A multi-level SOC cooperative multi-string battery pack active equalization control method and system

Abstract

The application discloses a kind of multi-level SOC collaborative multi-string battery pack active equalization control method and system, the method includes the following steps: first, the terminal voltage, temperature and charge-discharge current of each single body in battery pack are collected in real time;Then, the improved double extended Kalman filter algorithm is used to estimate the state of charge of each single body, and the global SOC standard deviation and global SOC range are calculated;Determine the current equalization level through multi-threshold hierarchical trigger mechanism, and the equalization level at least includes group equalization level and full topology collaborative equalization level;According to the determined equalization level, energy transfer is carried out through group internal flyback converter topology and / or inter-group multi-winding transformer topology;Under dynamic working condition, predict the trend of state of charge change and start equalization action in advance.The application can solve the problems of inaccurate equalization criterion, insufficient multi-string collaboration and poor adaptability to dynamic working condition in the prior art, and has the advantages of high equalization precision, high speed, high energy efficiency and good scalability.

Description

Technical Field

[0001] This invention relates to the field of battery balancing technology, and in particular to a multi-level SOC-coordinated active balancing control method and system for multi-string battery packs. Background Technology

[0002] In applications such as electric vehicles and energy storage systems, lithium-ion batteries are typically configured in series to form battery packs, meeting the demands for high voltage and large capacity. However, due to inherent differences in manufacturing processes, material consistency, operating temperature, and self-discharge rates among individual battery cells, the state of charge (SOC) of each cell gradually deviates during charge-discharge cycles. This inconsistency triggers the "weakest link" effect, limiting the usable capacity of the battery pack to the cell with the lowest SOC, resulting in capacity loss. Simultaneously, cells with higher SOCs are prone to overcharging towards the end of the charging phase, while cells with lower SOCs are prone to over-discharging towards the end of the discharging phase, severely impacting battery safety and lifespan. Furthermore, this inconsistency accelerates the overall aging and degradation of the battery pack, increasing the probability of safety incidents such as thermal runaway. Therefore, battery balancing technology is a crucial component of the Battery Management System (BMS).

[0003] Based on different energy transfer methods, existing battery balancing technologies are mainly divided into two categories: passive balancing and active balancing. Passive balancing dissipates excess energy from high-SOC cells through resistors. It features simple circuitry and low cost, but low energy utilization, with a balancing current typically only 50–200 mA. It is suitable for small-scale balancing at the end of charging and significantly increases the thermal management burden due to its high energy consumption. Active balancing, on the other hand, transfers energy from high-SOC cells to low-SOC cells or redistributes it from the battery pack as a whole through energy conversion circuits. It boasts high energy utilization and a balancing current reaching hundreds of milliamps to several amperes, enabling balancing throughout the entire charging and discharging process. Common active balancing topologies include switched capacitor methods, flying capacitor methods, transformer methods (such as flyback converters, forward converters, LLC resonant converters, etc.), and buck-boost converter methods.

[0004] In recent years, numerous patents related to BMS equalization technology have been published both domestically and internationally. For example, CN201820668340.5 proposes an active equalization scheme based on a switched capacitor array, which can achieve energy transfer between adjacent cells. However, the equalization speed is limited by the capacitor charging and discharging time constant, and it only supports energy transfer between adjacent cells, resulting in low equalization efficiency in multi-string scenarios. CN111478410B discloses a bidirectional active equalization circuit using a multi-winding transformer, which supports bidirectional energy transfer between individual cells and the battery pack, with a large equalization current. However, the transformer has a high size and cost, and the control strategy does not fully consider the global optimal equalization problem in multi-string scenarios. Some patents (such as active equalization circuits related to Sungrow Power) use bidirectional DC / DC modules to achieve equalization between battery packs, which is suitable for module-level equalization, but its adaptability to multi-string (such as 16-string, 32-string, and above) equalization scenarios at the individual cell level is insufficient. In addition, CN117669781A attempted to introduce Model Predictive Control (MPC) into the equalization strategy, but the computational complexity was high and it failed to effectively resolve the contradiction between the accuracy of state of charge estimation and the real-time performance of equalization in multi-string scenarios.

[0005] In summary, current BMS active load balancing technology still has the following shortcomings: 1. The balancing strategy is relatively simple. Most schemes use voltage difference as the balancing criterion and fail to make full use of SOC information for accurate decision-making. During the battery plateau period (SOC is about 30%-70%), the voltage change is slow, which leads to a significant decrease in balancing accuracy. 2. There is a lack of global collaborative balancing strategies for large-scale multi-string series battery packs. Existing solutions are mostly designed for two adjacent strings or a small number of series batteries, and their collaborative capabilities are insufficient. 3. There is a contradiction between equalization speed and energy efficiency. Increasing the equalization current in pursuit of rapid equalization will lead to increased energy loss, and there is a lack of adaptive optimization control of the equalization current. 4. Poor adaptability to dynamic operating conditions. Under dynamic operating conditions such as vehicle driving where the current changes rapidly, the battery voltage and SOC fluctuate violently, and the existing balancing strategy is unable to track and make accurate judgments in real time. 5. The balanced topology and control strategy are often designed separately, and the coordinated optimization of hardware topology and control algorithm is not achieved, which limits the improvement of the overall system performance.

[0006] Therefore, how to achieve high-precision active balancing based on SOC, design a hierarchical collaborative balancing strategy suitable for large-scale multi-string scenarios, reduce energy loss while ensuring balancing speed, adapt to dynamic operating conditions, and achieve collaborative optimization of topology and algorithm are technical problems that urgently need to be solved in this field. Summary of the Invention

[0007] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention proposes a multi-level SOC-coordinated active balancing control method for multi-string battery packs, which is capable of...

[0008] The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to the present invention includes the following steps: S1. Real-time acquisition of terminal voltage, temperature and charging / discharging current of each cell in the battery pack; S2. State of Charge Estimation and Situation Assessment: The improved dual extended Kalman filter algorithm is used to estimate the state of charge of each individual cell, and the global state of charge standard deviation and global state of charge range are calculated. S3. Layered triggering: The current equilibrium level is determined through a multi-threshold layered triggering mechanism based on the global standard deviation of the state of charge and the global range of the state of charge. The equilibrium level includes at least the intra-group equilibrium level and the full-topology collaborative equilibrium level. S4. Balanced Execution: Based on the determined balance level, energy transfer is performed through intra-group flyback converter topology and / or inter-group multi-winding transformer topology. S5. Dynamic pre-compensation: Under dynamic operating conditions, predict the trend of charge state change and initiate balancing action in advance.

