principle of battery balancing is to use power electronics technology to keep the voltage deviation of lithium-ion battery cells or battery packs within the expected range, thereby ensuring that each single cell remains in the same state during normal use to avoid overcharging or over-discharging. Battery balancing is generally divided into two types: active balancing and passive balancing.

[0003] Active balancing is to transfer energy between cells with different charge states in the battery pack to make them reach the same charge state. Energy transfer can be achieved in various ways, such as using devices such as resistors, switches or converters to consume or release energy. When the charge state of a cell exceeds or falls below that of other cells, the balancing system will transfer energy from the cell with a higher charge state to the cell with a lower charge state to balance the charge state of the entire battery pack.

[0004] Passive balancing is to reduce the voltage of the battery cell with higher voltage by consuming energy. Specifically, when the battery pack is charging, a battery cell with higher voltage is detected and a resistor is connected in parallel to it to reduce the voltage of the battery cell by consuming part of the energy.

[0005] After using the battery cells for a period of time, the capacity of the battery cells will be different. Of course, since the heating conditions of each battery cell are different, the internal resistance of each battery cell is also different, which accelerates the gap between the battery cells. When using the balancing method to make up for this gap, the capacity of the battery is like the barrel principle, which is based on the worst battery cell. If it is not balanced, the battery will soon be useless;

[0006] However, in the prior art, when calculating SOC (the state of charge of a car battery, also known as the remaining capacity, which indicates the battery's ability to continue working), the estimation accuracy is limited, resulting in errors during balancing, which affects the life of the battery pack. In view of this, we propose a battery pack balancing method that improves performance and life. Summary of the invention

[0007] The object of the present invention is to provide a battery pack balancing method for improving performance and life, so as to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides a battery pack balancing method for improving performance and life, comprising the following steps:

[0009] S1. Use neural network algorithm to establish capacity estimation model based on the real-time data of voltage, current and temperature of battery cells under different conditions in historical data. Use QC software algorithm to monitor the voltage, current and temperature of each battery cell, and input them into the capacity estimation model to predict the actual capacity of the battery cell.

[0010] S2. Based on the actual capacity of the output battery cell, the open circuit voltage method data is used to fit the SOC (state of charge) and OCV (open circuit voltage) function mapping, and the voltage, current and temperature of each battery cell are monitored again through the QC software algorithm to map the SOC of each battery cell;

[0011] S3. Calculate the deviation between the SOC of each battery cell and the average power, and dynamically adjust the balancing strategy throughout the life cycle. For batteries with SOC higher than the average, consume excess energy through resistive discharge. For batteries with SOC lower than the average, obtain energy from high SOC batteries through energy transfer.

[0012] As a further improvement of the technical solution, the establishment of the capacity estimation model in S1 includes the following steps:

[0013] A large amount of voltage, current and temperature data of battery cells at different temperatures, charge and discharge rates and aging degrees are collected as training sets; a neural network algorithm is used, and the neural network model is set to be f. The voltage V, current I, temperature T and aging degree A of the battery cell are input, and the output is the actual capacity of the battery cell C = f(V, I, T, A).

[0014] As a further improvement of the technical solution, the open circuit voltage method in S2 comprises the following steps:

[0015] Select a battery pack with multiple cells of the same model and specification, perform multiple charge and discharge cycles on the battery, and record the OCV and corresponding SOC after each rest, where the SOC is calculated by the accumulated power during the charge and discharge process;

[0016] Abnormal data points were removed and the SOC values ​​were normalized to the range of 0 to 1. According to the changing trends of OCV and SOC, the data were divided into several segments, and a different fitting method was used for each segment to establish the SOC and OCV function mapping.

[0017] As a further improvement of the technical solution, the QC software algorithm is used to monitor and control the state of the battery to ensure that the charge state of each battery cell is accurate, including the following steps:

[0018] Data collection: Use sensors to obtain the voltage, current and temperature of the battery cell. The sensors include voltage sensors to collect the voltage of each battery cell, current sensors to collect the battery charge and discharge current, and temperature sensors to collect the temperature of each battery cell.

