Lithium battery equalization power supply method for middle line ups

Abstract

The application belongs to the technical field of power supply control, and particularly relates to a neutral line UPS lithium battery equalization power supply method. The method comprises the following steps: acquiring single-state data, neutral line operation data and load power of a first battery pack and a second battery pack, and determining a candidate neutral line bias current set; generating a boundary domination chain and a boundary time ordering sequence according to the candidate neutral line bias current set, the single-state data and the load power; determining a corrected boundary risk surrogate value according to the boundary domination chain and the boundary time ordering sequence, and mapping the corrected boundary risk surrogate value into a virtual neutral line impedance, and determining a target neutral line bias current from the virtual neutral line impedance; and adjusting a neutral line current according to the target neutral line bias current, so as to distribute the load power of the first battery pack and the second battery pack. The application can unify the use of dominated single cells, dominated boundary types, boundary arrival times and boundary approximation rates under the constraint of single cell safety boundaries, so that the load power distribution of the two groups of batteries is adapted to the boundary risk.

Description

Technical Field

[0001] This invention relates to the field of power control technology, specifically to a neutral-line UPS lithium battery balanced power supply method. Background Technology

[0002] A neutral-line UPS typically consists of a first battery pack and a second battery pack connected to opposite sides of a split DC bus. The two battery packs share the load power, and their power supply status is affected not only by the battery pack terminal voltage and state of charge, but also by the voltage, temperature, capacity decay, and neutral current changes of each individual cell.

[0003] Existing technologies typically set fixed thresholds based on the voltage difference, state-of-charge difference, or maximum and minimum single-cell voltage between the two battery banks, and adjust the power sharing between the two banks through relay switching, current limiting, or fixed droop parameters. This approach primarily addresses the overall state of the battery pack and struggles to differentiate the order in which different cells within the same battery bank approach their voltage, temperature, or capacity safety boundaries. When a weak cell approaches its safety boundary prematurely, the average state of the battery pack may still be within acceptable limits, preventing timely power transfer to the lower-risk side. Conversely, directly using the maximum or minimum values ​​as control criteria could lead to sudden power fluctuations across the entire pack caused by a single abnormal cell.

[0004] On the other hand, existing methods typically lack a unified description of the relationships between dominant cells, dominant boundary types, and boundary arrival times under different neutral bias currents. They also fail to compare the ranking of boundary arrival times for each cell, making it difficult to distinguish between locally weak cells and situations where the entire battery pack is approaching the safety boundary. When using a fixed threshold or fixed virtual impedance, the lack of a continuous mapping between boundary risk and neutral current regulation can easily lead to lag in power distribution between the two battery packs, repeated switching, or prematurely reaching the safety boundary on one side of the battery pack, thereby affecting the power balance and continuous power supply capability of the split DC bus. Summary of the Invention

[0005] This invention provides a neutral-line UPS lithium battery balanced power supply method, which is used to at least solve the problem of how to correlate and evaluate the boundary risks of different candidate neutral bias currents and determine the target neutral bias current in order to allocate the load power of the two battery groups under the condition that the individual safety boundary states of the first battery group and the second battery group are inconsistent.

[0006] This invention provides a method for balanced power supply of lithium batteries in a neutral-line UPS, the method comprising: Acquire the individual cell status data, neutral line operation data and load power of the first and second battery packs to determine the candidate set of neutral line bias currents; Based on the candidate neutral bias current set, individual cell state data and load power, a boundary dominance chain and a boundary time sorting sequence are generated. The boundary dominance chain is associated with the dominant cell, dominant boundary type, boundary arrival time and boundary approach rate under each candidate neutral bias current. The modified boundary risk cost is determined based on the boundary dominance chain and boundary time sorting sequence. The modified boundary risk cost is then mapped to the virtual neutral impedance, and the target neutral bias current is determined by the virtual neutral impedance. Adjust the neutral current according to the target neutral bias current to distribute the load power of the first battery pack and the second battery pack.

[0007] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: By jointly acquiring the individual cell state data, neutral line operation data, and load power of the first and second battery packs, and determining the candidate neutral line bias current set, a unified evaluation of different load power allocation schemes is achieved. By generating a boundary dominance chain, each candidate neutral line bias current is correlated with the dominated cell, the dominance boundary type, the boundary arrival time, and the boundary approach rate, enabling the identification of the cell that first limits continuous power supply and the safety boundary in both battery packs. By constructing a boundary time sorting sequence and determining the modified boundary risk cost, the risk of the dominated cell and the boundary arrival time distribution of multiple cells can be utilized simultaneously, reducing control deviations caused by relying solely on the average state of the battery pack or a single extreme value. By mapping the modified boundary risk cost to a virtual neutral line impedance, and determining the target neutral line bias current from the virtual neutral line impedance, the boundary risk can continuously affect neutral line current regulation and load power allocation. Therefore, the first and second battery packs can share load power according to the individual cell safety boundary state, reducing the possibility of one battery pack interrupting power supply due to a weak cell prematurely reaching the safety boundary. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the UPS lithium battery balanced power supply method of the present invention. Figure 2 This is a diagram showing the state data of key individual units in an embodiment of the present invention; Figure 3 This is a diagram showing the candidate midline bias current and the predicted battery pack current in an embodiment of the present invention. Figure 4 This is a boundary dominance chain diagram in an embodiment of the present invention; Figure 5 This is a boundary time sorting sequence diagram in an embodiment of the present invention; Figure 6 This is a diagram showing the selection of candidate evaluation quantities and baseline centerline bias current in an embodiment of the present invention. Figure 7 This is a diagram showing the adjustment of the redundant switch state operation time and the convergence of the neutral current in an embodiment of the present invention. Detailed Implementation

[0009] The embodiments of the present invention will be described below with reference to the accompanying drawings. These embodiments are used to illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art can make adaptive adjustments to the relevant processing order, data structure, or parameter values ​​without changing the technical concept of the present invention.

[0010] In this embodiment, the neutral-line UPS lithium battery system consists of a first battery pack, a second battery pack, a split DC bus, and a neutral line. The balanced power supply process addresses the current distribution relationship between the two battery packs under load power. The controller obtains individual cell status data from the battery management system, neutral line operation data from the neutral line sampling circuit and the positive and negative bus sampling circuits, and load power from the power conversion control unit. Individual cell status data characterizes the state of each cell as it approaches the voltage safety boundary, temperature safety boundary, and capacity safety boundary. Neutral line operation data characterizes the current neutral line current and midpoint potential offset. Load power determines the predicted battery pack current borne by the two battery packs under different neutral line bias currents. After data alignment and validity checks, the controller determines a candidate neutral line bias current set and passes the candidate neutral line bias current set, individual cell status data, and load power to subsequent processing steps. This is used to generate boundary dominance chains, boundary time sorting sequences, and corrected boundary risk values, thereby determining the target neutral line bias current and adjusting the load power distribution.

