Industrial and commercial energy storage battery active equalization method and storage medium

Abstract

The invention relates to an industrial and commercial energy storage battery active equalization method and a storage medium. The method comprises the following steps: constructing a battery safe working area, collecting dynamic characteristic parameters of each battery unit in real time, training a working condition migration model by using a long short-term memory network, and predicting a unit working condition vector at the next moment; and mapping the current working condition vector to a global feature space to generate a dynamic working condition point, and calculating a working condition migration direction. Determining an equalization critical point by combining a safe working area of the battery, and further calculating equalization potential energy and global unbalance degree of each unit; when all the working condition points are located in a safety area and the global unbalance degree exceeds a threshold value, the electric charge quantity is transferred through a bidirectional DC-DC or a switch matrix; if a working condition point outside the safe working area of the battery exists, the input current is reduced, and the parallel branch balance impedance is switched if necessary. Through dynamic working condition prediction and potential energy quantification, the system safety is improved accurately and actively in a balanced mode, and the problem of battery inconsistency is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of battery balancing, and more particularly to an active balancing method and storage medium for industrial and commercial energy storage batteries. Background Technology

[0002] As the scale of commercial and industrial energy storage systems expands, the performance imbalance of individual cells within battery packs due to differences in manufacturing processes, aging levels, and operating environments is becoming increasingly prominent. Existing active balancing technologies primarily rely on static thresholds based on voltage or state of charge (SOC) to redistribute energy through passive response strategies. However, this type of technology has significant drawbacks:

[0003] Limitations of single-dimensional parameters: Using only voltage or SOC as the basis for balancing ignores the coupling effects of key characteristic parameters such as temperature gradient, internal resistance drift, and dynamic current path voltage drop, resulting in a disconnect between balancing decisions and the actual safety boundary of the battery.

[0004] Lack of predictability of operating conditions: Traditional methods cannot predict battery migration trends and only perform hysteresis compensation after imbalance occurs, making it difficult to prevent the battery pack from entering dangerous operating conditions such as overvoltage, overheating or sudden changes in internal resistance.

[0005] Insufficient strategy adaptability: Existing technologies use fixed thresholds to trigger equalization, which cannot distinguish the differentiated control needs of battery packs and is prone to ineffective equalization or safety risks.

[0006] The aforementioned defects lead to two major technical challenges for energy storage systems: first, the lag in balancing exacerbates the risk of battery aging and even thermal runaway; second, the timing of balancing actions and the amount of energy transferred lack scientific basis, resulting in system energy efficiency loss and reduced lifespan. Summary of the Invention

[0007] To address the shortcomings of existing technologies, embodiments of the present invention provide an active balancing method for industrial and commercial energy storage batteries, comprising the following steps: collecting historical operating condition data of the energy storage battery pack, and constructing a battery safe operating area based on the historical operating condition data;

[0008] The basic characteristic parameters of each battery cell are collected in real time at each time point within a preset time window. Based on the basic characteristic parameters, the dynamic characteristic parameters of each battery cell at each time point within the preset time window are obtained. The dynamic characteristic parameters are then arranged in ascending order of time to generate a dynamic characteristic sequence of each battery cell.

[0009] A long short-term memory network model trained based on dynamic feature sequences is used as a working condition transfer model.

[0010] The dynamic feature parameters of each battery cell at the current moment are extracted to generate the cell operating condition vector of each battery cell at the current moment. After normalization, the cell operating condition vector is mapped to the global feature space to generate the dynamic operating condition point of each battery cell at the current moment. The origin of the global feature space represents that all battery cells of the battery pack are in an ideal equilibrium state.

[0011] Input the cell operating condition vector of each battery cell at the current time into the operating condition transfer model to output the cell operating condition vector of each battery cell at the next time.

[0012] The difference between the current unit working condition vector and the next unit working condition vector is taken as the working condition transition vector, and the direction of the working condition transition vector is taken as the working condition transition direction.

[0013] If the dynamic operating point is located within the battery safe operating area, the point with the smallest Euclidean distance at the intersection of the ray along the operating condition migration direction starting from the dynamic operating point and the boundary of the battery safe operating area is selected as the equilibrium critical point; if the dynamic operating point is located outside the battery safe operating area, the point closest to the boundary of the battery safe operating area from the dynamic operating point is selected as the equilibrium critical point.

[0014] Calculate the normalized Euclidean distance from the dynamic operating point to the equilibrium critical point, and calculate the equilibrium potential energy of each battery cell based on the normalized Euclidean distance; obtain the maximum value and standard deviation of the equilibrium potential energy of all battery cells, and then perform a weighted summation of the maximum value and standard deviation to obtain the global imbalance degree.

[0015] When all dynamic operating points are within the battery safe operating area and the global imbalance is greater than the preset imbalance threshold, the following steps are performed: select the battery cell with the smallest equalization potential energy as the equalization target cell and the battery cell with the largest equalization potential energy as the equalization source cell.

[0016] The amount of charge to be transferred is determined based on the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit, and the amount of charge to be transferred from the equilibrium source unit to the equilibrium target unit is transferred through a bidirectional DC-DC converter or a switching matrix.