[0009] In some embodiments of the present invention, a grouping step is further included: The battery cells are evenly divided into There are 3 equilibrium groups, each containing 10 equilibrium groups. Battery series; grouping follows at least one of the following principles: physical adjacency principle, consistency screening principle, thermal coupling principle.

[0010] In some embodiments of the present invention, the improved dual extended Kalman filter algorithm in step S2 includes: Two extended Kalman filters are run simultaneously: the first is used to identify battery model parameters online, and the second is used for state of charge estimation. The battery model uses a second-order RC equivalent circuit model: , , , in, and These are the polarization voltages of the two parallel RC circuits, respectively. The internal resistance of the battery is ohms. and These are the polarization internal resistance and polarization capacitance of the first RC parallel circuit, respectively. and These are the polarization internal resistance and polarization capacitance of the second RC parallel circuit, respectively. The charging and discharging current of the battery is specified, with the discharge direction being positive. This is the open-circuit voltage corresponding to the state of charge (SOC). This refers to the battery terminal voltage. The improvement includes at least one of the following: Temperature compensation corrections are made to the open-circuit voltage, ohmic internal resistance, and RC parameters of the two parallel RC circuits based on the unit temperature T. Based on the terminal voltage residual, current change rate, and residual covariance, the process noise covariance matrix Q and the observation noise covariance matrix R are adjusted online. The SOC increment is calculated using the ampere-hour integration method within a preset short time window, and the SOC increment is then weighted and fused with the extended Kalman filter output for correction.

[0011] In some embodiments of the present invention, step S2 further calculates group-level state of charge characterization quantities: For each equilibrium group, define:

[0012] , ; in, For the first The average state of charge of the group ; The standard deviation of the state of charge within the group; This represents the global average state of charge.

[0013] In some embodiments of the present invention, the multi-threshold hierarchical triggering mechanism in step S3 may include: Level 1 equalization, also known as monitoring-level equalization: When the global state of charge standard deviation is less than 1%, only the battery state is recorded, and equalization action is not initiated. Secondary balancing, also known as intra-group fine-tuning balancing: Intra-group balancing is initiated when the global state of charge standard deviation is greater than or equal to 1% and less than 3%, and the intra-group state of charge standard deviation is greater than or equal to 1.5%. Level 3 equilibrium, namely strong equilibrium within the group: when the global state of charge standard deviation is greater than or equal to 3% and less than 5%, or the global state of charge range is greater than or equal to 4%, the intra-group equilibrium is initiated and supplemented by inter-group auxiliary equilibrium. Level 4 equilibrium, also known as global emergency equilibrium: when the global standard deviation of charge state is greater than or equal to 5%, or the global range of charge state is greater than or equal to 8%, the whole topology collaborative equilibrium is initiated. The trigger judgment period is 100 ms; when the trigger conditions of multiple levels are met at the same time, the higher level of equalization is executed first; a hysteresis interval of 0.5% is set between the start threshold and the exit threshold.

[0014] In some embodiments of the present invention, during the energy transfer performed by the flyback converter topology within the group in step S4, the equalization current... satisfy: , in, The duty cycle of the primary side is the on-state. For magnetizing inductance, The switching frequency; The inter-group multi-winding transformer topology operates in the following two modes: Mode 1: Direct inter-group transfer, transferring energy from a high-charge state group to a low-charge state group; Mode 2: Intra-group redistribution, assisting in intra-group balancing when the difference in state of charge between groups is below a threshold.

[0015] In some embodiments of the present invention, the following path planning strategy is employed in step S4: Intra-group path planning: The maximum difference priority strategy is adopted, and the individual cells in the group are divided into donors and acceptors according to the deviation of the state of charge. A greedy matching algorithm is used to form a balanced energy transfer path. Inter-group path planning: With minimizing the global state of charge variance as the objective function, a constrained optimization problem is established, and a hierarchical solution strategy is adopted: first, at the group level, quadratic programming is used to determine the direction and magnitude of energy transfer between groups, and then at the intra-group level, the allocated energy is further distributed to specific individuals.

[0016] In some embodiments of the present invention, an adaptive current equalization control step is also included: Based on the magnitude of the state-of-charge difference, the battery state-of-charge range, and the temperature T, the magnitude of the balancing current is adaptively adjusted, resulting in the final balancing current. satisfy: , in, This is the current reference value. , The equilibrium rate coefficient, in units of h1, ranges from 0.2 h1 to 0.5 h1. The magnitude of the difference in state of charge; For the state-of-charge interval correction factor: when hour, ;when or hour, ; For temperature correction factor, when hour, ;when hour, ;when hour, When T < 5℃ or T > 55℃, βT = 0, and equilibrium is suspended.

[0017] The maximum allowable current for the equalization circuit.

[0018] In some embodiments of the present invention, the step S5 of predicting the trend of state of charge change and initiating the balancing action in advance includes: State tracking: The dual extended Kalman filter algorithm is used to track and continuously update the estimated state of charge of each individual cell in real time, and the process noise covariance matrix is ​​adaptively adjusted. and observation noise covariance matrix ; Trend prediction: Based on historical trends in state of charge, a linear extrapolation method is used to predict future trends. Changes in state of charge within the time window: ， in, for Time of the first The charge state of an individual cell, For the first Individual in Rate of change of state of charge at time t; prediction window Adjustments based on dynamic operating conditions: Under steady-state operating conditions Under dynamic operating conditions .

[0019] Pre-compensation control: If it is predicted that the state of charge of a certain single cell will be... If the equalization trigger threshold is exceeded, the equalization action of that single unit will be initiated ahead of schedule; pre-compensation intensity coefficient. It is directly proportional to the prediction confidence level: , in, The coefficient of determination for linear extrapolation.