[0019] Monitor battery status: monitor the voltage of each battery cell, monitor the charge and discharge current of the battery, and monitor the temperature of each battery cell to ensure that they are within a safe range. When an abnormal situation is detected, trigger an alarm and take measures such as stopping charging and reducing the charge and discharge current. Transmit data to the main control unit through the communication interface and send control commands based on the monitoring results.

[0020] As a further improvement of the technical solution, the open circuit voltage method also includes a waiting time adjustment algorithm, including the following steps:

[0021] The current sensor is used to determine whether the battery is charging or discharging. If the current is 0 or close to 0, the output battery is in a static state. When the battery enters the static state, the timer is started to record the static time. If the battery starts to charge and discharge again during the static period, the timer is reset. When the static time reaches the set value, the battery voltage is collected. The voltage at this time is defined as the actual OCV. The current temperature is collected through the temperature sensor, the OCV-SOC curve is adjusted, and the pre-calibrated temperature-OCV-SOC function is used for compensation. At the same time, the OCV is converted to SOC, and the SOC value converted by OCV is compared with the current SOC estimation value for calibration.

[0022] As a further improvement of the technical solution, the dynamic adjustment and balancing strategy of the life cycle includes the following postures:

[0023] Charging starts: At the beginning of charging, the neural network model is used to predict the initial SOC of each battery cell, and the charging current is adjusted according to the prediction results to ensure that all batteries are charged at the same speed;

[0024] During charging: monitor the voltage and temperature of each battery cell in real time, and adjust the charging strategy through the QC software algorithm to prevent overcharging;

[0025] When charging is almost full: When the battery is nearly fully charged, the neural network model is used to predict the final SOC of each cell and fine-tune it to ensure that the SOC of all cells is consistent;

[0026] During discharge: monitor the discharge status of each battery cell in real time, and adjust the discharge strategy through the QC software algorithm to prevent over-discharge;

[0027] When the battery is almost discharged: When the battery is almost discharged, the neural network model is used to predict the remaining capacity of each cell and make fine adjustments to ensure that the discharge status of all cells is consistent;

[0028] In static state: When the battery is static, the neural network model is used to predict the self-discharge of each cell and perform necessary balancing operations.

[0029] As a further improvement of the technical solution, the calculation of the deviation between the SOC of each battery cell and the average power in S3 includes the following steps:

[0030] Calculate the average SOC of all cells, the expression is:

[0031] Among them, SOC avg is the average SOC, N is the total number of cells, SOC i is the SOC of the ith cell;

[0032] Calculate the deviation of each cell’s SOC from the average SOC, ΔSOC i =SOC i -SOC avg , where ΔSOC i is the SOC deviation of the i-th cell.

[0033] As a further improvement of the technical solution, in S3, for the battery cells whose SOC is higher than the average value, the excess energy is consumed by resistor discharge, and the expression is:

[0034] I discharge，i = k * ΔSOC i , where I discharge，i is the discharge current of the ith cell, k is the proportionality factor, which is adjusted according to the battery characteristics and requirements;

[0035] For cells with SOC lower than the average value, energy is obtained from high SOC cells through energy transfer. The expression is:

[0036] I transfer，i = -k*ΔSOC i , where I transfer，i is the energy transfer current of the ith cell.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] In the battery pack balancing method for improving performance and life, a neural network algorithm is used to predict the real capacity of the battery cell, and based on the real capacity of the output battery cell, the SOC and OCV function mapping is fitted according to the open circuit voltage method data to map the SOC of each battery cell; compared with the prior art that directly predicts the SOC value but lacks a detailed understanding of the internal state of the battery, the method of predicting the SOC value based on the real capacity first estimates the current capacity of the battery, and then calculates the SOC based on the capacity, thereby improving the accuracy of the SOC value measurement;

[0039] Moreover, errors are prone to accumulation during long-term operation, especially when the battery ages or the environment changes, the error will gradually increase. By estimating the current capacity of the battery first, the accumulated error can be corrected more accurately, the capacity estimation can be adjusted dynamically, and the accumulated error can be reduced. By dynamically estimating the current capacity of the battery, the aging of the battery can be better adapted. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is the overall flow chart of the present invention;

[0041] Figure 2 It is a schematic diagram of the OCV-SOC curve of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] There are two types of balancing in the market: passive balancing and active balancing. However, after using the battery for a period of time, the capacity of the battery will be different. Of course, since the heating conditions of each battery are different, the internal resistance of each battery is also different, which accelerates the gap between the batteries. Therefore, the balancing method should be used to make up for this gap. The capacity of the battery is like the principle of a wooden barrel, which is based on the worst battery. If it is not balanced, the battery will soon be useless.