[0017] When a dynamic operating point exists outside the battery's safe operating range, perform the following steps:

[0018] The input current of the energy storage battery pack is reduced based on the equilibrium potential energy. If, after adjusting the input current, there is still a dynamic operating point outside the safe operating area of ​​the battery, the parallel branch impedance is switched to balance the impedance.

[0019] According to a preferred embodiment, the basic characteristic parameters include SOC value, voltage value, temperature value, and internal resistance value; the dynamic characteristic parameters include temperature rise rate, charging current integral, current path voltage drop, voltage drop amplitude, internal resistance change rate, and resting period self-discharge rate.

[0020] According to a preferred embodiment, constructing a battery safety operating area based on historical operating condition data includes:

[0021] Traverse the dynamic feature parameters and use the dynamic feature parameter being traversed as the target dynamic feature parameter.

[0022] The target dynamic characteristic parameter value is gradually increased or decreased according to the preset boundary test step size until the battery cell triggers an abnormal state such as over-limit voltage, temperature alarm or internal resistance change.

[0023] Record the values ​​of the target's dynamic characteristic parameters when an abnormal state is triggered, and use them as the safe boundary points of the target's dynamic characteristic parameters;

[0024] Repeat the above steps until the safe boundary point corresponding to each dynamic feature parameter is obtained;

[0025] Map all safety boundary points to a global feature space with dynamic feature parameters as coordinate axes;

[0026] Connecting all mapped safe boundary points forms a closed multidimensional convex polyhedron boundary.

[0027] Verify the boundary continuity of the multidimensional convex polyhedron boundary; if the Euclidean distance between adjacent safe boundary points exceeds the preset tolerance, then insert a linear interpolation point between the two points.

[0028] The space enclosed by the boundary of the multidimensional convex polyhedron is defined as the initial battery safe operating area.

[0029] According to a preferred embodiment, constructing a battery safety operating area based on historical operating condition data includes:

[0030] Several historical operating condition points are identified by using historical operating condition data. Each historical operating condition point is constructed from the voltage, temperature, current, SOC and internal resistance of a single battery cell at the same time.

[0031] Based on preset voltage threshold, temperature threshold, internal resistance deviation threshold and BMS alarm records, historical operating points are filtered: historical operating points that trigger any abnormal condition are abnormal operating points, and historical operating points that are in the standard charging and discharging range and have no alarm records are normal operating points.

[0032] The number of normal operating points that are within the initial battery safe operating area is counted, and the proportion of the number of normal operating points is calculated. If the proportion is less than a preset proportion threshold, it is determined that the initial battery safe operating area has not passed the first verification.

[0033] Check if any abnormal operating point is located within the initial battery safe operating area. If so, the initial battery safe operating area is deemed to have failed the second verification.

[0034] If the initial battery safe working area fails either the first verification or the second verification, the boundary test step size is halved, and the safe boundary point generation process is re-executed based on the halved boundary test step size, thereby reconstructing the initial battery safe working area using the newly generated safe boundary points.

[0035] Repeat the above steps of generating safety boundary points, constructing the initial battery safe working area, and verifying the initial battery safe working area until the reconstructed initial battery safe working area passes both the first and second verifications. At this point, it is determined as the final battery safe working area.

[0036] According to a preferred embodiment, determining the amount of transferred charge based on the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit includes:

[0037] Calculate the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit;

[0038] Calculate the capacity correction factor based on the rated capacity and SOC value of the target unit;

[0039] Calculate the voltage difference between the equalization source unit and the equalization target unit;

[0040] The amount of basic transferred charge is calculated based on the potential energy difference, capacity correction factor, and voltage difference.

[0041] Obtain the preset safety factor and maximum transferred charge, and determine the final transferred charge based on the base transferred charge, safety factor, and maximum transferred charge;

[0042] If the voltage difference is less than the voltage difference threshold, the final transferred charge is transferred from the equalization source unit to the equalization target unit through the switching matrix; otherwise, the final transferred charge is transferred from the equalization source unit to the equalization target unit through bidirectional DC-DC.

[0043] After the charge transfer is completed, update the SOC and voltage values ​​of the equalization source cell and the equalization target cell.

[0044] According to a preferred embodiment, reducing the input current of the energy storage battery pack based on the equilibrium potential energy includes:

[0045] Battery cells whose dynamic operating point exceeds the boundary of the battery safe operating area are recorded as out-of-bounds cells.

[0046] Calculate the average equilibrium potential energy of all out-of-bounds units;

[0047] Calculate the Euclidean distance between the origin of the global feature space and each safety boundary point of the battery safety working area, and use the minimum Euclidean distance as the boundary threshold;

[0048] The current adjustment ratio is calculated based on the average equilibrium potential energy and the boundary threshold.

[0049] Calculate the target current to be adjusted based on the current input current and the current adjustment ratio;

[0050] The input current is adjusted to the target current in a stepwise manner using a preset current change rate, and the dynamic operating point of the out-of-range unit is monitored in real time.