[0020] The present invention also provides a multi-level SOC-coordinated active balancing control system for multi-string battery packs, used to execute the above method, comprising: The sensing layer includes a voltage acquisition circuit, a temperature acquisition circuit, and a current sensor, which are used to acquire the terminal voltage, temperature, and charging / discharging current of each battery cell in real time; among them, the voltage acquisition accuracy reaches ±1mV, and the sampling frequency is not less than 100Hz. The decision-making layer includes a main control MCU and a multi-level SOC collaborative balancing algorithm module running on it. The algorithm module includes an accurate SOC estimation module, a balancing demand assessment module, a balancing path planning module, and a balancing current optimization module. The main control MCU communicates with the sensing layer via a CAN bus or isolated SPI. The execution layer includes a two-level balancing topology that groups to the global level. The two-level balancing topology includes intra-group balancing units and inter-group balancing units. The intra-group balancing units adopt a flyback converter topology to realize energy transfer between individual units within the group. The inter-group balancing units adopt a multi-winding transformer topology to realize energy redistribution between balancing groups. All balancing switches are independently controlled by the main control MCU through PWM signals.

[0021] Compared with the prior art, the multi-level SOC coordinated active balancing control method and system for multi-string battery packs provided in this invention have the following advantages: First, this invention replaces the traditional voltage criterion with an equilibrium criterion based on state of charge (SOC), fundamentally overcoming the problem of inaccurate equilibrium criterion caused by the gradual voltage change during the battery plateau period (SOC 30%-70%). Combined with an improved dual extended Kalman filter (DEKF) algorithm, it achieves an SOC estimation accuracy within ±1%.

[0022] Secondly, this invention employs a two-level balanced topology from grouping to global, coupled with a multi-threshold hierarchical triggering mechanism. This allows for adaptive selection of the balancing mode based on the severity of inconsistencies: fine-tuning within groups is initiated for mild inconsistencies, while full-topology collaborative balancing is activated for severe inconsistencies. The balancing speed is improved by 30%-50% compared to traditional solutions.

[0023] Third, this invention employs an active balancing topology, where the energy conversion efficiency of the intra-group flyback converter and the inter-group multi-winding transformer can reach 85%-92%. Compared to passive balancing schemes that dissipate excess energy through resistors, energy utilization is improved by more than 60%, significantly reducing energy loss during the balancing process. Simultaneously, the adaptive balancing current control mechanism dynamically adjusts the balancing current based on the SOC difference, SOC range, and temperature, avoiding ineffective energy consumption caused by blindly increasing the current.

[0024] Fourth, the grouping strategy and hierarchical optimization algorithm proposed in this invention decompose the large-scale multi-string balancing problem into two levels: intra-group balancing and inter-group balancing. Intra-group path planning employs a greedy algorithm, while inter-group path planning uses quadratic programming, avoiding the direct construction of pairwise global single-unit paths and reducing the solution scale of the optimization problem. This design enables the system to efficiently handle balancing scenarios with 16, 32, and more series-connected batteries.

[0025] Fifth, the "tracking-prediction-pre-compensation" three-stage dynamic balancing strategy proposed in this invention enables the balancing system to adapt to dynamic operating conditions such as vehicle movement. The DEKF algorithm maintains high-precision SOC tracking under dynamic current through adaptive noise adjustment, while the linear extrapolation prediction module predicts and initiates pre-compensation control in advance based on the SOC change trend, temporarily freezing the balancing action under high current conditions. Compared with traditional voltage criterion schemes, this strategy effectively avoids balancing lag and false triggering problems under dynamic conditions.

[0026] Sixth, the grouping architecture of this invention gives the system excellent scalability. When the number of battery strings needs to be increased, only the number of equalization groups needs to be increased (i.e., the flyback converter and multi-winding transformer windings are added), without redesigning the entire equalization system. This modular design reduces system upgrade and maintenance costs and is suitable for battery pack applications with different numbers of strings. Attached Figure Description

[0027] Figure 1 This is the overall architecture diagram of the multi-level SOC collaborative multi-string battery pack active balancing control system of the present invention; Figure 2 This is a schematic diagram of the grouped to global two-stage equalization topology circuit of the present invention; Figure 3 This is a flowchart of the multi-level SOC coordinated active balancing control method for multi-string battery packs according to the present invention; Figure 4 This is a flowchart of the multi-level state of charge estimation algorithm of the present invention; Figure 5 This is the state transition diagram of the balanced demand assessment and triggering mechanism of this invention; Figure 6 This is a schematic diagram of the intra-group balanced path planning algorithm of the present invention; Figure 7 This is a block diagram of the adaptive current equalization control of the present invention; Figure 8 This is a timing diagram of the three-stage strategy under dynamic operating conditions of the present invention.

[0028] Figure 9 This is a performance comparison chart between the present invention and existing balancing methods. Detailed Implementation

[0029] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0030] The following is for reference. Figures 1 to 9A multi-cell battery pack active balancing control method with multi-level SOC coordination according to an embodiment of the present invention is described, comprising the following steps: S1. Real-time acquisition of terminal voltage, temperature and charging / discharging current of each cell in the battery pack; S2. State of Charge Estimation and Situation Assessment: The improved dual extended Kalman filter algorithm is used to estimate the state of charge of each individual cell, and the global state of charge standard deviation and global state of charge range are calculated. S3, Layered Triggering: Based on the global standard deviation of charge state and the global range of charge state, the current equilibrium level is determined through a multi-threshold layered triggering mechanism. The equilibrium level includes at least the intra-group equilibrium level and the full-topology collaborative equilibrium level. S4. Balanced Execution: Based on the determined balance level, energy transfer is performed through intra-group flyback converter topology and / or inter-group multi-winding transformer topology. S5. Dynamic pre-compensation: Under dynamic operating conditions, predict the trend of charge state change and initiate balancing action in advance.

[0031] For example, the global SOC (state of charge) mean. Global SOC range Global SOC standard deviation Global SOC Difference Matrix These are all evaluation indicators at the global level, defined as follows: , , , , in, This represents the total number of individual cells in the battery pack. For the first The state of charge of each individual cell, and These represent the maximum and minimum values ​​of the state of charge (SOC) values ​​for N individual cells. ; and The first The equilibrium group and the first The average state of charge of each equilibrium group and .

[0032] These indicators together constitute a global SOC situation map, providing a basis for decision-making in balanced path planning.

[0033] In some embodiments of the present invention, such as Figures 1 to 7 As shown, it may also include a grouping step: The battery cells are evenly divided into There are 3 equilibrium groups, each containing 10 equilibrium groups. Series-connected batteries are grouped according to at least one of the following principles: physical adjacency, consistency screening, and thermal coupling. For example, a 32-cell battery pack can be divided into 4 groups of 8 cells each. The physical adjacency principle prioritizes grouping physically adjacent cells together to reduce wiring complexity. The consistency screening principle considers factory capacity and internal resistance data, grouping cells with good consistency together. The thermal coupling principle considers heat dissipation structure, grouping thermally coupled cells together to reduce the impact of temperature differences on battery equalization.