[0011] like Figure 1 As shown, a neutral-line UPS lithium battery balanced power supply method includes the following processing steps. Figure 1 The power supply data in the report is a collective term for individual unit status data, neutral line operation data, and load power.

[0012] In this embodiment, when the neutral-line UPS lithium battery system is in battery-powered mode, it acquires the individual status data of the first and second battery packs, neutral-line operation data, and load power at the same sampling time. The neutral-line UPS lithium battery system performs time alignment and validity checks on all types of data to ensure that the individual status data and neutral-line operation data correspond to the same load condition.

[0013] Based on the current neutral current, the current neutral point potential offset, and the allowable neutral operating boundary, a range of candidate neutral bias currents is determined, and a set of candidate neutral bias currents arranged in the current direction is formed within this range. The set of candidate neutral bias currents, together with the load power, is used to determine the first and second predicted battery pack currents, thereby forming the cell boundary margin for each individual cell under different candidate neutral bias currents. This cell boundary margin is then transferred to the boundary dominance chain generation stage.

[0014] In one embodiment, the individual cell state data consists of the cell voltage, cell temperature, cell state of charge, cell internal resistance, and cell polarization voltage of each cell in the first and second battery packs. The cell voltage is acquired through a battery sampling link within a uniform sampling period, and the cell temperature is acquired from temperature sampling points arranged on the cell surface or cell connection area.

[0015] The state of charge (SOC) of a single cell can be determined jointly by the current integral and the settling voltage correction. The current integral reflects the capacity change during power supply, while the settling voltage correction limits the cumulative deviation formed by long-term integration. The internal resistance of a single cell is determined based on the change in single-cell voltage and the change in battery pack current under load changes. When the change in battery pack current is below the identifiable range, the most recent valid single-cell internal resistance is used. The polarization voltage of a single cell is determined based on the difference between the open-circuit voltage corresponding to the single cell's SOC, the single-cell terminal voltage, and the voltage drop across the single cell's internal resistance, and is recursively updated between adjacent sampling times.

[0016] The neutral line operation data consists of neutral line current, positive bus voltage, negative bus voltage, positive bus current, and negative bus current. The neutral line current is used to characterize the power supply bias direction and bias amplitude between the first and second battery packs. The positive and negative bus currents are used to verify the actual current carrying relationship between the first and second battery packs. The positive and negative bus voltages are used to determine the current neutral point potential offset.

[0017] The load power can be determined by the sum of the AC output active power and the power conversion loss, or it can be calculated from the DC side bus voltage and bus current. When both acquisition methods are available, cross-validation is performed using data with consistent timestamps. If the difference exceeds the allowable power deviation, the current sampling period is marked as an invalid calculation period, and the candidate neutral bias current set is not updated.

[0018] Individual cell status data, midline operation data, and load power are aligned according to the sampling time before being included in subsequent calculations. When the individual cell temperature sampling period is longer than the current sampling period, the most recent valid individual cell temperature is maintained between adjacent temperature sampling times. When there are sudden changes in individual cell voltage or current, validity checks are performed based on sensor range, adjacent sampling rate of change, and battery pack series relationship.

[0019] Data that fails the validity check is not included in the determination of the individual cell boundary margin for the current sampling period. The neutral UPS lithium battery system uses the corresponding data from the previous valid sampling period and limits the current amplitude of the candidate neutral bias current set to avoid abrupt changes in power distribution caused by abnormal sampling. After alignment and validity checks are completed, the individual cell status data, neutral operation data, and load power are used together in the process of determining the first predicted battery pack current, the second predicted battery pack current, and the individual cell boundary margin.

[0020] In one embodiment, the candidate neutral bias current set is used to represent multiple neutral current bias schemes that are allowed to be applied under the current power supply condition. The neutral UPS lithium battery system reads the current neutral current, positive bus voltage, and negative bus voltage from the neutral operation data, and determines the current neutral point potential offset according to the difference between the absolute values ​​of the positive bus voltage and the negative bus voltage:

[0021] in, This represents the current midpoint potential offset. Positive bus voltage The current neutral point potential offset is set according to the voltage sampling polarity of the neutral UPS lithium battery system, so that the positive and negative directions of the candidate neutral bias current can respectively indicate the transfer of load power to the first battery pack or the second battery pack.

[0022] The neutral current limit is determined based on the minimum value among the allowable current of the neutral conductor, the allowable current of the power devices, and the neutral current sampling range. The neutral point potential offset limit is determined based on the allowable positive and negative bus voltage imbalance range of the three-level inverter, and is used to limit further offset of the candidate neutral bias current from the neutral point potential.

[0023] The neutral-line UPS lithium battery system determines the allowable current range by combining the current neutral current with the neutral current limit. Based on the relationship between changes in neutral current and neutral point potential offset within adjacent control cycles, it predicts the neutral point potential offset caused by different neutral bias currents. Only current values ​​whose predicted neutral current does not exceed the neutral current limit and whose predicted neutral point potential offset does not exceed the neutral point potential offset limit are included in the candidate neutral bias current range.

[0024] The intersection formed by the two types of constraints is used to prevent the candidate current from meeting the current constraint but causing the midpoint potential to continue to be unbalanced, or meeting the midpoint potential constraint but exceeding the midline current carrying capacity.

[0025] Within the range of candidate neutral bias currents, multiple candidate neutral bias currents are generated at preset bias intervals:

[0026] in, For the first One candidate neutral bias current. This represents the lower limit of the candidate neutral bias current range. For the preset offset interval, This is the candidate number.

[0027] The preset bias interval is determined based on the neutral current adjustment resolution, sampling noise amplitude, and the allowable power change in a single control cycle. Its value should be greater than the equivalent current fluctuation caused by the sampling noise and less than the minimum controllable current that can cause a change in the power sharing between the first and second battery packs.

[0028] Multiple candidate neutral bias currents are arranged in order of current direction from negative to positive. When the zero bias current is within the range of candidate neutral bias currents, it is retained as a candidate value. If the range of candidate neutral bias currents cannot form a non-zero candidate value, only the candidate value that does not increase the current midpoint potential offset is retained, and the resulting set of candidate neutral bias currents is passed to the process of determining the predicted battery pack current.

[0029] In one embodiment, the first predicted battery pack current and the second predicted battery pack current represent the supply current undertaken by the first and second battery packs after the candidate neutral bias current is implemented. The neutral UPS lithium battery system determines the reference battery pack current based on the load power, positive bus voltage, and negative bus voltage, and then adjusts the allocation according to the candidate neutral bias current.

[0030]

[0031] in, For the first The first predicted battery pack current under candidate midline bias currents. For the second predicted battery pack current, and These are the reference battery pack currents for the first and second battery packs, respectively. For the first One candidate neutral bias current. This is the neutral current distribution factor. The neutral current distribution factor is determined based on the calibration relationship between the change in neutral current and the changes in positive bus current and negative bus current.