[0051] If all dynamic operating points return to the battery's safe operating range after current adjustment, then maintain operation at the target current.

[0052] If there are still out-of-bounds units, then perform a switch to balance the impedance of the parallel branches.

[0053] According to a preferred embodiment, the specific steps for performing the switching parallel branch impedance balancing operation include:

[0054] Identify all parallel branches containing out-of-bounds units and form a set of branches to be adjusted.

[0055] For each parallel branch in the set of branches to be adjusted, perform the following operations:

[0056] The balance impedance value of the parallel branch is increased by a preset impedance step size;

[0057] The dynamic operating point of each battery cell in the parallel branch is updated in real time;

[0058] Calculate the equilibrium potential energy of each battery cell in the parallel branch;

[0059] If the dynamic operating point of all battery cells in the parallel branch is within the safe operating range of the battery, then stop increasing the balancing impedance of the parallel branch and record the current impedance value.

[0060] If the balance impedance of the parallel branch increases to the preset limit impedance, and there are still battery cells in the parallel branch whose dynamic operating point exceeds the boundary of the battery safe operating area, then the corresponding parallel branch will be marked as a faulty branch.

[0061] Synchronously adjust the balance impedance value of the non-adjusted branch to keep the total input current of the energy storage battery pack at the adjustment target current;

[0062] If a faulty branch exists, the protection shutdown mechanism of the energy storage battery pack will be triggered;

[0063] If the dynamic operating points of all out-of-bounds units return to the battery's safe operating range, then the current balanced impedance values ​​of each branch will be maintained.

[0064] According to a preferred embodiment, training a long short-term memory network model based on dynamic feature sequences as a working condition transfer model includes:

[0065] Extract the dynamic feature parameters of each battery cell at each time point from the dynamic feature sequence, and generate the cell operating condition vector of each battery cell at each time point based on the dynamic feature parameters of each battery cell at each time point;

[0066] The cell condition vector of the same battery cell at the previous moment is used as the input sample, and the cell condition vector at the next moment is used as the output label to form a training sample pair.

[0067] Construct a long short-term memory network model, wherein: the number of neurons in the input layer is equal to the feature dimension of the unit condition vector; the number of neurons in the output layer is the same as the number of neurons in the input layer.

[0068] The training sample pairs are input into the Long Short-Term Memory network model in chronological order, with the previous time step unit load vector as the input and the next time step unit load vector as the true label. The network parameters are optimized through the backpropagation algorithm to minimize the mean square error between the predicted load vector and the true label.

[0069] Training will stop when any of the following conditions are met: the prediction error on the validation set decreases by less than a preset convergence threshold for several consecutive training cycles; or the total number of training cycles reaches a preset maximum number of iterations.

[0070] The present invention also provides a storage medium storing a program that can be read and written, wherein the program, when executed, implements the above-described active battery balancing method.

[0071] The present invention has the following beneficial effects:

[0072] 1. This application dynamically constructs a battery safe operating area and monitors the offset state of each battery cell in the feature space in real time to accurately identify potential risks. Based on migration trends, preventive regulation is implemented to reduce the risk of thermal runaway and simultaneously reduce irreversible damage caused by the battery operating outside the battery safe operating area.

[0073] 2. This application utilizes the direction of operating condition migration combined with dynamic characteristic sequence and global imbalance analysis to accurately locate the source and target cells that need to be optimized, and determine the optimal amount of transferred charge, avoiding the blind energy dissipation of traditional equalization. Moreover, the equalization process is gentle and gradual, reducing the impact on the battery. Attached Figure Description

[0074] Figure 1 A flowchart of an active balancing method for industrial and commercial energy storage batteries is provided as an exemplary embodiment. Detailed Implementation

[0075] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0076] The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0077] It should be understood that although the terms first, second, third, etc., may be used in this invention to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of this invention, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0078] See Figure 1 The active balancing method for industrial and commercial energy storage batteries according to the present invention includes:

[0079] S1. Collect historical operating condition data of the energy storage battery pack and construct a battery safety working area based on the historical operating condition data.

[0080] Optionally, the energy storage battery pack consists of several battery cells.

[0081] In a preferred embodiment, constructing the battery safety operating area based on historical operating condition data includes:

[0082] Traverse the dynamic feature parameters and use the dynamic feature parameter being traversed as the target dynamic feature parameter.

[0083] The target dynamic characteristic parameter value is gradually increased or decreased according to the preset boundary test step size until the battery cell triggers an abnormal state such as over-limit voltage, temperature alarm or internal resistance change.

[0084] Record the values ​​of the target's dynamic characteristic parameters when an abnormal state is triggered, and use them as the safe boundary points of the target's dynamic characteristic parameters;

[0085] Repeat the above steps until the safe boundary point corresponding to each dynamic feature parameter is obtained;

[0086] Map all safety boundary points to a global feature space with dynamic feature parameters as coordinate axes;

[0087] Connecting all mapped safe boundary points forms a closed multidimensional convex polyhedron boundary.