[0034] In some embodiments of the present invention, the improved dual extended Kalman filter algorithm in step S2 includes: Two extended Kalman filters are run simultaneously: the first is used to identify battery model parameters online, and the second is used for state of charge estimation. The battery model uses a second-order RC equivalent circuit model: , , , in, and These are the polarization voltages of the two parallel RC circuits, respectively. The internal resistance of the battery is ohms. and These are the polarization internal resistance and polarization capacitance of the first RC parallel circuit, respectively. and These are the polarization internal resistance and polarization capacitance of the second RC parallel circuit, respectively. The charging and discharging current of the battery is specified, with the discharge direction being positive. This is the open-circuit voltage corresponding to the state of charge (SOC). This refers to the battery terminal voltage. The improvements include at least one of the following: (1) Temperature compensation correction: After collecting the unit temperature T, the open circuit voltage, ohmic internal resistance and the resistance and capacitance parameters of the two parallel RC links in the second-order RC equivalent circuit model are corrected in real time so that the parameters output by the first extended Kalman filter are updated synchronously with the temperature change. (2) Adaptive noise covariance matrix adjustment: Based on the residual between the measured terminal voltage and the model estimate, the rate of change of current, and the residual covariance, the process noise covariance matrix Q and the observation noise covariance matrix R are adjusted online; (3) Short-time correction based on ampere-hour integration: The SOC increment is obtained based on the integration of charging and discharging current within a preset short-time window, and the SOC increment is weighted and fused with the SOC estimate of the second extended Kalman filter to reduce the estimation lag when large current changes.

[0035] Specifically, such as Figure 4 As shown, the temperature compensation factor can be calculated according to... Introduced, among which Indicates OCV, , , , or Any model parameter in, The current monomer temperature, To calibrate the reference temperature, This represents the temperature coefficient of the corresponding model parameters.

[0036] Adaptive noise adjustment can be pressed , Execution, among which, This represents the change in current between adjacent sampling periods. The residual between the measured terminal voltage and the model estimate. and These are calibration coefficients.

[0037] The ampere-hour integral short-time correction can be set according to Calculate, and through Integration, in which, For Coulomb efficiency, The sampling period is Nominal capacity To integrate weights; thereby enabling temperature compensation, adaptive noise covariance adjustment, and ampere-hour integral correction in Figure 4 The process shown operates sequentially on parameter identification, state estimation, and short-time correction stages.

[0038] In some embodiments of the present invention, step S2 further calculates group-level state of charge characterization quantities: For each equilibrium group, define: , , Where SOCgroup_k is the average state of charge of the kth equilibration group, k∈{1,2,...,M}; σk is the standard deviation of the state of charge within the kth equilibration group, used to characterize the degree of inconsistency of individual cells within the group and to determine whether to initiate equilibration within the group.

[0039] In some embodiments of the present invention, such as Figure 5As shown, the multi-threshold hierarchical triggering mechanism in step S3 can initiate different levels of balancing based on the severity of SOC differences. These different levels of balancing include: Level 1 equilibrium, also known as monitoring-level equilibrium: when the global state of charge standard deviation... When the battery level is less than 1%, only the battery status is recorded, and the balancing action is not initiated. Secondary equilibrium, also known as intra-group fine-tuning equilibrium: when the global state of charge standard deviation... When the percentage of the charge level is greater than or equal to 1% and less than 3%, and the standard deviation of the charge state within the group is greater than or equal to 1.5%, the group balancing is initiated, with a balancing current of 0.5-1A. Level 3 equilibrium, also known as strong equilibrium within the group: when the global state of charge standard deviation is greater than or equal to 3% and less than 5%, or the global state of charge range is... When the equivalence is greater than or equal to 4%, intra-group balancing is initiated and inter-group auxiliary balancing is supplemented, with a balancing current of 1-3A. Level 4 equilibrium, also known as global emergency equilibrium: when the global state of charge standard deviation... Greater than or equal to 5%, or extremely poor global state of charge. When the percentage is greater than or equal to 8%, full-topology collaborative balancing is initiated, with a balancing current of 3-5A. The trigger judgment period is 100 ms; when the trigger conditions of multiple levels are met at the same time, the higher level of equalization is executed first; a hysteresis interval of 0.5% is set between the start threshold and the exit threshold.

[0040] Understandably, by employing the aforementioned hierarchical triggering mechanism, the system updates the global state of charge (SOC) assessment every 100ms and determines whether the balancing level needs adjustment, achieving a rapid response to changes in battery pack inconsistencies. Different levels correspond to different balancing current ranges, allowing the system to operate with lower power consumption during mild imbalances and to quickly eliminate differences with higher current during severe imbalances, thus achieving a dynamic balance between balancing speed and energy consumption. Simultaneously, the 0.5% hysteresis interval between the start and stop thresholds effectively avoids frequent start-stop operations due to SOC estimation fluctuations or noise interference, significantly reducing switching losses of the switching transistors and extending system lifespan.

[0041] In some embodiments of the present invention, during the energy transfer performed by the flyback converter topology within the group in step S4, the equalization current... satisfy: , in, The duty cycle of the primary side is the on-state. For magnetizing inductance, The switching frequency; Inter-group multi-winding transformer topologies operate in the following two modes: Mode 1: Direct inter-group transfer, transferring energy from a high-charge state group to a low-charge state group; Mode 2: Intra-group redistribution, assisting in intra-group balancing when the difference in state of charge between groups is below a threshold.

[0042] Specifically, the group-based balancing employs a bidirectional balancing topology based on a flyback transformer. Each group is equipped with one flyback converter, whose primary side is connected to each individual cell within the group via a MOSFET switch array, and its secondary side is connected to each individual cell within the group via another MOSFET switch array. When energy needs to be transferred from cell i to cell j, the primary-side MOSFET is turned on, energizing the primary side of the flyback transformer from cell i; then, the primary-side MOSFET is turned off, and the secondary-side MOSFET is turned on, allowing the flyback transformer to release energy to cell j. The balancing current can be controlled by adjusting the primary-side duty cycle D. The size of the equalization unit is considered. Simultaneously, inter-group equalization employs a multi-winding transformer coupling topology. Each winding of a multi-winding transformer is connected to the total positive and total negative terminals of each equalization group. Energy transfer between different equalization groups is achieved by controlling the switching timing of the MOSFETs on each winding side. When there is a significant difference in the state of charge between two equalization groups, energy is directly transferred from the high-charge-state group to the low-charge-state group, resulting in higher efficiency. When the difference in the state of charge between equalization groups is small, but the difference within each group is large, the inter-group equalization unit assists in intra-group equalization to improve the overall equalization speed.