[0032] When the direction of the discharge current is taken as the positive direction, the open-circuit voltage of the cell is first determined based on the cell's state of charge and temperature, and the predicted cell voltage and voltage boundary margin are determined according to the following formula:

[0033]

[0034] in, For the first The first battery pack The monomer in the first The candidate midline bias current and the first Predicted single-cell voltage at each prediction step The open-circuit voltage of a single cell is determined based on its state of charge and temperature. It is a single-cell charged state. To predict monomer temperature, For the first The battery pack in the first Predicted battery pack current under candidate midline bias currents For the internal resistance of a single monomer, The polarization voltage of a single cell. This is the voltage boundary margin. This is the lowest voltage for a single unit.

[0035] The temperature boundary margin is obtained by subtracting the predicted cell temperature from the highest temperature of the individual cell. The predicted cell temperature is determined based on the current cell temperature, the internal resistance heat generated by the predicted battery pack current, and the temperature change trend. The capacity boundary margin is determined by the difference between the available capacity of the individual cell and the difference between the individual cell's state of charge and the minimum permissible state of charge. The safety boundary is set based on the cell specifications, permissible discharge rate, and power supply duration requirements.

[0036] Voltage boundary margin, temperature boundary margin and capacity boundary margin are stored in association with battery pack number, cell number, candidate serial number and safety boundary type, and the cell boundary margin is formed by the three types of boundary margin.

[0037] If a candidate midline bias current causes the predicted battery pack current to exceed the battery pack's allowable current, or causes any single cell boundary margin to be negative, the candidate midline bias current is marked as unusable; the remaining single cell boundary margins are passed to the prediction process of boundary arrival time and boundary approach rate.

[0038] The controller of the neutral-line UPS lithium battery system receives the candidate neutral bias current set, individual cell status data, load power, and individual cell boundary margin formed in the previous stage. For each candidate neutral bias current, it recursively calculates the voltage boundary margin, temperature boundary margin, and capacity boundary margin of each individual cell in the first and second battery packs within a preset prediction time domain.

[0039] The controller determines the boundary arrival time and boundary approach rate based on the boundary margins of various individual cells. Within the first and second battery packs, it identifies the dominant cells and dominant boundary types, and forms a boundary dominance chain according to the current arrangement order in the candidate midline bias current set. Simultaneously, the controller sorts the shortest boundary arrival times of each cell to form a boundary time sorting sequence, and then transmits the boundary dominance chain and boundary time sorting sequence to the stage of determining the corrected boundary risk cost value.

[0040] In one embodiment, the preset prediction time domain is determined based on the power supply observation time under load power, and the prediction step size is determined according to the individual voltage sampling period.

[0041] Regarding the first The controller uses the first and second predicted battery pack currents to recursively deduce the state of charge, polarization voltage, and temperature of each cell in the first and second battery packs, respectively, and then updates the voltage boundary margin, temperature boundary margin, and capacity boundary margin.

[0042] Within the preset prediction time domain, the state of charge, polarization voltage, and temperature of each individual cell are recursively derived according to the following relationships:

[0043]

[0044]

[0045] in, The charge states of a single cell are represented by numbers from 0 to 1. For charging and discharging efficiency, The available capacity of a single cell is expressed in Ah. The prediction step size is expressed in seconds; when the state of charge of a single cell in the single cell state data is expressed as a percentage, it is converted to a dimensionless number in the range of 0 to 1 before participating in the recursive calculation. and The recursive coefficient for the polarization voltage of a single cell. For the heat capacity of a single unit, For the thermal resistance of a single unit, This refers to ambient temperature. The above parameters were calibrated based on cell specifications and historical operating data. The first battery pack The monomer in the first The candidate midline bias current and the first The first prediction step The boundary margin of a class of single entities is denoted as .in, Number the battery pack. For individual unit numbering, This is the candidate neutral bias current sequence number. This is a single-unit boundary margin type. This is the prediction step number.

[0046] When the boundary margin of a single unit changes from positive to negative, the boundary arrival time is:

[0047] in, For the time of arrival at the boundary, The prediction step size is set as follows: if the safety boundary is not reached within the preset prediction time domain, the upper limit of the preset prediction time domain is used as the boundary arrival time, and the single boundary margin at the end of the prediction time domain is used as the sorting criterion when the boundary arrival times are the same.

[0048] The boundary approach rate is determined by the ratio of the decrease in the individual boundary margin within the continuous prediction interval before the safety boundary is reached to the prediction time:

[0049] in, For the boundary approximation rate, and These are the starting and ending prediction step numbers for the continuous prediction interval, respectively. The continuous prediction interval is used to suppress fluctuations in single sampling.

[0050] The controller compares the boundary arrival time of each unit under different security boundary types, takes the shortest boundary arrival time of each unit as the boundary arrival time of each unit, and takes the security boundary type that produces the shortest boundary arrival time as the security boundary type that each unit first touches.

[0051] For each candidate neutral bias current, the controller determines the cell with the shortest boundary arrival time in the first and second battery packs as the dominant cell, and determines the type of the safe boundary that the dominant cell first touches as the dominant boundary type of the corresponding battery pack.

[0052] The controller associates and stores each candidate neutral bias current with the dominant cell, dominant boundary type, boundary arrival time and boundary approach rate of the first and second battery packs according to the current arrangement order in the candidate neutral bias current set, forming a boundary domination chain.

[0053] Simultaneously, the controller sorts the shortest boundary arrival times of each cell in the first and second battery packs in ascending order, forming a boundary time sorting sequence. The boundary dominance chain and the boundary time sorting sequence are jointly transmitted to the process of determining the modified boundary risk cost value.

[0054] The neutral-line UPS lithium battery system determines the first and second boundary risk values ​​based on the boundary dominance chain and boundary time sequence. These values ​​are then corrected based on the first and second sequence difference values ​​to obtain the first and second corrected boundary risk values. These values ​​are used to determine the first and second virtual neutral impedances, respectively. The impedance difference between the first and second virtual neutral impedances is used to correct the reference neutral bias current, resulting in the target neutral bias current.

[0055] The difference between the first virtual neutral impedance and the second virtual neutral impedance is converted into a neutral bias current correction amount, and the reference neutral bias current is corrected to obtain the target neutral bias current, which is used for neutral current adjustment.

[0056] In one embodiment, a preset observation time is used to screen individual cells that require focused evaluation. The preset observation time is less than or equal to a preset prediction time domain and is determined based on the time required for the neutral current to complete one stable adjustment, the rate of change of load power, and the allowable response time of the individual cell protection action.