[0088] Verify the boundary continuity of the multidimensional convex polyhedron boundary; if the Euclidean distance between adjacent safe boundary points exceeds the preset tolerance, then insert a linear interpolation point between the two points.

[0089] The space enclosed by the boundary of the multidimensional convex polyhedron is defined as the initial battery safe operating area.

[0090] Furthermore, the battery safety working area is constructed based on historical operating condition data, including:

[0091] Several historical operating condition points are identified by using historical operating condition data. Each historical operating condition point is constructed from the voltage, temperature, current, SOC and internal resistance of a single battery cell at the same time.

[0092] Based on preset voltage threshold, temperature threshold, internal resistance deviation threshold and BMS alarm records, historical operating points are filtered: historical operating points that trigger any abnormal condition are abnormal operating points, and historical operating points that are in the standard charging and discharging range and have no alarm records are normal operating points.

[0093] The number of normal operating points that are within the initial battery safe operating area is counted, and the proportion of the number of normal operating points is calculated. If the proportion is less than a preset proportion threshold, it is determined that the initial battery safe operating area has not passed the first verification.

[0094] Check if any abnormal operating point is located within the initial battery safe operating area. If so, the initial battery safe operating area is deemed to have failed the second verification.

[0095] If the initial battery safe working area fails either the first verification or the second verification, the boundary test step size is halved, and the safe boundary point generation process is re-executed based on the halved boundary test step size, thereby reconstructing the initial battery safe working area using the newly generated safe boundary points.

[0096] Repeat the above steps of generating safety boundary points, constructing the initial battery safe working area, and verifying the initial battery safe working area until the reconstructed initial battery safe working area passes both the first and second verifications. At this point, it is determined as the final battery safe working area.

[0097] Optionally, dynamic characteristic parameters include temperature rise rate, charging current integral, current path voltage drop, voltage drop amplitude, internal resistance change rate, and resting self-discharge rate.

[0098] Optionally, the preset tolerance, preset ratio threshold, and boundary test step size are preset according to the actual situation.

[0099] S2. Collect the basic characteristic parameters of each battery cell at each time point within a preset time window in real time. Based on the basic characteristic parameters, obtain the dynamic characteristic parameters of each battery cell at each time point within the preset time window. Arrange the dynamic characteristic parameters in ascending order of time to generate the dynamic characteristic sequence of each battery cell.

[0100] Optionally, the basic characteristic parameters include SOC value, voltage value, temperature value, and internal resistance value; the dynamic characteristic parameters include temperature rise rate, charging current integral, current path voltage drop, voltage drop amplitude, internal resistance change rate, and resting period self-discharge rate.

[0101] Optionally, the dynamic feature sequence is the dynamic feature parameters of battery cells arranged in chronological order.

[0102] S3. A long short-term memory network model trained based on dynamic feature sequences is used as a working condition transfer model.

[0103] Preferably, training a long short-term memory network model based on dynamic feature sequences as a working condition transfer model includes:

[0104] Extract the dynamic feature parameters of each battery cell at each time point from the dynamic feature sequence, and generate the cell operating condition vector of each battery cell at each time point based on the dynamic feature parameters of each battery cell at each time point;

[0105] The cell condition vector of the same battery cell at the previous moment is used as the input sample, and the cell condition vector at the next moment is used as the output label to form a training sample pair.

[0106] Construct a long short-term memory network model, wherein: the number of neurons in the input layer is equal to the feature dimension of the unit condition vector; the number of neurons in the output layer is the same as the number of neurons in the input layer.

[0107] The training sample pairs are input into the Long Short-Term Memory network model in chronological order, with the previous time step unit load vector as the input and the next time step unit load vector as the true label. The network parameters are optimized through the backpropagation algorithm to minimize the mean square error between the predicted load vector and the true label.

[0108] Training will stop when any of the following conditions are met: the prediction error on the validation set decreases by less than a preset convergence threshold for several consecutive training cycles; or the total number of training cycles reaches a preset maximum number of iterations.

[0109] Optionally, the preset convergence threshold and the preset maximum number of iterations can be set in advance according to the actual situation.

[0110] S4. Extract the dynamic feature parameters of each battery cell at the current time to generate the cell operating condition vector of each battery cell at the current time. After normalizing the cell operating condition vector, map it to the global feature space to generate the dynamic operating condition point of each battery cell at the current time.

[0111] Optionally, the dynamic operating point represents the current actual state of the battery cell.

[0112] Optionally, the origin of the global feature space represents that all battery cells in the battery pack are in an ideal equilibrium state.

[0113] S5. Input the cell operating condition vector of each battery cell at the current time into the operating condition migration model to output the cell operating condition vector of each battery cell at the next time. Take the difference between the cell operating condition vector at the current time and the cell operating condition vector at the next time as the operating condition migration vector, and take the direction of the operating condition migration vector as the operating condition migration direction.

[0114] S6. If the dynamic operating point is located within the battery safe operating area, select the point with the smallest Euclidean distance from the intersection of the ray along the operating condition migration direction starting from the dynamic operating point and the boundary of the battery safe operating area as the equilibrium critical point; if the dynamic operating point is located outside the battery safe operating area, select the point closest to the boundary of the battery safe operating area from the dynamic operating point as the equilibrium critical point.