[0043] In some embodiments of the present invention, the following path planning strategy may be adopted in step S4: Intra-group path planning: The maximum difference priority strategy is adopted, and the individual cells in the group are divided into donors and acceptors according to the deviation of the state of charge. A greedy matching algorithm is used to form a balanced energy transfer path. Inter-group path planning: With minimizing the global state of charge variance as the objective function, a constrained optimization problem is established, and a hierarchical solution strategy is adopted: first, at the group level, quadratic programming is used to determine the direction and magnitude of energy transfer between groups, and then at the intra-group level, the allocated energy is further distributed to specific individuals.

[0044] Specifically, such as Figure 6 As shown, the intra-group balanced path planning adopts the maximum difference priority strategy, which may include the following steps: 1) Calculate the state of charge of each individual unit within the group. Compared with the average state of charge within the group deviation , ,in, ; 2) The monomer is labeled as "donor". The monomer is labeled as "receptor"; 3) Press Donors and recipients are sorted in descending order of value; 4) Using a greedy matching algorithm, the largest donor is paired with the largest acceptor to form a balanced energy transfer path; 5) Calculate the expected transfer energy for each path. , ,in, Nominal capacity Nominal voltage; 6) Determine the equalization time for each path based on the power capability of the flyback converter.

[0045] Inter-group balancing employs a global optimization strategy, using the balanced state of charge of each individual unit as the optimization variable and minimizing the global state of charge variance as the objective function. And establish the following constraints: Condition 1: Equalization Current Not exceeding the device's rated value ,Right now ; Condition 2: The charge state of a single cell does not exceed the safe range, i.e. ; Condition 3: Equalizing Power Not exceeding the system's heat dissipation capacity ,Right now ; Condition 4: Equilibrium Time No more than the allowed window ,Right now .

[0046] Based on this, the inter-group balancing adopts a hierarchical solution strategy. First, at the group level, quadratic programming is used to determine the direction and magnitude of energy transfer between groups. Then, at the intra-group level, the allocated energy is further distributed to specific individuals.

[0047] In some embodiments of the present invention, such as Figure 7 As shown, it may also include a current equalization adaptive control step: Based on the magnitude of the state of charge difference, the battery state of charge range, and temperature The equalization current is adaptively adjusted to achieve the final equalization current. satisfy: , in, This is the current reference value. , The equilibrium rate coefficient, in units of h1, ranges from 0.2 h1 to 0.5 h1. The magnitude of the difference in state of charge; This is a correction factor for the state of charge interval, when hour, ;when or hour, ; For temperature correction factor, when hour, ;when hour, ;when hour, When T < 5℃ or T > 55℃, βT = 0, and equilibrium is suspended.

[0048] The maximum allowable current for the equalization circuit.

[0049] It is understandable that the magnitude of the state of charge (SOC) difference is positively correlated with the balancing current: the greater the SOC difference, the greater the required balancing current. Therefore, this invention sets a current reference value. Through the proportionality coefficient The equalization current is linearly adjusted according to the difference in SOC, thereby accelerating the equalization speed when the difference is significant.

[0050] Meanwhile, to avoid the risk of overcharging or over-discharging in the extreme SOC range, this invention introduces an SOC range correction coefficient. This is to reduce the balancing current in the extreme range of SOC, specifically: when SOC is in the normal range of 10% to 90%, Maintain full balance current; when SOC is below 10% or above 90%, To ensure battery safety, the equalization current should be appropriately reduced.

[0051] Furthermore, the increased internal resistance of the battery at low temperatures leads to a corresponding increase in Joule heat generated by the equalization current, which may exacerbate the risk of lithium plating at low temperatures. Therefore, this invention introduces a temperature correction coefficient. The equalization current is linearly reduced in the low temperature (5°C to 15°C) and high temperature (45°C to 55°C) ranges, and maintained within the ideal temperature range of 15°C to 45°C. To achieve a balance between efficiency and thermal safety.

[0052] Final Equalization Current The minimum value obtained after adjusting for SOC range and temperature, based on the above reference value, is then compared with the maximum allowable current of the circuit. The current is determined and output to the equalization MOSFET in the form of PWM duty cycle to achieve precise control of the equalization current.

[0053] In some embodiments of the present invention, such as Figure 8As shown, step S5, which predicts the trend of state of charge change and initiates the balancing action in advance, includes: State tracking: The dual extended Kalman filter algorithm is used to track and continuously update the estimated state of charge of each individual cell in real time, and the process noise covariance matrix is ​​adaptively adjusted. and observation noise covariance matrix ; Trend prediction: Based on historical trends in state of charge, a linear extrapolation method is used to predict future trends. Changes in state of charge within the time window: ， in, for Time of the first The charge state of each individual cell, For the first Individual in Rate of change of state of charge at time t; prediction window Adjustments based on dynamic operating conditions: Under steady-state operating conditions Under dynamic operating conditions .

[0054] Pre-compensation control: If it is predicted that the state of charge of a certain single cell will be... If the equalization trigger threshold is exceeded, the equalization action of that single unit will be initiated ahead of schedule; pre-compensation intensity coefficient. It is directly proportional to the prediction confidence level: , in, The coefficient of determination for linear extrapolation.

[0055] It is understandable that, such as Figure 8 As shown, the state tracking stage adaptively adjusts the noise covariance matrix to ensure that the SOC estimation does not diverge during large current surges, providing high-precision input for prediction. The trend prediction stage adaptively adjusts the prediction window based on the dynamic nature of the operating conditions, balancing steady-state smoothness and dynamic response speed. The pre-compensation control stage adjusts the compensation intensity based on the prediction confidence level, acting proactively when the prediction is reliable and conservatively adjusting when the prediction is uncertain. These three stages form a closed-loop mechanism of "precise tracking → adaptive prediction → confidence-weighted compensation," enabling the equalization system to adapt to dynamic operating condition changes throughout the charging and discharging process. This effectively avoids problems such as equalization failure due to response lag under current surges or operating condition switching conditions, and false triggering due to SOC estimation fluctuations, which are common in traditional solutions, thus improving the equalization effect in dynamic scenarios.