[0057] For each candidate neutral bias current, the neutral UPS lithium battery system checks the boundary arrival time of each cell in the first and second battery packs. Cells with boundary arrival times shorter than a preset observation time are added to either the first or second near-boundary cell set, and the cell's first encountered safety boundary type, boundary approach rate, and boundary approach damage are correlated. If a cell meets the screening criteria under multiple safety boundary types, only the safety boundary type with the shortest boundary arrival time is retained.

[0058] Boundary approach damage is used to record the cumulative state of a single unit approaching the safety boundary and experiencing current stress. If a historical record exists, the most recent boundary approach damage is read; otherwise, it is set to zero. The boundary approach damage serves as input for risk calculation and is updated during the neutral current regulation phase before being written to the next control cycle.

[0059] To eliminate the influence of different dimensions of the individual boundary margin under different safety boundary types, the boundary approximation rate is divided by the reference boundary margin of the corresponding safety boundary type to obtain the normalized boundary approximation rate, which is denoted as […]. The reference boundary margin is determined based on the allowable variation range of the corresponding safety boundary type. The voltage safety boundary, temperature safety boundary, and capacity safety boundary are determined using the corresponding allowable voltage range, allowable temperature range, and available capacity range, respectively.

[0060] No. The battery pack in the first The boundary risk cost under each candidate neutral bias current is determined according to the following formula:

[0061] In the formula, Choose 1 or 2. This is the candidate neutral bias current sequence number. For individual unit numbering, For safety boundary type, Value for boundary risks It is either the first set of near-boundary monomers or the second set of near-boundary monomers. For boundary type weights, For the time of arrival at the boundary, To preset the observation time, To normalize the boundary approximation rate, To ensure the boundary is close to the damage amount, For the boundary approximation rate coefficient, The boundary is the damage coefficient.

[0062] The boundary type weight is determined based on the power supply impact after the safety boundary type triggers the protection action. When the normalized boundary approach rate is less than zero, the normalized boundary approach rate is set to zero. When the first near-boundary unit set is empty, the risk value of the first boundary is set to zero; when the second near-boundary unit set is empty, the risk value of the second boundary is set to zero.

[0063] The first boundary risk value and the second boundary risk value are associated one-to-one with the candidate midline bias current and are passed to the correction process of the boundary time sorting sequence.

[0064] In one embodiment, for each candidate neutral bias current, the neutral UPS lithium battery system compares the boundary arrival times of each cell in the first and second battery packs under voltage, temperature, and capacity boundaries, takes the minimum value as the shortest boundary arrival time of the cell, and retains the corresponding safety boundary type.

[0065] The shortest boundary arrival times in the first battery pack are sorted in ascending order to form the first boundary time sequence; the shortest boundary arrival times in the second battery pack are sorted in ascending order to form the second boundary time sequence. When the boundary arrival times are the same, the order is determined by the normalized boundary approximation rate in descending order.

[0066] The preset quantity is determined based on the number of near-boundary cells and the data processing volume of a single control cycle. Its value is simultaneously no greater than the smaller of the number of cells in the first and second battery packs and the length of the reference boundary arrival time series. The reference boundary arrival time series is generated from historical power supply data where the load power is stable and the safety boundary has not been triggered. Within the same load power range, the boundary arrival time at each sorting position in the historical effective control cycles is extracted, and the median of the boundary arrival times at each sorting position is used to form the reference boundary arrival time series.

[0067] No. The battery pack in the first The sequence difference value under each candidate neutral bias current is determined according to the following formula:

[0068] In the formula, Choose 1 or 2. This is the candidate neutral bias current sequence number. For sorting position, The difference value is either the first sequence difference value or the second sequence difference value. For the preset quantity, The first in the boundary time sorting sequence The arrival time of the bit boundary, The first time series to reach the reference boundary The arrival time of the bit boundary, To prevent positive numbers with a denominator of zero.

[0069] The risk value of the first modified boundary and the risk value of the second modified boundary are determined according to the following formula:

[0070] In the formula, The risk substitution value for the first or second correction boundary. The value of the first boundary risk or the value of the second boundary risk. This is the sequence difference correction factor. The sequence difference correction factor is determined based on the historical relationship between the sequence difference value and the safety boundary trigger frequency.

[0071] When the reference boundary arrival time series is missing, no series difference correction is performed, and the corresponding boundary risk value is used as the corrected boundary risk value. The first and second corrected boundary risk values ​​are then passed to the virtual midline impedance mapping process.

[0072] In each control cycle, the neutral-line UPS lithium battery system receives the boundary dominance chain, the first corrected boundary risk value, the second corrected boundary risk value, and neutral-line operation data. The controller identifies the migration location of the dominant cell and the dominance boundary type along the current arrangement order of the candidate neutral-line bias current set, and determines the reference neutral-line bias current by combining the arrival time difference, value difference, and current amplitude of the candidate neutral-line bias current between the first and second battery packs.

[0073] The risk values ​​of the first and second modified boundary are converted into the first virtual neutral impedance and the second virtual neutral impedance, respectively. The reference neutral bias current is corrected by the impedance difference to form the target neutral bias current.

[0074] The controller adjusts the duration of redundant switching states based on the deviation between the target neutral bias current and the actual neutral current, so that the first and second battery packs distribute the load power in the target direction, and transmits the updated boundary approach damage quantity to the next control cycle.

[0075] In one embodiment, a dominance migration penalty is used to limit frequent switching between the dominant cell and the dominance boundary type during changes in the candidate neutral bias current. The controller compares the dominant cells and dominance boundary types associated with adjacent candidate neutral bias currents according to the order of the candidate neutral bias current sets.

[0076] When the dominant cell changes, the dominant cell migration is recorded. When the dominant boundary type switches between voltage safety boundary, temperature safety boundary, and capacity safety boundary, the dominant boundary type migration is recorded. The dominant migration penalty is determined jointly by the two types of migration. When the dominant boundary type migration has a significant impact, a larger penalty coefficient is set for it.

[0077]

[0078] in, To control migration penalties, To dominate the single entity, For the dominant boundary type, To dominate the monomer migration coefficient, For the dominance boundary type migration coefficient, This is an indicator function that takes the value 1 when the condition is true and 0 when the condition is false.

[0079] The above dominance migration penalty formula applies to the candidate neutral bias currents that are the second and subsequent positions in the candidate neutral bias current set; for the candidate neutral bias current that is the first position, its dominance migration penalty is set to zero.

[0080] Regarding the first The cost of selecting one candidate neutral bias current is determined according to the following formula:

[0081] In the formula, To select the cost value; and These are the arrival times of the dominance boundaries of the first and second battery packs, respectively. and These are the risk substitution values ​​of the first and second correction boundary, respectively. To punish those who dominate migration; Candidate neutral bias current; The difference coefficient for arrival time of the dominance boundary; The coefficient of difference in substitution value; The dominant migration penalty coefficient; This is the current amplitude coefficient; To prevent positive values ​​of zero from being entered as logarithms.