[0115] Optionally, the equilibrium critical point is the theoretical limit point that a battery cell is closest to the safety boundary under its current state and predicted trend. The equilibrium critical point represents the closest critical state to the safety boundary that a battery cell is allowed to reach under specific conditions, and it serves as the benchmark for calculating the equilibrium potential energy.

[0116] S7. Calculate the normalized Euclidean distance from the dynamic operating point to the equilibrium critical point, and calculate the equilibrium potential energy of each battery cell based on the normalized Euclidean distance; obtain the maximum value and standard deviation of the equilibrium potential energy of all battery cells, and then perform a weighted summation of the maximum value and standard deviation to obtain the global imbalance degree.

[0117] Optionally, the equilibrium potential energy reflects the degree to which the current state of the battery cell deviates from its "safe critical state". The larger the equilibrium potential energy, the closer the dynamic operating point is to the equilibrium critical point (i.e., very close to the safety boundary), the dangerous state needs to be pulled back. The smaller the equilibrium potential energy, the farther the dynamic operating point is from the equilibrium critical point (i.e., farther from the safety boundary, closer to the ideal origin), the safe state is close to the ideal.

[0118] Optionally, the larger the equilibrium potential energy, the more unbalanced the battery cell is, which may face safety risks due to overcharging, over-discharging, and excessive temperature rise. Therefore, it should be given priority as a charge output to participate in active balancing.

[0119] Alternatively, equilibrium potential energy = 1 - normalized Euclidean distance.

[0120] Optionally, the global imbalance degree can be obtained by weighted summation of the maximum value and the standard deviation. The weighting coefficient of the maximum value is generally 0.7-0.9, and the weighting coefficient of the standard deviation is generally 0.3-0.1.

[0121] Optionally, global imbalance is used to quantify the overall imbalance of the entire battery pack.

[0122] Optionally, the global imbalance degree effectively and comprehensively reflects the overall imbalance state and safety risk level of the battery pack by integrating the two key dimensions of "most dangerous cell state" and "overall state dispersion".

[0123] Optionally, the maximum equilibrium potential energy represents the battery cell within the energy storage battery pack that is in the most dangerous state and closest to the safety boundary. A high equilibrium potential energy value means that at least one battery cell is in danger of collapsing, the overall system risk is high, and equilibrium must be triggered. This is a key indicator for ensuring safety.

[0124] Alternatively, the standard deviation of the equilibrium potential energy represents the degree of dispersion of the states of all battery cells relative to their respective critical points. A large standard deviation means that there are large differences in the states between battery cells and poor overall coordination. Even without extremely dangerous cells, equilibration is needed to improve overall consistency.

[0125] S8. When all dynamic operating points are within the battery safe operating area and the global imbalance is greater than the preset imbalance threshold, perform the following steps: Select the battery cell with the smallest equilibrium potential energy as the equilibrium target cell and the battery cell with the largest equilibrium potential energy as the equilibrium source cell; determine the amount of charge to be transferred based on the difference in equilibrium potential energy between the equilibrium source cell and the equilibrium target cell, and transfer the amount of charge from the equilibrium source cell to the equilibrium target cell through bidirectional DC-DC or a switching matrix.

[0126] Optionally, the preset imbalance threshold is obtained based on experimental calibration.

[0127] Optionally, determining the amount of transferred charge based on the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit includes:

[0128] Calculate the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit;

[0129] Calculate the capacity correction factor based on the rated capacity and SOC value of the target unit;

[0130] Calculate the voltage difference between the equalization source unit and the equalization target unit;

[0131] The amount of basic transferred charge is calculated based on the potential energy difference, capacity correction factor, and voltage difference.

[0132] Obtain the preset safety factor and maximum transferred charge, and determine the final transferred charge based on the base transferred charge, safety factor, and maximum transferred charge;

[0133] If the voltage difference is less than the voltage difference threshold, the final transferred charge is transferred from the equalization source unit to the equalization target unit through the switching matrix; otherwise, the final transferred charge is transferred from the equalization source unit to the equalization target unit through bidirectional DC-DC.

[0134] After the charge transfer is completed, update the SOC and voltage values ​​of the equalization source cell and the equalization target cell.

[0135] Optionally, a capacity correction factor is calculated based on the rated capacity and SOC value of the target unit, using the following formula:

[0136]

[0137] Among them, C nom To balance the rated capacity of the target cells, SOC min To balance the SOC value of the target cell, SOC max To equalize the SOC value of the source cells, K c This is the capacity correction factor.

[0138] Optionally, the basic transferred charge is calculated based on the potential energy difference, capacity correction factor, and voltage difference using the following formula:

[0139]

[0140] Where ΔE is the potential energy difference, ΔV is the voltage difference, ε is the minimum constant, and K is the constant for preventing zero. c Q is the capacity correction factor. base The amount of charge transferred is based on this.