[0056] like Figures 1 to 7 As shown, the present invention also provides a multi-level SOC-coordinated active balancing control system for multi-string battery packs, used to execute the above method, including: The sensing layer includes a voltage acquisition circuit, a temperature acquisition circuit, and a current sensor, which are used to acquire the terminal voltage, temperature, and charging / discharging current of each battery cell in real time; among them, the voltage acquisition accuracy reaches ±1mV, and the sampling frequency is not less than 100Hz. The decision-making layer includes the main control MCU and the multi-level SOC collaborative balancing algorithm module running on it. The algorithm module includes an accurate SOC estimation module, a balancing demand assessment module, a balancing path planning module, and a balancing current optimization module. The main control MCU communicates with the sensing layer through a CAN bus or isolated SPI. The execution layer includes a two-level balancing topology from group to global. The two-level balancing topology includes intra-group balancing units and inter-group balancing units. The intra-group balancing units adopt a flyback converter topology to realize energy transfer between individual units within the group. The inter-group balancing units adopt a multi-winding transformer topology to realize energy redistribution between balancing groups. All balancing switches are independently controlled by the main control MCU through PWM signals.

[0057] Specifically, in the intra-group equalization unit, each group is equipped with a flyback converter, whose primary side is connected to each individual unit in the group through a first MOSFET switch array, and its secondary side is connected to each individual unit in the group through a second MOSFET switch array; in the inter-group equalization unit, each winding of the multi-winding transformer is connected to the total positive terminal and the total negative terminal of each equalization group through its own MOSFET switch array.

[0058] The balanced path planning module can include intra-group path planning units and inter-group path planning units. The intra-group path planning unit adopts the maximum difference priority strategy, classifying each individual in the group into donors and acceptors according to the charge state deviation, and uses a greedy matching algorithm to form a balanced energy transfer path. The inter-group path planning unit models the inter-group energy transfer as a constrained optimization problem with minimizing the global charge state variance as the objective function. At the group level, quadratic programming is used to determine the direction and magnitude of energy transfer between groups, and at the intra-group level, the allocated energy is further distributed to specific individuals.

[0059] Example 1: Balancing of a 32-cell lithium iron phosphate battery pack This embodiment uses a 32-cell lithium iron phosphate (LiFePO4) battery pack used in an electric vehicle as an application scenario. The nominal capacity of the cells is 100Ah, the nominal voltage is 3.2V, and the nominal voltage of the battery pack is 102.4V.

[0060] I. System Configuration This embodiment builds a balanced control system according to the three-layer architecture of "perception layer - decision layer - execution layer" described in this invention, with the specific configuration as follows: Grouping scheme: The 32 battery strings are evenly divided into 4 equal groups, each containing 8 battery strings, and the grouping follows the principle of physical adjacency; Intra-group balancing topology: Each group is equipped with one flyback converter, with a switching frequency of 100kHz, a magnetizing inductance of 10μH, and supports balancing current of 0.5-3A; Inter-group balancing topology: Equipped with one 4-winding multi-winding transformer with a rated power of 500W, supporting 1-5A balancing current; Main control MCU: adopts ARM Cortex-M4 core, main frequency 168MHz, and runs multi-level SOC collaborative balancing algorithm; Sensing layer hardware: 16-channel ADC, voltage acquisition accuracy ±0.5mV, sampling rate 200SPS, meeting the requirements of ±1mV and above 100Hz as described in this invention.

[0061] II. Initial State Assuming that after a certain number of charge-discharge cycles, the state of charge (SOC) of each cell in the battery pack becomes inconsistent, with an initial SOC distribution range of 75%–82%, the states of each equalization group are as follows: Group 1 (Series 1-8): Average State of Charge within the Group Within-group standard deviation ; Group 2 (9-16 strings): Average state of charge within the group Within-group standard deviation ; Group 3 (strings 17-24): Average state of charge within the group Within-group standard deviation ; Group 4 (25-32 strings): Average state of charge within the group Within-group standard deviation .

[0062] III. Balanced Triggering and Execution Process 1. Overall Situation Assessment and Balance Level Determination The sensing layer collects the terminal voltage, temperature, and charging / discharging current of each individual cell. The decision layer estimates the state of charge (SOC) of each individual cell based on the improved DEKF algorithm, and calculates the global SOC standard deviation. Global SOC range .

[0063] According to the multi-threshold hierarchical triggering mechanism of the present invention, the current state satisfies the triggering condition of secondary equilibrium (intra-group fine-tuning): And Group 3 The system determines that it has entered the secondary equalization stage and initiates intra-group equalization, with the equalization current range being 0.5-1A.

[0064] 2. Intra-group balanced path planning and execution Due to the within-group standard deviation of Group 3 At its highest level, the system prioritizes performing intra-group balancing on this group. A "maximum difference first" strategy is used for path planning. First, calculate the deviation of the SOC of each individual cell in Group 3 from the average SOC of the group (i.e., the average state of charge within the group, hereinafter the same), that is, calculate the deviation of the individual cells' SOC. to and With a deviation of 77.3%, the largest donor was identified as the 20th string. , The largest receptor is the 19th string ( , Then, a greedy matching algorithm is used to pair the 20th string with the 19th string to form a balanced energy transfer path. Next, based on the SOC difference magnitude and nominal capacity, combined with a temperature correction coefficient... (Assuming ambient temperature) , ), calculate the final equalization current The current is set to 1.5A. Finally, the decision layer controls the flyback converter of the third group through the PWM signal to transfer energy from the 20th string to the 19th string.

[0065] As the equilibration process continued, the SOC differences within Group 3 gradually decreased. It gradually decreased from 2.1%.

[0066] 3. Balanced intervention between groups When the intra-group equilibrium in Group 3 has been achieved to a certain extent, the intra-group differences have been partially eliminated, but the differences in SOC between groups are still significant. At this point, the average SOC difference between Group 3 and Group 2 is greater than 2%, triggering the need for inter-group equilibrium.

[0067] The system starts the inter-group multi-winding transformer and operates in mode one (direct inter-group transfer), transferring energy from the second group (80.1%) with a higher average SOC to the third group (77.3%) with a lower average SOC, in order to reduce the difference between groups.