[0082] Each coefficient is determined and normalized based on the safety boundary triggering conditions, the change in neutral current, and the load power fluctuation range. The difference in dominance boundary arrival time reflects the time difference between the two sets of dominance cells reaching the safety boundary; the difference in cost value reflects the risk difference between the two sets of near-boundary cells; the dominance migration penalty is used to limit the switching of controlled objects; and the current amplitude is used to limit the neutral bias amplitude.

[0083] The controller compares the selection costs of each candidate neutral bias current and determines the candidate neutral bias current with the lowest selection cost as the reference neutral bias current. When multiple candidate neutral bias currents have the same selection cost, the candidate neutral bias current with the smaller current amplitude and the same direction as the current neutral current is selected.

[0084] If any single-cell boundary margin associated with a candidate neutral bias current is negative, or if the predicted midpoint potential offset exceeds the midpoint potential offset limit, the candidate neutral bias current will not be selected. If all non-zero candidate neutral bias currents are unavailable, the candidate neutral bias current with zero bias current or that does not increase the current midpoint potential offset will be determined as the reference neutral bias current and passed to the virtual neutral impedance correction process.

[0085] In one embodiment, the first virtual neutral impedance and the second virtual neutral impedance convert the first and second corrected boundary risk values ​​into a neutral bias current correction. As the corrected boundary risk value increases, the corresponding virtual neutral impedance increases, allowing the higher-risk battery pack to bear less load power.

[0086] The first virtual neutral impedance, the second virtual neutral impedance, and the target neutral bias current are determined by the following formula:

[0087]

[0088]

[0089]

[0090] In the formula, This is the impedance of the first virtual neutral line; This is the impedance of the second virtual neutral line; Based on the virtual neutral impedance; These are the impedance mapping coefficients; The risk value of the first correction boundary; The risk value of the second correction boundary; To prevent positive values ​​with a denominator of zero; This is the correction amount for the neutral bias current; This is the impedance difference conversion factor; This is the reference neutral bias current; The target neutral bias current; This is a limiting operator used to restrict the input value to the lower limit of the candidate neutral bias current range. Upper limit of candidate neutral bias current range between.

[0091] The basic virtual neutral impedance is determined based on the allowable variation range of the virtual neutral impedance and the calibration relationship between the change in neutral current and the load power distribution changes of the first and second battery packs. The impedance mapping coefficient is determined based on the allowable variation range of the virtual neutral impedance, and the impedance difference conversion coefficient is determined based on the calibration relationship between the change in neutral current and the impedance difference.

[0092] The positive direction of the neutral bias current is to reduce the load power borne by the first battery pack and increase the load power of the second battery pack. When the first correction boundary risk value is large, the correction is to reduce the load power of the first battery pack; when the second correction boundary risk value is large, the correction direction is opposite. When the correction boundary risk values ​​of both sets are zero, the neutral bias current correction is zero.

[0093] After limiting the target neutral bias current, check the offset between the predicted neutral current and the predicted midpoint potential. If either result exceeds the limit, reduce the neutral bias current correction until both results are within the allowable range. The process for determining the adjustment amount for the target neutral bias current to enter the redundant switching state is also described.

[0094] In one embodiment, the neutral-line UPS lithium battery system reads the actual neutral-line current from the neutral-line operation data and determines the difference between the target neutral-line bias current and the actual neutral-line current as the neutral-line current deviation. The three-level inverter selects redundant switching states that synthesize the same output voltage but have different effects on the direction of the neutral-line current, and reallocates the action time of the redundant switching states while keeping the total action time of the output voltage vector constant.

[0095] The adjustment amount for the redundancy switch state action time is determined according to the following formula:

[0096] In the formula, For the first The adjustment amount of the redundant switch state action time of each control cycle; For the first The deviation of the centerline current in each control cycle; This is the proportionality coefficient; The integral coefficient; This refers to the control cycle number in the integral calculation; This is the current control cycle number; To control the cycle, the proportional coefficient is determined based on the actual neutral current's response gain to changes in the application time, and the integral coefficient is determined based on the neutral current's convergence time and steady-state deviation.

[0097] The adjustment amount for the redundant switch state action time is limited by the switching cycle and the minimum effective action time. When the adjustment amount exceeds the allocable action time, it is limited, and the accumulation of neutral current deviation is stopped.

[0098] The controller adjusts the duration of one redundant switching state based on its duration, increasing the duration of that state and decreasing the duration of another redundant switching state that produces the same output voltage, thus converging the actual neutral current towards the target neutral bias current. Once the actual neutral current enters the allowable deviation range, the current duration allocation is maintained. The allowable deviation range is determined based on the neutral current sampling error and the minimum controllable change in neutral current.

[0099] During power supply, the controller determines the degree of proximity to the safety boundary based on the normalized voltage boundary margin, temperature boundary margin, and capacity boundary margin. Current stress is determined based on the ratio of the battery pack current to the rated battery pack current, and the cumulative amplitude is increased when the battery pack current exceeds the rated battery pack current.

[0100] To ensure a consistent comparison of individual boundary margins under different safety boundary types, the ratios between the voltage boundary margin, temperature boundary margin, and capacity boundary margin and their corresponding reference boundary margins are calculated, and each ratio is constrained between 0 and 1 to obtain the normalized voltage boundary margin, normalized temperature boundary margin, and normalized capacity boundary margin. The degree of proximity of the safety boundaries is determined according to the following formula:

[0101] in, To the degree of proximity to the safety boundary, , and These are the three types of normalized monomer boundary margins.

[0102] The boundary approach damage quantity is updated jointly by the boundary approach damage quantity, safe boundary approach degree, and current stress from the previous control cycle. The boundary approach damage quantity from the previous control cycle decays according to the historical damage retention coefficient, while the safe boundary approach degree and current stress are accumulated according to their respective coefficients. The historical damage retention coefficient is determined based on the decay period of the boundary approach damage quantity, while the safe boundary approach degree coefficient and current stress coefficient are calibrated based on the individual unit temperature rise, individual unit voltage drop, and safe boundary triggering records.

[0103]

[0104] in, The current control cycle boundary is close to the damage amount. Historical damage retention coefficient, The proximity coefficient to the safety boundary. The current stress coefficient, This is the ratio between the actual battery pack current and the rated battery pack current of the corresponding battery pack.

[0105] The updated first battery pack boundary proximity damage value is used to determine the first boundary risk cost value in the next control cycle, and the updated second battery pack boundary proximity damage value is used to determine the second boundary risk cost value in the next control cycle, forming a feedback closed loop.

[0106] In this embodiment, the neutral-line UPS lithium battery system uses a first battery pack and a second battery pack to form a split DC bus. Both the first and second battery packs are composed of 120 lithium iron phosphate cells connected in series, with a rated capacity of 100Ah per cell. The neutral-line UPS lithium battery system is in battery-powered mode with a load power of 60.0kW, a control cycle of 10ms, and a cell status data sampling period of 100ms.