[0141] Optionally, the final transferred charge is determined based on the base transferred charge, the safety factor, and the maximum transferred charge, using the following formula:

[0142] Q transfer =min(α×Q) base Q max )

[0143] Among them, Q transfer The final transferred charge is Q, where α is the safety factor and Q is the final amount of charge transferred. base Based on the amount of charge transferred, Q max The maximum amount of charge transferred is preset.

[0144] Optionally, the maximum transferable charge is the maximum amount of charge that the battery pack system is allowed to transfer.

[0145] S9. When there is a dynamic operating point outside the battery's safe operating area, perform the following steps: reduce the input current of the energy storage battery pack according to the equalization potential energy. If there is still a dynamic operating point outside the battery's safe operating area after adjusting the input current, switch the parallel branch balance impedance.

[0146] Optionally, reducing the input current of the energy storage battery pack based on the equilibrium potential energy includes:

[0147] Battery cells whose dynamic operating point exceeds the boundary of the battery safe operating area are recorded as out-of-bounds cells.

[0148] Calculate the average equilibrium potential energy of all out-of-bounds units;

[0149] Calculate the Euclidean distance between the origin of the global feature space and each safety boundary point of the battery safety working area, and use the minimum Euclidean distance as the boundary threshold;

[0150] The current adjustment ratio is calculated based on the average equilibrium potential energy and the boundary threshold.

[0151] Calculate the target current to be adjusted based on the current input current and the current adjustment ratio;

[0152] The input current is adjusted to the target current in a stepwise manner using a preset current change rate, and the dynamic operating point of the out-of-range unit is monitored in real time.

[0153] If all dynamic operating points return to the battery's safe operating range after current adjustment, then maintain operation at the target current.

[0154] If there are still out-of-bounds units, then perform a switch to balance the impedance of the parallel branches.

[0155] Optionally, the current adjustment ratio is calculated based on the average equilibrium potential energy and the boundary threshold, using the following formula:

[0156]

[0157] Among them, K I E is the current adjustment ratio. avg To achieve the equilibrium potential mean, E bound β is the boundary threshold, and β is the preset maximum adjustment coefficient.

[0158] Optionally, the target current can be calculated based on the current input current and the current adjustment ratio, as shown in the following formula:

[0159] I adj =I in ×K I

[0160] Among them, I in K is the current input current. I For current adjustment ratio, Iadj To adjust the target current.

[0161] According to a preferred embodiment, the specific steps for performing the switching parallel branch impedance balancing operation include:

[0162] Identify all parallel branches containing out-of-bounds units and form a set of branches to be adjusted.

[0163] For each parallel branch in the set of branches to be adjusted, perform the following operations:

[0164] The balance impedance value of the parallel branch is increased by a preset impedance step size;

[0165] The dynamic operating point of each battery cell in the parallel branch is updated in real time;

[0166] Calculate the equilibrium potential energy of each battery cell in the parallel branch;

[0167] If the dynamic operating point of all battery cells in the parallel branch is within the safe operating range of the battery, then stop increasing the balancing impedance of the parallel branch and record the current impedance value.

[0168] If the balance impedance of the parallel branch increases to the preset limit impedance, and there are still battery cells in the parallel branch whose dynamic operating point exceeds the boundary of the battery safe operating area, then the corresponding parallel branch will be marked as a faulty branch.

[0169] Synchronously adjust the balance impedance value of the non-adjusted branch to keep the total input current of the energy storage battery pack at the adjustment target current;

[0170] If a faulty branch exists, the protection shutdown mechanism of the energy storage battery pack will be triggered;

[0171] If the dynamic operating points of all out-of-bounds units return to the battery's safe operating range, then the current balanced impedance values ​​of each branch will be maintained.

[0172] This application dynamically constructs a battery safe operating area and monitors the offset state of each battery cell in the feature space in real time to accurately identify potential risks. Based on migration trends, preventative control is implemented to reduce the risk of thermal runaway and minimize irreversible damage caused by battery operation outside the safe operating area. This application utilizes the operating condition migration direction combined with dynamic feature sequences and global imbalance analysis to accurately locate the source and target cells requiring optimization and determine the optimal amount of transferred charge. This avoids the blind energy dissipation of traditional equalization processes, and the equalization process is gentle and gradual, reducing the impact on the battery.

[0173] This embodiment also provides a storage medium storing a readable and writable program. When the program is executed, it implements the aforementioned active battery balancing method. Specifically, the active battery balancing method provided by this invention can be programmed or software and stored on a storage medium. In actual use, the program stored on the storage medium is used to execute the various steps of the active battery balancing method.

[0174] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention.

[0175] The present invention discloses a non-transitory computer-readable storage medium storing computer instructions, which, when executed by a processor, cause the processor to perform the above-described method.

[0176] Those skilled in the art will understand that all or part of the steps in the above methods can be implemented by a program instructing related hardware (e.g., processor, FPGA, ASIC, etc.), and the program can be stored in a readable storage medium, such as a read-only memory, a disk, or an optical disk. All or part of the steps in the above embodiments can also be implemented using one or more integrated circuits. Accordingly, each module in the above embodiments can be implemented in hardware, such as by using integrated circuits to implement its corresponding function, or it can be implemented as a software functional module, such as by a processor executing a program / instruction stored in memory to implement its corresponding function. The embodiments of the present invention are not limited to any particular combination of hardware and software.