[0044] And, if Figure 2 From the OCV-SOC curve, only the curves at the two ends are steeper, and the curve in the middle is flatter. It is not easy to find the difference between the cells. In addition, the heat generated by the cell itself will affect the change of internal resistance, and the voltage is even more difficult to truly reflect the real capacity of the cell. Therefore, using the same voltage method plus the actual charging capacity of each cell can accurately reflect the difference between the cells. In this way, the measurement is accurate and the balancing effect is better. It is equivalent to measuring in two dimensions: voltage to determine the large range / scale, and capacity to accurately measure. The present invention combines battery hardware balancing with battery power calculation, and uses the real capacity of the cell to measure to ensure the accuracy of balancing. Please refer to Figure 1 As shown, this embodiment provides a battery pack balancing method for improving performance and life, comprising the following steps:

[0045] S1. Use neural network algorithm to establish capacity estimation model based on the real-time data of voltage, current and temperature of battery cells under different conditions in historical data. Use QC software algorithm to monitor the voltage, current and temperature of each battery cell, and input them into the capacity estimation model to predict the real capacity of the battery cell. When training the model, use a large amount of experimental data, including battery cell behavior under different temperatures, charge and discharge rates, aging degree and other conditions, to ensure the generalization ability and accuracy of the model.

[0046] S2. Based on the actual capacity of the output battery cell, the open circuit voltage method data is used to fit the SOC (state of charge) and OCV (open circuit voltage) function mapping, and the voltage, current and temperature of each battery cell are monitored again through the QC software algorithm to map the SOC of each battery cell;

[0047] Compared with the existing technology that directly predicts the SOC value, which lacks a detailed understanding of the internal state of the battery, the method of predicting the SOC value based on the real capacity first estimates the current capacity of the battery, and then calculates the SOC based on the capacity, thereby improving the accuracy of the SOC value measurement;

[0048] Moreover, errors tend to accumulate over long periods of operation, especially as the battery ages or the environment changes, the errors will gradually increase. By estimating the current capacity of the battery first, the accumulated errors can be corrected more accurately, and the capacity estimation can be adjusted dynamically to reduce the accumulated errors. By dynamically estimating the current capacity of the battery, the aging of the battery can be better adapted.

[0049] Specifically, establishing the capacity estimation model in S1 includes the following steps:

[0050] A large amount of voltage, current and temperature data of battery cells at different temperatures, charge and discharge rates and aging degrees are collected as training sets; a neural network algorithm is used, and the neural network model is set to be f, and the voltage V, current I, temperature T and aging degree A of the battery cell are input, and the output is the true capacity C=f(V, I, T, A) of the battery cell. The neural network learns the capacity change law of the battery cell under different conditions through nonlinear transformation of multiple layers of neurons, and ensures the generalization ability and accuracy of the model through cross-validation and other methods. Assuming that the voltage of a battery cell at 25°C, 1C charge and discharge rate, 50% aging degree is 3.7V, the current is 1A, and the temperature is 25°C, the neural network model predicts its true capacity to be 2.5Ah.

[0051] Further, the open circuit voltage method in S2 includes the following steps:

[0052] Select a battery pack with multiple cells of the same model and specification, perform multiple charge and discharge cycles on the battery, and record the OCV and corresponding SOC after each rest, where the SOC is calculated by the accumulated power during the charge and discharge process;

[0053] Remove abnormal data points, such as those caused by measurement errors or battery failures, and normalize the SOC value to the range of 0 to 1 to facilitate subsequent mathematical processing. According to the changing trends of OCV and SOC, divide the data into several segments, and use different fitting methods for each segment to establish the SOC and OCV function mapping;

[0054] The fitting methods include the following:

[0055] Linear fitting: For some battery types, the relationship between OCV and SOC can be approximated as a linear relationship in certain intervals, and the least squares method is used for linear fitting;

[0056] Polynomial fitting: For nonlinear relationships, use polynomial fitting, and the commonly used polynomial order is 2 or 3;

[0057] Spline interpolation: For complex relationships, you can use spline interpolation methods, such as cubic spline interpolation, which provides a smoother fitting curve;

[0058] Fit each segment of data and select a suitable fitting method to improve the accuracy of the SOC and OCV function mapping. In addition, use part of the data for verification to ensure the accuracy of the fitting curve and calculate the fitting error, such as the root mean square error, to obtain the functional relationship between OCV and SOC, thereby accurately estimating the battery state of charge in practical applications.