[0068] 4. Complete in a balanced manner like Figure 8 As shown, after approximately 45 minutes of stratified collaborative equilibration (alternating or simultaneous within-group and between-group sessions), the global SOC standard deviation was... If the battery level drops below 0.5%, which is below the 1% monitoring threshold, the system automatically downgrades to Level 1 monitoring, stops balancing operations, and only continuously monitors the battery status.

[0069] This embodiment verifies the effectiveness of the technical solution described in this invention: the SOC-based hierarchical triggering mechanism accurately identifies the intra-group imbalance state of the third group, avoiding potential misjudgments that may occur during the plateau period of lithium iron phosphate using traditional voltage criteria. The greedy matching algorithm quickly locates the maximum donor and acceptor, maximizing energy transfer efficiency. The hierarchical strategy of first addressing intra-group differences and then inter-group differences eliminates intra-group discrepancies while also considering inter-group differences, avoiding global efficiency reduction caused by local equilibration. The equilibration time of 45 minutes is reduced by approximately 30% compared to traditional schemes, and the energy conversion efficiency of the active equilibration topology reaches over 85% (see...). Figure 8 ).

[0070] Example 2: Dynamic balancing of a 16-cell ternary lithium battery pack This embodiment uses a power battery pack consisting of 16 series ternary lithium (NCM) cells used in a pure electric passenger vehicle as the application scenario. The nominal capacity of the cells is 60Ah and the nominal voltage is 3.7V.

[0071] 1. Dynamic working conditions When a vehicle is driving in urban conditions, frequent acceleration and deceleration cause the battery current to fluctuate between -150A (regenerative braking) and +100A (acceleration). Traditional equalization solutions, due to the drastic voltage fluctuations under these conditions, frequently trigger equalization actions erroneously. Furthermore, when equalization is truly needed, the rapid voltage changes make it difficult to effectively determine the cause, resulting in poor equalization performance or even failure.

[0072] 2. Work Process 1) State tracking The sensing layer collects real-time data on the terminal voltage, temperature, and charging / discharging current of each individual cell. The decision layer employs an improved DEKF algorithm to continuously track the estimated state of charge (SOC) of each cell under dynamic current conditions. By adaptively adjusting the process noise covariance matrix Q and the observation noise covariance matrix R, the algorithm maintains high tracking performance even under sudden large current changes. In this embodiment, the DEKF algorithm controls the SOC estimation error under dynamic operating conditions to within 1.5%, providing a reliable data foundation for subsequent prediction and equilibrium decisions.

[0073] 2) Trend Forecasting Based on historical SOC (Solar Oxygen Circulation) trends, the system uses linear extrapolation to predict SOC changes within future time windows. Considering the dynamic nature of urban operating conditions, the prediction window... The prediction formula is set to 1 second (dynamic operating condition): , The system updates the predicted value every 100ms, tracking the changing trend of the SOC of each individual unit in real time.

[0074] 3) Pre-compensation control When it is predicted that the SOC of a certain cell will exceed the equalization trigger threshold in 1 second, the system initiates the equalization action of that cell in advance, achieving proactive control. In this embodiment, the current SOC of a certain cell is 82%, and it is predicted that it will reach 83.5% in 1 second (exceeding the level 3 equalization trigger threshold of 83%). The system uses a pre-compensation intensity coefficient Initiate the balancing action in advance. The value and the coefficient of determination of linear extrapolation The higher the reliability of the prediction, the greater the pre-compensation.

[0075] 4) High current protection Considering the significant battery polarization effect under high current conditions, and the susceptibility of voltage and SOC estimation to interference, this embodiment introduces a high-current protection mechanism. When the absolute value of the charging / discharging current exceeds 80A (i.e., |I|>80A), the system temporarily freezes the equalization operation to prevent energy transfer under extreme current conditions. This protection mechanism effectively prevents erroneous equalization operations caused by estimation errors, while also avoiding the safety hazards associated with energy transfer under high current conditions.

[0076] 5) Balanced recovery and path replanning Once the current returns to the normal range (|I|≤80A), the system re-evaluates the global situation and plans the balancing path based on the latest SOC estimate. Based on the updated SOC distribution, the decision-making layer determines the current balancing level and generates new energy transfer paths according to the intra-group "maximum difference priority" strategy and the inter-group "global optimization" strategy, and continues to execute the balancing action.

[0077] This embodiment was simulated and verified under WLTC (Worldwide Harmonized Light Vehicles Test Cycle) conditions. The results are as follows: Compared with traditional voltage criterion equalization schemes, the present invention's SOC-based hierarchical triggering mechanism combined with a three-stage dynamic equalization strategy improves equalization accuracy by 40%, effectively avoiding false triggering caused by voltage fluctuations. The pre-compensation control mechanism initiates equalization in advance based on the predicted SOC change trend, effectively overcoming the equalization failure problem caused by response lag in traditional schemes. By temporarily freezing the equalization action under extreme current conditions, maloperation caused by estimation errors is avoided, improving the safety and reliability of the system. Through adaptive equalization current control and a protection freezing mechanism under high current conditions, this scheme reduces energy loss during the equalization process by 25%. Throughout the entire WLTC cycle, no false triggering events occurred, significantly improving the stability and reliability of the system.

[0078] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0079] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0080] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0081] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0082] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0083] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A multi-level SOC-coordinated active balancing control method for multi-string battery packs, characterized in that, Includes the following steps: S1. Real-time acquisition of terminal voltage, temperature and charging / discharging current of each cell in the battery pack; S2. State of Charge Estimation and Situation Assessment: The improved dual extended Kalman filter algorithm is used to estimate the state of charge of each individual cell, and the global state of charge standard deviation and global state of charge range are calculated. S3. Layered triggering: The current equilibrium level is determined through a multi-threshold layered triggering mechanism based on the global standard deviation of the state of charge and the global range of the state of charge. The equilibrium level includes at least the intra-group equilibrium level and the full-topology collaborative equilibrium level. S4. Balanced Execution: Based on the determined balance level, energy transfer is performed through intra-group flyback converter topology and / or inter-group multi-winding transformer topology. S5. Dynamic pre-compensation: Under dynamic operating conditions, predict the trend of charge state change and initiate balancing action in advance.

2. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, It also includes a grouping step: The battery cells are evenly divided into There are 3 equilibrium groups, each containing Series batteries; Grouping follows at least one of the following principles: physical adjacency principle, consistency screening principle, and thermal coupling principle.

3. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, The improved dual extended Kalman filter algorithm described in step S2 includes: Two extended Kalman filters are run simultaneously; the first is used for online identification of battery model parameters. , , , The second one is used for state of charge estimation; where the battery model is a second-order RC equivalent circuit model: ， ， ， in, and These are the polarization voltages of the two parallel RC circuits, respectively. The internal resistance of the battery is ohms. and These are the polarization internal resistance and polarization capacitance of the first RC parallel circuit, respectively. and These are the polarization internal resistance and polarization capacitance of the second RC parallel circuit, respectively. The charging and discharging current of the battery is specified, with the discharge direction being positive. This is the open-circuit voltage corresponding to the state of charge (SOC). This refers to the battery terminal voltage. The improvement includes at least one of the following: Temperature compensation corrections are made to the open-circuit voltage, ohmic internal resistance, and RC parameters of the two parallel RC circuits based on the unit temperature T. Based on the terminal voltage residual, current change rate, and residual covariance, the process noise covariance matrix Q and the observation noise covariance matrix R are adjusted online. The SOC increment is calculated using the ampere-hour integration method within a preset short time window, and the SOC increment is then weighted and fused with the extended Kalman filter output for correction.

4. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, In step S2, the group-level state of charge characterization quantities are also calculated: For each equilibrium group, define: ， ； in, For the first The average state of charge of the group ; The standard deviation of the state of charge within the group; This represents the global average state of charge.

5. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, The multi-threshold hierarchical triggering mechanism mentioned in step S3 includes: Level 1 equalization, also known as monitoring-level equalization: When the global state of charge standard deviation is less than 1%, only the battery state is recorded, and equalization action is not initiated. Secondary balancing, also known as intra-group fine-tuning balancing: Intra-group balancing is initiated when the global state of charge standard deviation is greater than or equal to 1% and less than 3%, and the intra-group state of charge standard deviation is greater than or equal to 1.5%. Level 3 equilibrium, namely strong equilibrium within the group: when the global state of charge standard deviation is greater than or equal to 3% and less than 5%, or the global state of charge range is greater than or equal to 4%, the intra-group equilibrium is initiated and supplemented by inter-group auxiliary equilibrium. Level 4 equilibrium, also known as global emergency equilibrium: when the global standard deviation of charge state is greater than or equal to 5%, or the global range of charge state is greater than or equal to 8%, the whole topology collaborative equilibrium is initiated. The trigger judgment period is 100 ms; when the trigger conditions of multiple levels are met at the same time, the higher level of equalization is executed first; a hysteresis interval of 0.5% is set between the start threshold and the exit threshold.

6. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, In step S4, when the flyback converter topology within the group performs energy transfer, the equalization current... satisfy: ， in, The duty cycle of the primary side is the on-state. For magnetizing inductance, The switching frequency; The inter-group multi-winding transformer topology operates in the following two modes: Mode 1: Direct inter-group transfer, transferring energy from a high-charge state group to a low-charge state group; Mode 2: Intra-group redistribution, assisting in intra-group balancing when the difference in state of charge between groups is below a threshold.

7. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, In step S4, the following path planning strategy is adopted: Intra-group path planning: The maximum difference priority strategy is adopted, and the individual cells in the group are divided into donors and acceptors according to the deviation of the state of charge. A greedy matching algorithm is used to form a balanced energy transfer path. Inter-group path planning: With minimizing the global state of charge variance as the objective function, a constrained optimization problem is established, and a hierarchical solution strategy is adopted: first, at the group level, quadratic programming is used to determine the direction and magnitude of energy transfer between groups, and then at the intra-group level, the allocated energy is further distributed to specific individuals.

8. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, It also includes a current equalization adaptive control step: Based on the magnitude of the state-of-charge difference, the battery state-of-charge range, and the temperature T, the magnitude of the balancing current is adaptively adjusted, resulting in the final balancing current. satisfy: ， in, This is the current reference value. , The equilibrium rate coefficient, in units of h1, ranges from 0.2 h1 to 0.5 h1. The magnitude of the difference in state of charge; For the state-of-charge interval correction factor: when hour, ;when or hour, ; For temperature correction factor, when hour, ;when hour, ;when hour, When T < 5℃ or T > 55℃, βT = 0, and equilibrium is suspended. The maximum allowable current for the equalization circuit.

9. The multi-level SOC-coordinated active balancing control method for multi-string battery packs according to claim 1, characterized in that, The step S5, which involves predicting the trend of state of charge change and initiating the balancing action in advance, includes: State tracking: The dual extended Kalman filter algorithm is used to track and continuously update the estimated state of charge of each individual cell in real time, and the process noise covariance matrix is ​​adaptively adjusted. and observation noise covariance matrix ; Trend prediction: Based on historical trends in state of charge, a linear extrapolation method is used to predict future trends. Changes in state of charge within the time window: ， in, for Time of the first The charge state of an individual cell, For the first Individual in Rate of change of state of charge at time t; prediction window Adjustments based on dynamic operating conditions: Under steady-state operating conditions Under dynamic operating conditions . Pre-compensation control: If it is predicted that the state of charge of a certain single cell will be... If the equalization trigger threshold is exceeded, the equalization action of that single unit will be initiated ahead of schedule; pre-compensation intensity coefficient It is proportional to the prediction confidence level: ， in, The coefficient of determination for linear extrapolation.

10. A multi-level SOC-coordinated active balancing control system for multi-string battery packs, used to execute the method according to any one of claims 1 to 9, characterized in that, include: The sensing layer includes a voltage acquisition circuit, a temperature acquisition circuit, and a current sensor, which are used to acquire the terminal voltage, temperature, and charging / discharging current of each battery cell in real time; among them, the voltage acquisition accuracy reaches ±1mV, and the sampling frequency is not less than 100Hz. The decision-making layer includes a main control MCU and a multi-level SOC collaborative balancing algorithm module running on it. The algorithm module includes an accurate SOC estimation module, a balancing demand assessment module, a balancing path planning module, and a balancing current optimization module. The main control MCU communicates with the sensing layer via a CAN bus or isolated SPI. The execution layer includes a two-level balancing topology that groups to the global level. The two-level balancing topology includes intra-group balancing units and inter-group balancing units. The intra-group balancing units adopt a flyback converter topology to realize energy transfer between individual units within the group. The inter-group balancing units adopt a multi-winding transformer topology to realize energy redistribution between balancing groups. All balancing switches are independently controlled by the main control MCU through PWM signals.

Images