[0107] At sampling time 12:00:00.000, the positive bus voltage is 378V, the negative bus voltage is 382V, the positive bus current is 78.9A, the negative bus current is 77.3A, the neutral current is 1.60A, and the current neutral point potential offset is -4.0V. The neutral current limit is set to 10A, and the neutral point potential offset limit is set to 8V.

[0108] This control cycle performs calculations on the individual cell boundary margins and boundary arrival times for all cells in the first and second battery packs. Figure 2 Only A1-037, A1-052, and A1-061 from the first battery pack, ranked by boundary arrival time, and B2-084, B2-019, and B2-093 from the second battery pack, ranked by boundary arrival time, are shown in this embodiment. The cells used in the illustrations are referred to as critical cells. A1-037 has a cell voltage of 3.080V, a cell temperature of 43.1℃, a cell state of charge of 16.8%, a cell internal resistance of 2.80mΩ, a cell polarization voltage of 21mV, and a boundary approach damage of 0.46. A1-052 has a cell voltage of 3.112V, a cell temperature of 45.0℃, a cell state of charge of 21.3%, a cell internal resistance of 2.50mΩ, a cell polarization voltage of 18mV, and a boundary approach damage of 0.34. The A1-061 has a single-cell voltage of 3.126V, a single-cell temperature of 41.8℃, a single-cell state of charge of 22.7%, a single-cell internal resistance of 2.35mΩ, a single-cell polarization voltage of 16mV, and a boundary near-damage of 0.22.

[0109] The individual cell voltage of B2-084 is 3.130V, the individual cell temperature is 39.4℃, the individual cell state of charge is 18.7%, the individual cell internal resistance is 2.20mΩ, the individual cell polarization voltage is 15mV, and the boundary approach damage is 0.28. The individual cell voltage of B2-019 is 3.146V, the individual cell temperature is 44.2℃, the individual cell state of charge is 23.1%, the individual cell internal resistance is 2.30mΩ, the individual cell polarization voltage is 16mV, and the boundary approach damage is 0.24. The individual cell voltage of B2-093 is 3.151V, the individual cell temperature is 40.6℃, the individual cell state of charge is 24.0%, the individual cell internal resistance is 2.10mΩ, the individual cell polarization voltage is 14mV, and the boundary approach damage is 0.15.

[0110] like Figure 2 As shown in the figure, each row corresponds to a key cell in the first and second battery packs, and each column represents the cell number, cell voltage, cell temperature, cell state of charge, cell internal resistance, cell polarization voltage, and boundary proximity damage, respectively. Figure 2 The lower section records the sampling time, load power, neutral current, and current neutral point potential offset. This data is generated from battery management system sampling data and neutral operation data collected at the same sampling time, and is used to determine candidate neutral bias currents and cell boundary margins.

[0111] Based on the current neutral current, neutral current limit, current midpoint potential offset, and midpoint potential offset limit, the candidate neutral bias current range is determined to be -6A to +6A. The preset bias interval is set to 3A, thus forming five candidate neutral bias currents: -6A, -3A, 0A, +3A, and +6A. The positive direction of the candidate neutral bias current indicates a decrease in the load power borne by the first battery pack and an increase in the load power borne by the second battery pack.

[0112] When the candidate neutral bias current is -6A, the first predicted battery pack current is 82.0A and the second predicted battery pack current is 74.4A; when the candidate neutral bias current is -3A, the first predicted battery pack current is 80.5A and the second predicted battery pack current is 75.9A; when the candidate neutral bias current is 0A, the first predicted battery pack current is 79.0A and the second predicted battery pack current is 77.4A; when the candidate neutral bias current is +3A, the first predicted battery pack current is 77.5A and the second predicted battery pack current is 78.9A; when the candidate neutral bias current is +6A, the first predicted battery pack current is 76.0A and the second predicted battery pack current is 80.4A.

[0113] like Figure 3As shown, the horizontal axis represents the candidate neutral bias current, and the vertical axis represents the predicted battery pack current. The red curve represents the first predicted battery pack current, and the blue curve represents the second predicted battery pack current. The green dashed line corresponds to the +3A candidate neutral bias current, and the green ring marks the first and second predicted battery pack currents under this candidate neutral bias current. Figure 3 The upper part simultaneously records the neutral current limit, the neutral point potential offset limit, and the preset bias interval.

[0114] The prediction time domain is set to 900s, the prediction step size to 5s, and the preset observation time to 600s. The minimum single-cell voltage is set to 2.85V, the maximum single-cell temperature to 55℃, and the minimum allowable single-cell state of charge to 10%. Based on the first and second predicted battery pack currents, the controller recursively calculates the single-cell voltage, single-cell temperature, single-cell state of charge, and single-cell polarization voltage of each cell, and thereby determines the voltage boundary margin, temperature boundary margin, and capacity boundary margin.

[0115] When the candidate neutral bias current is -6A, the dominant cell of the first battery pack is A1-037, the dominant boundary type is capacity safety boundary, and the boundary arrival time is 220s; the dominant cell of the second battery pack is B2-084, the dominant boundary type is capacity safety boundary, and the boundary arrival time is 690s.

[0116] When the candidate neutral bias current is -3A, the dominant cell of the first battery pack is A1-037, the dominant boundary type is capacity safety boundary, and the boundary arrival time is 290s; the dominant cell of the second battery pack is B2-084, the dominant boundary type is capacity safety boundary, and the boundary arrival time is 610s.

[0117] When the candidate neutral bias current is 0A, the dominant cell of the first battery pack is A1-037, the dominant boundary type is voltage safety boundary, and the boundary arrival time is 360s; the dominant cell of the second battery pack is B2-084, the dominant boundary type is capacity safety boundary, and the boundary arrival time is 500s.

[0118] When the candidate neutral bias current is +3A, the dominant cell of the first battery pack is A1-037, the dominant boundary type is voltage safety boundary, and the boundary arrival time is 445s; the dominant cell of the second battery pack is B2-084, the dominant boundary type is voltage safety boundary, and the boundary arrival time is 420s. The difference in arrival time between the two dominant boundaries is 25s.

[0119] When the candidate neutral bias current is +6A, the dominant cell of the first battery pack migrates to A1-052, the dominant boundary type migrates to the temperature safety boundary, and the boundary arrival time is 530s; the dominant cell of the second battery pack migrates to B2-019, the dominant boundary type migrates to the temperature safety boundary, and the boundary arrival time is 285s.

[0120] like Figure 4 As shown, the horizontal axis represents the candidate midline bias current, and the vertical axis represents the dominance boundary arrival time. The red curve represents the dominance boundary arrival time of the first battery pack, and the blue curve represents the dominance boundary arrival time of the second battery pack. The label boxes next to each node record the dominant cell and the dominance boundary type. The horizontal dashed line represents the preset observation time of 600s, and the green dotted line corresponds to the +3A candidate midline bias current.