[0059] It is worth noting that the QC software algorithm is used to monitor and control the state of the battery to ensure that the state of charge of each battery cell is accurate, including the following steps:

[0060] Data collection: Use sensors to obtain the voltage, current and temperature of the battery cell. The sensors include voltage sensors to collect the voltage of each battery cell, current sensors to collect the charge and discharge current of the battery, and temperature sensors to collect the temperature of each battery cell. At the same time, use filtering algorithms (such as low-pass filtering, sliding average filtering, etc.) to remove noise and improve data accuracy. Calibrate sensor data to ensure data accuracy. For example, voltage sensors may need to be temperature compensated to detect abnormal values ​​in the data, such as voltage mutations, high or low temperatures, etc., to process the collected data.

[0061] Monitor battery status: monitor the voltage of each battery cell, for example, the voltage of a lithium-ion battery is usually between 3.0V and 4.2V, monitor the charge and discharge current of the battery, for example, to avoid overcharging or over-discharging, and monitor the temperature of each battery cell, for example, the temperature of a lithium-ion battery is usually between -20℃ and 60℃, all to ensure that it is within a safe range, and when an abnormal situation is detected, trigger an alarm and take measures to stop charging and reduce the charge and discharge current, transmit data to the main control unit through a communication interface (such as CAN, I2C, SPI, etc.), and send control commands based on the monitoring results, such as adjusting the charging current, starting the cooling system, etc., to achieve real-time monitoring of the voltage, current and temperature of each battery cell to ensure safe operation and optimized performance of the battery.

[0062] Among them, lithium-ion batteries have a polarization effect. The impact of the polarization effect is that the voltage of the battery is not constant for a period of time after the charging and discharging stops, but changes slowly. The voltage obtained after about 1 to 2 hours of standing is the real open circuit voltage under the current SOC. In addition, lithium-ion batteries have another significant feature, that is, their characteristics are particularly susceptible to the influence of ambient temperature. The most preferred temperature range is about -20℃~50℃ (similar to the temperature range that humans can adapt to). At different temperatures, its discharge capacity, open circuit voltage, internal resistance and other parameters are all different. Therefore, the open circuit voltage method also includes a waiting time adjustment algorithm, including the following steps:

[0063] The current sensor is used to determine whether the battery is charging or discharging. If the current is 0 or close to 0, the output battery is in a static state. When the battery enters the static state, the timer is started to record the static time. The static time is set to 1-2 hours. If the battery starts to charge and discharge again during the static period, the timer is reset. When the static time reaches the set value, the battery voltage is collected. The voltage at this time is defined as the actual OCV. The current temperature is collected through the temperature sensor, the OCV-SOC curve is adjusted, and the pre-calibrated temperature-OCV-SOC function is used for compensation. At the same time, the OCV is converted to SOC, and the SOC value converted by OCV is compared with the current SOC estimation value for calibration. During the vehicle parking period, the battery is in a static state. When the vehicle is started the next morning, an accurate SOC value can be obtained by the OCV method for calibrating the SOC estimation value in the BMS. In the energy storage system, the battery may not be charged or discharged for a long time. By setting the static time, the SOC can be calibrated regularly to eliminate the influence of the polarization effect and improve the accuracy of the SOC estimation. The influence of temperature on battery performance is considered to improve the reliability of the SOC estimation.