[0177] Furthermore, the functional units in the various embodiments of this document can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0178] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this paper, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this paper. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0179] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for active balancing of industrial and commercial energy storage batteries, characterized in that, Includes the following steps: Collect historical operating condition data of energy storage battery packs and construct a battery safety operating area based on the historical operating condition data; The basic characteristic parameters of each battery cell are collected in real time at each time point within a preset time window. Based on the basic characteristic parameters, the dynamic characteristic parameters of each battery cell at each time point within the preset time window are obtained. The dynamic characteristic parameters are then arranged in ascending order of time to generate a dynamic characteristic sequence of each battery cell. A long short-term memory network model trained based on dynamic feature sequences is used as a working condition transfer model. The dynamic feature parameters of each battery cell at the current moment are extracted to generate the cell operating condition vector of each battery cell at the current moment. After normalization, the cell operating condition vector is mapped to the global feature space to generate the dynamic operating condition point of each battery cell at the current moment. The origin of the global feature space represents that all battery cells of the battery pack are in an ideal equilibrium state. Input the cell operating condition vector of each battery cell at the current time into the operating condition transfer model to output the cell operating condition vector of each battery cell at the next time. The difference between the current unit working condition vector and the next unit working condition vector is taken as the working condition transition vector, and the direction of the working condition transition vector is taken as the working condition transition direction. If the dynamic operating point is located within the battery safe operating area, select the point with the smallest Euclidean distance from the intersection of the ray along the operating point migration direction and the boundary of the battery safe operating area as the equilibrium critical point. If the dynamic operating point is located outside the battery safe operating area, the closest point from the dynamic operating point to the boundary of the battery safe operating area is taken as the equilibrium critical point. Calculate the normalized Euclidean distance from the dynamic operating point to the equilibrium critical point, and calculate the equilibrium potential energy of each battery cell based on the normalized Euclidean distance; obtain the maximum value and standard deviation of the equilibrium potential energy of all battery cells, and then perform a weighted summation of the maximum value and standard deviation to obtain the global imbalance degree. When all dynamic operating points are within the battery safe operating area and the global imbalance is greater than the preset imbalance threshold, the following steps are performed: select the battery cell with the smallest equalization potential energy as the equalization target cell and the battery cell with the largest equalization potential energy as the equalization source cell. The amount of charge to be transferred is determined based on the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit, and the amount of charge to be transferred from the equilibrium source unit to the equilibrium target unit is transferred through a bidirectional DC-DC converter or a switching matrix. When a dynamic operating point exists outside the battery's safe operating range, perform the following steps: The input current of the energy storage battery pack is reduced based on the equilibrium potential energy. If, after adjusting the input current, there is still a dynamic operating point outside the safe operating area of ​​the battery, the parallel branch impedance is switched to balance the impedance.

2. The active balancing method according to claim 1, characterized in that, The basic characteristic parameters include SOC value, voltage value, temperature value, and internal resistance value; the dynamic characteristic parameters include temperature rise rate, charging current integral, current path voltage drop, voltage drop amplitude, internal resistance change rate, and self-discharge rate during rest period.

3. The active balancing method according to claim 2, characterized in that, The battery safety working area is constructed based on historical operating condition data, including: Traverse the dynamic feature parameters and use the dynamic feature parameter being traversed as the target dynamic feature parameter. The target dynamic characteristic parameter value is gradually increased or decreased according to the preset boundary test step size until the battery cell triggers an abnormal state such as over-limit voltage, temperature alarm or internal resistance change. Record the values ​​of the target's dynamic characteristic parameters when an abnormal state is triggered, and use them as the safe boundary points of the target's dynamic characteristic parameters; Repeat the above steps until the safe boundary point corresponding to each dynamic feature parameter is obtained; Map all safety boundary points to a global feature space with dynamic feature parameters as coordinate axes; Connecting all mapped safe boundary points forms a closed multidimensional convex polyhedron boundary. Verify the boundary continuity of the multidimensional convex polyhedron boundary; if the Euclidean distance between adjacent safe boundary points exceeds the preset tolerance, then insert a linear interpolation point between the two points. The space enclosed by the boundary of the multidimensional convex polyhedron is defined as the initial battery safe operating area.

4. The active balancing method according to claim 3, characterized in that, The battery safety working area is constructed based on historical operating condition data, including: Several historical operating condition points are identified by using historical operating condition data. Each historical operating condition point is constructed from the voltage, temperature, current, SOC and internal resistance of a single battery cell at the same time. Based on preset voltage threshold, temperature threshold, internal resistance deviation threshold and BMS alarm records, historical operating points are filtered: historical operating points that trigger any abnormal condition are abnormal operating points, and historical operating points that are in the standard charging and discharging range and have no alarm records are normal operating points. The number of normal operating points that are within the initial battery safe operating area is counted, and the proportion of the number of normal operating points is calculated. If the proportion is less than a preset proportion threshold, it is determined that the initial battery safe operating area has not passed the first verification. Check if any abnormal operating condition point is located within the initial battery safe operating area. If so, the initial battery safe operating area is deemed to have failed the second verification. If the initial battery safe working area fails either the first verification or the second verification, the boundary test step size is halved, and the safe boundary point generation process is re-executed based on the halved boundary test step size, thereby reconstructing the initial battery safe working area using the newly generated safe boundary points. Repeat the above steps of generating safety boundary points, constructing the initial battery safe working area, and verifying the initial battery safe working area until the reconstructed initial battery safe working area passes both the first and second verifications. At this point, it is determined as the final battery safe working area.