[0121] Under a +3A candidate neutral bias current, the arrival times of each individual boundary are sorted in ascending order according to their shortest boundary arrival times. The first boundary time sequence is 445s, 510s, 575s, and 650s, and the second boundary time sequence is 420s, 485s, 560s, and 630s. The reference boundary arrival time sequence is formed from historical power supply data where the load power is stable and the safety boundary has not been triggered, and its values ​​are 470s, 520s, 580s, and 640s.

[0122] like Figure 5 As shown, red nodes represent the first boundary time sequence, blue nodes represent the second boundary time sequence, and orange nodes represent the reference boundary arrival time sequence. The horizontal axis represents the boundary arrival time. The difference between the first boundary time sequence and the reference boundary arrival time sequence is 4.91 × 10^-5, and the difference between the second boundary time sequence and the reference boundary arrival time sequence is 1.20 × 10^-4.

[0123] In this embodiment, the preset quantity is 4. The arrival time of the fourth boundary in the sorting sequence is used to characterize the overall sorting status of non-near-boundary units and does not participate in the accumulation of boundary risk cost of the first or second near-boundary unit set.

[0124] The first set of near-boundary units consists of A1-037, A1-052, and A1-061, with boundary arrival times of 445s, 510s, and 575s, respectively. Their normalized boundary approximation rates for calculating boundary risk value are 0.42, 0.28, and 0.20, respectively. The second set of near-boundary units consists of B2-084, B2-019, and B2-093, with boundary arrival times of 420s, 485s, and 560s, respectively. Their normalized boundary approximation rates for calculating boundary risk value are 0.35, 0.26, and 0.18, respectively.

[0125] Under a candidate neutral bias current of +3A, the safety boundary types first encountered by A1-037, A1-052, and A1-061 are voltage safety boundary, voltage safety boundary, and capacity safety boundary, respectively; while the safety boundary types first encountered by B2-084, B2-019, and B2-093 are voltage safety boundary, capacity safety boundary, and capacity safety boundary, respectively. The boundary type weights for voltage safety boundary, temperature safety boundary, and capacity safety boundary are set to 0.2523, 0.2200, and 0.1573, respectively; the boundary approach rate coefficient is set to 0.5; the boundary approach damage coefficient is set to 0.4; and the preset observation time is 600s.

[0126] Based on the boundary arrival time, normalized boundary approach rate, safe boundary type, and boundary approach damage of each individual, the risk cost of the first boundary is 0.384, and the risk cost of the second boundary is 0.322. With the sequence difference correction coefficient set to 800, after sequence difference correction, the risk cost of the first corrected boundary is 0.399, and the risk cost of the second corrected boundary is 0.353.

[0127] For the five candidate neutral bias currents, the selected cost is formed by summing the difference in arrival time of the dominance boundary, the difference in risk cost of the correction boundary, the dominance migration penalty, and the current amplitude. The four values ​​for the -6A candidate neutral bias current are 0.330, 0.270, 0, and 0.116, with a selection cost of 0.716; the four values ​​for the -3A candidate neutral bias current are 0.140, 0.120, 0, and 0.063, with a selection cost of 0.323; the four values ​​for the 0A candidate neutral bias current are 0.035, 0.015, 0, and 0.020, with a selection cost of 0.070; the four values ​​for the +3A candidate neutral bias current are 0.008, 0.006, 0, and 0.013, with a selection cost of 0.027; and the four values ​​for the +6A candidate neutral bias current are 0.075, 0.110, 0.030, and 0.073, with a selection cost of 0.288.

[0128] like Figure 6 As shown, the five stacked bars correspond to five candidate neutral bias currents. Different colored areas represent the difference in arrival time of the dominance boundary, the difference in risk cost of the correction boundary, the penalty for dominance migration, and the current amplitude, respectively. The +3A candidate neutral bias current has the lowest selection cost and is therefore determined as the reference neutral bias current.

[0129] The basic virtual neutral impedance is set to 0.18Ω, and the impedance mapping coefficient is set to 0.6. Based on the first corrected boundary risk value of 0.399, the unrounded calculated value of the first virtual neutral impedance is 0.2373Ω. Figure 6The value is displayed to three decimal places as 0.237Ω; based on the second corrected boundary risk value of 0.353, the unrounded calculated value of the second virtual neutral impedance is 0.2307Ω. Figure 6 The value is displayed to three decimal places as 0.231Ω. The impedance difference calculated by the controller using the unrounded value is approximately 0.0066Ω. The impedance difference conversion factor is set to 18.2A / Ω, resulting in a neutral bias current correction of approximately +0.12A. After correction, the target neutral bias current is 3.12A.

[0130] The actual neutral current is 1.60A, and the deviation between the target neutral bias current and the actual neutral current is 1.52A. The proportional coefficient is set to 3.0μs / A. The proportional term corresponding to the neutral current deviation of 1.52A is 4.56μs, and the integral accumulation term of the current control cycle is 0.64μs. Together, they form a redundant switching state action time adjustment of 5.20μs.

[0131] In a three-level inverter, a pair of redundant switches that synthesize the same output voltage but have different effects on the direction of the neutral current have a positive action time and a negative action time of 24.0 μs when the control cycle number is 0. After the action time adjustment is completed, the positive action time of control cycles 1 to 5 is 29.2 μs and the negative action time is 18.8 μs, with the total action time remaining at 48.0 μs.

[0132] like Figure 7 As shown, the stacked bars represent the positive and negative action times within each control cycle, the green curve represents the actual neutral current, and the horizontal dashed line represents the target neutral bias current of 3.12A. The actual neutral currents in control cycles 0 to 5 are 1.60A, 2.15A, 2.58A, 2.91A, 3.08A, and 3.12A, respectively, thereby completing the neutral current regulation and the redistribution of load power between the first and second battery packs.

[0133] After the current control cycle is completed, the boundary proximity damage value of A1-037 is updated from 0.460 to 0.471, and the boundary proximity damage value of B2-084 is updated from 0.280 to 0.296. The updated boundary proximity damage values ​​are used to determine the first boundary risk cost value and the second boundary risk cost value in the next control cycle, thus forming a closed loop between candidate neutral bias current selection, neutral current adjustment, and boundary proximity damage value write-back.

[0134] In this embodiment, the historical damage retention coefficient is set to 0.95, the safety boundary proximity coefficient is set to 0.03, and the current stress coefficient is set to 0.02. The safety boundary proximity of A1-037 is 0.72, and the current stress is 0.60; the safety boundary proximity of B2-084 is 0.59, and the current stress is 0.62.