[0064] S3. Calculate the deviation between the SOC of each battery cell and the average power, and dynamically adjust the balancing strategy throughout the life cycle. For batteries with SOC higher than the average, consume excess energy through resistance discharge. For batteries with SOC lower than the average, obtain energy from high SOC batteries through energy transfer. By comparing the power of each battery cell, determine the battery cell that needs to be balanced. When implementing the balancing operation, adjust the current flow between the batteries to transfer part of the power of the battery cell with higher power to the battery cell with lower power, thereby achieving power balance.

[0065] Obtaining energy from high SOC cells through energy transfer means transferring electrical energy from cells with SOC higher than the average to cells with SOC lower than the average, so as to achieve power balance.

[0066] Among them, the dynamic adjustment and balancing strategy of the life cycle includes the following postures:

[0067] Charging starts: At the beginning of charging, the neural network model is used to predict the initial SOC of each battery cell, and the charging current is adjusted according to the prediction results to ensure that all batteries are charged at the same speed;

[0068] During charging: monitor the voltage and temperature of each battery cell in real time, and adjust the charging strategy through the QC software algorithm to prevent overcharging;

[0069] When charging is almost full: When the battery is nearly fully charged, the neural network model is used to predict the final SOC of each cell and fine-tune it to ensure that the SOC of all cells is consistent;

[0070] During discharge: monitor the discharge status of each battery cell in real time, and adjust the discharge strategy through the QC software algorithm to prevent over-discharge;

[0071] When the battery is almost discharged: When the battery is almost discharged, the neural network model is used to predict the remaining capacity of each cell and make fine adjustments to ensure that the discharge status of all cells is consistent;

[0072] In static state: When the battery is static, the neural network model is used to predict the self-discharge of each cell and perform necessary balancing operations.

[0073] In addition, the deviation between the SOC of each battery cell and the average power is calculated in S3, including the following steps:

[0074] Calculate the average SOC of all cells, the expression is:

[0075] Among them, SOC avg is the average SOC, N is the total number of cells, SOC i is the SOC of the ith cell;

[0076] Calculate the deviation of each cell’s SOC from the average SOC, ΔSOC i =SOC i -SOC avg , where ΔSOC i is the SOC deviation of the i-th cell.

[0077] In S3, for cells with SOC higher than the average value, excess energy is consumed by resistor discharge, and the expression is:

[0078] I discharge，i = k * ΔSOC i , where I discharge，i is the discharge current of the ith cell, k is the proportionality factor, which is adjusted according to the battery characteristics and requirements;

[0079] For cells with SOC lower than the average value, energy is obtained from high SOC cells through energy transfer. The expression is:

[0080] I transfer，i = -k*ΔSOC i , where I transfer，i It is the energy transfer current of the ith battery cell. The battery balancing principle is to use power electronics technology to keep the voltage deviation of lithium-ion battery cells or battery pack voltage within the expected range, thereby ensuring that each single battery cell maintains the same state during normal use to avoid overcharging and over-discharging, achieve energy balance between the cells inside the battery pack, and improve the performance and life of the battery pack.

[0081] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A battery pack balancing method for improving performance and life, characterized in that: The following steps are involved: S1. Use neural network algorithm to establish capacity estimation model based on the real-time data of voltage, current and temperature of battery cells under different conditions in historical data. Use QC software algorithm to monitor the voltage, current and temperature of each battery cell, and input them into the capacity estimation model to predict the actual capacity of the battery cell. S2. Based on the actual capacity of the output battery cell, the open circuit voltage method data is used to fit the SOC and OCV function mapping, and the voltage, current and temperature of each battery cell are monitored again through the QC software algorithm to map the SOC of each battery cell; S3. Calculate the deviation between the SOC of each battery cell and the average power, and dynamically adjust the balancing strategy throughout the life cycle. For batteries with SOC higher than the average, consume excess energy through resistive discharge. For batteries with SOC lower than the average, obtain energy from high SOC batteries through energy transfer.

2. The battery pack equalization method for improving performance and life according to claim 1, characterized in that: The establishment of the capacity estimation model in S1 includes the following steps: A large amount of voltage, current and temperature data of battery cells at different temperatures, charge and discharge rates and aging degrees are collected as training sets; a neural network algorithm is used, and the neural network model is set to be f. The voltage V, current I, temperature T and aging degree A of the battery cell are input, and the output is the actual capacity of the battery cell C = f(V, I, T, A).