5. The active balancing method according to claim 4, characterized in that, The amount of transferred charge is determined based on the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit, including: Calculate the difference in equilibrium potential energy between the equilibrium source unit and the equilibrium target unit; Calculate the capacity correction factor based on the rated capacity and SOC value of the target unit; Calculate the voltage difference between the equalization source unit and the equalization target unit; The amount of basic transferred charge is calculated based on the potential energy difference, capacity correction factor, and voltage difference. Obtain the preset safety factor and maximum transferred charge, and determine the final transferred charge based on the base transferred charge, safety factor, and maximum transferred charge; If the voltage difference is less than the voltage difference threshold, the final transferred charge is transferred from the equalization source unit to the equalization target unit through the switching matrix; otherwise, the final transferred charge is transferred from the equalization source unit to the equalization target unit through bidirectional DC-DC. After the charge transfer is completed, update the SOC and voltage values ​​of the equalization source cell and the equalization target cell.

6. The active balancing method according to claim 5, characterized in that, Reducing the input current of the energy storage battery pack based on the equilibrium potential energy includes: Battery cells whose dynamic operating point exceeds the boundary of the battery safe operating area are recorded as out-of-bounds cells. Calculate the average equilibrium potential energy of all out-of-bounds units; Calculate the Euclidean distance between the origin of the global feature space and each safety boundary point of the battery safety working area, and use the minimum Euclidean distance as the boundary threshold; The current adjustment ratio is calculated based on the average equilibrium potential energy and the boundary threshold. Calculate the target current to be adjusted based on the current input current and the current adjustment ratio; The input current is adjusted to the target current in a stepwise manner using a preset current change rate, and the dynamic operating point of the out-of-range unit is monitored in real time. If all dynamic operating points return to the battery's safe operating range after current adjustment, then maintain operation at the target current. If there are still out-of-bounds units, then perform a switch to balance the impedance of the parallel branches.

7. The active balancing method according to claim 6, characterized in that, The specific steps for performing the impedance balancing operation of switching parallel branches include: Identify all parallel branches containing out-of-bounds units and form a set of branches to be adjusted. For each parallel branch in the set of branches to be adjusted, perform the following operations: The balance impedance value of the parallel branch is increased by a preset impedance step size; The dynamic operating point of each battery cell in the parallel branch is updated in real time; Calculate the equilibrium potential energy of each battery cell in the parallel branch; If the dynamic operating point of all battery cells in the parallel branch is within the safe operating range of the battery, then stop increasing the balancing impedance of the parallel branch and record the current impedance value. If the balance impedance of the parallel branch increases to the preset limit impedance, and there are still battery cells in the parallel branch whose dynamic operating point exceeds the boundary of the battery safe operating area, then the corresponding parallel branch will be marked as a faulty branch. Synchronously adjust the balance impedance value of the non-adjusted branch to keep the total input current of the energy storage battery pack at the adjustment target current; If a faulty branch exists, the protection shutdown mechanism of the energy storage battery pack will be triggered; If the dynamic operating points of all out-of-bounds units return to the battery's safe operating range, then the current balanced impedance values ​​of each branch will be maintained.

8. The active balancing method according to claim 7, characterized in that, Long Short-Term Memory (LSTM) network models trained based on dynamic feature sequences are used as condition transfer models, including: Extract the dynamic feature parameters of each battery cell at each time point from the dynamic feature sequence, and generate the cell operating condition vector of each battery cell at each time point based on the dynamic feature parameters of each battery cell at each time point; The cell condition vector of the same battery cell at the previous moment is used as the input sample, and the cell condition vector at the next moment is used as the output label to form a training sample pair. Construct a long short-term memory network model, wherein: the number of neurons in the input layer is equal to the feature dimension of the unit condition vector; the number of neurons in the output layer is the same as the number of neurons in the input layer. The training sample pairs are input into the Long Short-Term Memory network model in chronological order, with the previous time step unit load vector as the input and the next time step unit load vector as the true label. The network parameters are optimized through the backpropagation algorithm to minimize the mean square error between the predicted load vector and the true label. Training will stop when any of the following conditions are met: the prediction error on the validation set decreases by less than a preset convergence threshold for several consecutive training cycles; or the total number of training cycles reaches a preset maximum number of iterations.

9. A storage medium having a readable and writable program stored thereon, characterized in that, When the program is executed, it implements the active balancing method as described in any one of claims 1-8.

Images