[0135] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A neutral-line UPS lithium battery balanced power supply method, characterized in that, The method includes: Acquire the individual cell status data, neutral line operation data and load power of the first and second battery packs to determine the candidate set of neutral line bias currents; Based on the candidate neutral bias current set, the individual cell state data, and the load power, a boundary dominance chain and a boundary time sorting sequence are generated. The boundary dominance chain associates the dominating cell, dominance boundary type, boundary arrival time, and boundary approach rate under each candidate neutral bias current. The modified boundary risk cost is determined based on the boundary dominance chain and the boundary time sorting sequence. The modified boundary risk cost is mapped to a virtual neutral impedance. The target neutral bias current is determined by the virtual neutral impedance. Adjust the neutral current according to the target neutral bias current to distribute the load power of the first battery pack and the second battery pack.

2. The method according to claim 1, characterized in that, The individual cell status data includes individual cell voltage, individual cell temperature, individual cell state of charge, individual cell internal resistance, and individual cell polarization voltage. The neutral line operation data includes neutral line current, positive bus voltage, negative bus voltage, positive bus current, and negative bus current. The individual cell status data, the neutral line operation data, and the load power are used together to determine the individual cell boundary margin under each candidate neutral line bias current.

3. The method according to claim 1, characterized in that, The determination of the candidate neutral bias current set includes: The current centerline current and the current centerpoint potential offset are determined from the centerline operation data. The candidate neutral bias current range is determined based on the neutral current limit, the neutral point potential offset limit, the current neutral current, and the current neutral point potential offset. Multiple candidate neutral bias currents are generated in the candidate neutral bias current range according to a preset bias interval and arranged in the current direction, and the multiple candidate neutral bias currents form the candidate neutral bias current set.

4. The method according to claim 2, characterized in that, The determination of the individual unit boundary margin includes: Based on the candidate neutral bias current set and the load power, the first predicted battery pack current and the second predicted battery pack current under each candidate neutral bias current are determined respectively for the first battery pack and the second battery pack. Based on the first predicted battery pack current, the second predicted battery pack current, and the individual cell state data, the voltage boundary margin, temperature boundary margin, and capacity boundary margin of each individual cell in the first battery pack and the second battery pack are determined respectively, and the individual cell boundary margin is formed by the voltage boundary margin, the temperature boundary margin, and the capacity boundary margin.

5. The method according to claim 4, characterized in that, The generated boundary dominance chain includes: Within a preset prediction time domain, the boundary arrival time and boundary approach rate of each individual cell under each candidate neutral bias current are predicted based on the individual cell boundary margin. The boundary approach rate is the rate at which the individual cell boundary margin decreases with the prediction time. For each candidate centerline bias current, the cell with the shortest boundary arrival time in the first battery pack and the second battery pack is determined as the dominant cell, and the safe boundary first touched by each dominant cell is determined as the dominant boundary type. According to the current arrangement order in the candidate midline bias current set, each candidate midline bias current is associated with the dominant cell, the dominant boundary type, the boundary arrival time, and the boundary approach rate to generate the boundary domination chain.

6. The method according to claim 5, characterized in that, The determination of the risk value of the modified boundary includes: Based on the boundary arrival time of each cell, a first near-boundary cell set and a second near-boundary cell set are determined respectively. The first near-boundary cell set and the second near-boundary cell set are respectively composed of cells in the corresponding battery pack whose boundary arrival time is less than the preset observation time, and are respectively associated with the safety boundary type first touched by each cell. Obtain the boundary proximity damage amount for each individual unit; The first boundary risk cost is determined based on the boundary arrival time, boundary approach rate, safe boundary type, and boundary approach damage amount associated with each individual in the first near-boundary individual set, and the second boundary risk cost is determined based on the boundary arrival time, boundary approach rate, safe boundary type, and boundary approach damage amount associated with each individual in the second near-boundary individual set.

7. The method according to claim 6, characterized in that, The step of determining the corrected boundary risk value based on the boundary time sorting sequence includes: For each candidate midline bias current, determine the shortest boundary arrival time of each cell in the first battery pack and the second battery pack under different safety boundary types. The individual cells in the first battery pack and the second battery pack are sorted according to the shortest boundary arrival time to obtain a first boundary time sorting sequence and a second boundary time sorting sequence. The boundary time sorting sequence is composed of the first boundary time sorting sequence and the second boundary time sorting sequence. A first sequence difference value is determined based on the difference between the arrival times of the first boundary in the first boundary time sorting sequence and the reference boundary arrival time sequence, and a second sequence difference value is determined based on the difference between the arrival times of the first boundary in the second boundary time sorting sequence and the reference boundary arrival time sequence. The first boundary risk value is obtained by correcting the first sequence difference value, and the second boundary risk value is obtained by correcting the second sequence difference value.

8. The method according to claim 7, characterized in that, The determination of the target neutral bias current and virtual neutral impedance includes: The dominance migration penalty is determined based on the dominance cell variation and dominance boundary type variation between adjacent candidate neutral bias currents. The reference neutral bias current is determined from the set of candidate neutral bias currents based on the difference in arrival time of the dominance boundary between the first battery pack and the second battery pack under each candidate neutral bias current, the difference in risk value between the first modified boundary and the second modified boundary, the dominance migration penalty, and the current amplitude of each candidate neutral bias current. The virtual neutral impedance is determined based on the first modified boundary risk value and the second modified boundary risk value, and the reference neutral bias current is modified based on the virtual neutral impedance to obtain the target neutral bias current.

9. The method according to claim 8, characterized in that, The virtual neutral impedance includes a first virtual neutral impedance and a second virtual neutral impedance. The first virtual neutral impedance is determined based on the first modified boundary risk value, and the second virtual neutral impedance is determined based on the second modified boundary risk value. When the modified boundary risk value increases, the corresponding virtual neutral impedance increases. The neutral bias current correction amount is determined based on the impedance difference between the first virtual neutral impedance and the second virtual neutral impedance, and the reference neutral bias current is corrected based on the neutral bias current correction amount to obtain the target neutral bias current.

10. The method according to claim 9, characterized in that, The adjustment of the neutral current includes: The adjustment amount of the redundant switch state action time is determined based on the deviation between the target neutral bias current and the actual neutral current determined by the neutral operation data. Based on the adjustment amount of the redundant switch state action time, adjust the action time of the redundant switch states in the three-level inverter that synthesize the same output voltage and have different effects on the direction of the neutral current, so that the actual neutral current converges to the target neutral bias current. The safety boundary proximity is determined based on the voltage boundary margin, temperature boundary margin, and capacity boundary margin of each cell. The current stress is determined based on the ratio between the actual battery pack current and the rated battery pack current of each cell. The boundary proximity damage is updated based on the safety boundary proximity and the current stress, and the updated boundary proximity damage is used to determine the first boundary risk cost value and the second boundary risk cost value in the next control cycle.

Images