3. The battery pack balancing method for improving performance and life according to claim 2, characterized in that: The open circuit voltage method in S2 comprises the following steps: Select a battery pack with multiple cells of the same model and specification, perform multiple charge and discharge cycles on the battery, and record the OCV and corresponding SOC after each rest, where the SOC is calculated by the accumulated power during the charge and discharge process; Abnormal data points were removed and the SOC values ​​were normalized to the range of 0 to 1. According to the changing trends of OCV and SOC, the data were divided into several segments, and a different fitting method was used for each segment to establish the SOC and OCV function mapping.

4. The battery pack balancing method for improving performance and life according to claim 3, characterized in that: The QC software algorithm is used to monitor and control the state of the battery to ensure that the state of charge of each battery cell is accurate, including the following steps: Data collection: Use sensors to obtain the voltage, current and temperature of the battery cell. The sensors include voltage sensors to collect the voltage of each battery cell, current sensors to collect the battery charge and discharge current, and temperature sensors to collect the temperature of each battery cell. Monitor battery status: monitor the voltage of each battery cell, monitor the charge and discharge current of the battery, and monitor the temperature of each battery cell to ensure that they are within a safe range. When an abnormal situation is detected, trigger an alarm and take measures such as stopping charging and reducing the charge and discharge current. Transmit data to the main control unit through the communication interface and send control commands based on the monitoring results.

5. The battery pack balancing method for improving performance and life according to claim 4, characterized in that: The open circuit voltage method also includes a waiting time adjustment algorithm, including the following steps: The current sensor is used to determine whether the battery is charging or discharging. If the current is 0 or close to 0, the output battery is in a static state. When the battery enters the static state, the timer is started to record the static time. If the battery starts to charge and discharge again during the static period, the timer is reset. When the static time reaches the set value, the battery voltage is collected. The voltage at this time is defined as the actual OCV. The current temperature is collected through the temperature sensor, the OCV-SOC curve is adjusted, and the pre-calibrated temperature-OCV-SOC function is used for compensation. At the same time, the OCV is converted to SOC, and the SOC value converted by OCV is compared with the current SOC estimation value for calibration.

6. The battery pack balancing method for improving performance and life according to claim 5, characterized in that: The dynamic adjustment and balancing strategy of the life cycle includes the following postures: Charging starts: At the beginning of charging, the neural network model is used to predict the initial SOC of each battery cell, and the charging current is adjusted according to the prediction results to ensure that all batteries are charged at the same speed; During charging: monitor the voltage and temperature of each battery cell in real time, and adjust the charging strategy through the QC software algorithm to prevent overcharging; When charging is almost full: When the battery is nearly fully charged, the neural network model is used to predict the final SOC of each cell and fine-tune it to ensure that the SOC of all cells is consistent; During discharge: monitor the discharge status of each battery cell in real time, and adjust the discharge strategy through the QC software algorithm to prevent over-discharge; When the battery is almost discharged: When the battery is almost discharged, the neural network model is used to predict the remaining capacity of each cell and make fine adjustments to ensure that the discharge status of all cells is consistent; In static state: When the battery is static, the neural network model is used to predict the self-discharge of each cell and perform necessary balancing operations.

7. The battery pack balancing method for improving performance and life according to claim 6, characterized in that: Calculating the deviation between the SOC of each battery cell and the average power in S3 includes the following steps: Calculate the average SOC of all cells, the expression is: Among them, SOC avg is the average SOC, N is the total number of cells, SOC i is the SOC of the ith cell; Calculate the deviation of each cell’s SOC from the average SOC, ΔSOC i =SOC i -SOC avg , where ΔSOC i is the SOC deviation of the i-th cell.

8. The battery pack balancing method for improving performance and life according to claim 7, characterized in that: In S3, for cells with SOC higher than the average value, excess energy is consumed by resistor discharge, and the expression is: I discharge，i = k * ΔSOC i , where I discharge，i is the discharge current of the ith cell, k is the proportionality factor, which is adjusted according to the battery characteristics and requirements; For cells with SOC lower than the average value, energy is obtained from high SOC cells through energy transfer. The expression is: I transfer，i =-k*ΔSOC i , where I transfer，i is the energy transfer current of the ith cell.

Images