A group balancing control method for series battery packs based on K-means clustering

Abstract

The present invention discloses a group balancing control method for series battery packs based on K-means clustering. Assume that the working threshold of the balancing circuit is Δ ref The maximum single cell capacity of the series battery pack is SOC max , the minimum single cell power is SOC min , if SOC max With SOC min The difference is greater than Δ ref , the balancing circuit is started. Through K-means clustering, all cells in the series battery pack are divided into three groups; adjacent high-power cells and adjacent low-power cells are identified and clustered into groups respectively; the group with the largest number of cells to be discharged and high-power cells to be discharged is discharged and balanced, while the group with the largest number of cells to be charged and low-power cells to be charged is charged and balanced. When SOC max With SOC min The difference is less than or equal to Δ ref Compared with the cell-to-cell balancing control method, the present invention significantly improves the balancing speed while ensuring the balancing efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of battery pack balancing; specifically, to a group balancing control method for series battery packs based on K-means clustering, which is suitable for energy balancing of battery energy storage systems. Background Art

[0002] Lithium-ion battery packs are widely used in distribution network energy storage systems. Since the manufacturing process, materials and other aspects cannot be completely consistent, there are inevitable consistency differences in the capacity and impedance of single cells (referred to as single cells for short). These differences will increase with the running time of the battery pack, thereby reducing the available capacity and cycle life of the battery pack, and even causing overcharge and over-discharge, posing a safety hazard. In order to improve the energy utilization rate of the battery pack and extend its cycle life, effective balancing must be introduced to reduce the consistency differences of the battery pack. Compared with parallel battery packs, series battery packs are more susceptible to the consistency differences of single cells. The content of the present invention is developed with respect to the balancing of series battery packs.

[0003] Balancing control refers to controlling the transfer of balancing energy based on consistency metrics. Common consistency metrics include open circuit voltage, operating voltage, and state of charge (SOC). The open circuit voltage parameter is readily available, and when the battery is idle, there is a stable mapping between open circuit voltage and SOC. However, using open circuit voltage as a balancing metric requires the battery to be idle for a period of time before each measurement, resulting in longer balancing times. Compared to open circuit voltage, measuring the operating voltage is more convenient and quicker, and can be performed in real time during battery operation. However, it is dependent on changes in the battery pack's load current. Large current fluctuations can cause significant fluctuations in the battery's operating voltage, leading to malfunctions in the battery pack's balancing system, frequent starts and stops, and even over-balancing. Furthermore, different types of lithium batteries exhibit voltage plateaus. Between 10% and 90% remaining charge, the voltage changes very little. Consequently, when using voltage as a balancing metric, slight changes in voltage can lead to significant changes in remaining charge, hindering balancing accuracy. Using SOC as a balancing indicator can ensure that each cell in the battery pack is fully charged or discharged at the same time, thereby improving the energy utilization rate of the battery pack. It is a relatively ideal battery pack consistency difference indicator.

[0004] Research findings on balancing energy transfer methods include the following: Model predictive control and fuzzy logic control are proposed, which effectively improve balancing speed and avoid over-balancing compared to the common cell-to-cell control methods based on extreme or average values. State-of-charge balancing is achieved through the establishment of a consensus-based stochastic control algorithm, ensuring ideal balancing even when the SOC estimate is subject to noise. Fuzzy logic is used to dynamically adjust the balancing current to improve balancing speed and efficiency. A particle swarm optimization algorithm is used to optimize the balancing path of each cell to improve balancing efficiency and speed and avoid over-balancing. A hierarchical balancing control method is developed, dividing the entire battery pack into several modules and ultimately achieving overall pack balancing through intra-module and inter-module balancing. While these studies optimized the balancing energy transfer process using different algorithms, they all designed balancing control methods based on the principle of "one-to-one" balancing. Within each sampling cycle, balancing energy is still transferred between cells or between fixed battery modules. Some scholars have also proposed a "many-to-many" balancing control method based on the idea of ​​clustering. The main research results include: achieving cluster balancing based on the k-means clustering algorithm optimized by genetic algorithms; and achieving cluster balancing based on the adaptive balancing topology, introducing the concept of sample density on the basis of the traditional fuzzy C-means clustering algorithm, and designing a density-based fuzzy C-means clustering algorithm. Compared with the "one-to-one" balancing based on extreme values ​​or average values, the above two clustering balancing methods effectively improve the balancing speed and efficiency, but ignore the significant feature of the energy storage battery pack, which is the large number of cells. This greatly increases the probability of adjacent cells or adjacent modules with small consistency differences. As a result, adjacent cells or modules with different numbers of small consistency differences can participate in balancing as battery groups at the same time. In addition, a complex structural and control-complex balancing topology is introduced to achieve cluster balancing, resulting in low balancing efficiency and slow balancing speed. Summary of the Invention

[0005] The present invention proposes a group balancing control method for series battery packs based on K-means clustering. While introducing the idea of ​​cluster balancing, it also establishes a group balancing control method based on adjacent cells with small consistency differences, thereby realizing the transfer of balancing energy between battery groups containing different numbers of adjacent cells, thereby improving the balancing speed of the battery energy storage system.

[0006] The present invention relates to the research of a balancing control method. With SOC as the balancing indicator, adjacent high-power cells and low-power cells are clustered into groups based on the K-means clustering algorithm to achieve group balancing, and ultimately improve the balancing speed based on the established hardware circuit.

[0007] Based on this, the present invention takes SOC as the balancing indicator and proposes a group balancing control method for series battery packs based on K-means clustering.

[0008] The following steps are involved:

[0009] Assume that the working threshold of the equalization circuit is Δ ref The maximum single cell capacity of the series battery pack is SOC max , the minimum single cell power is SOC min , if SOC max With SOC min The difference is greater than Δ ref , the balancing circuit starts;

[0010] First, based on SOC, all cells in the series battery pack are divided into three groups using the K-means clustering algorithm: cells with higher SOC, cells with moderate SOC, and cells with lower SOC.

[0011] Secondly, the high-power cells with adjacent serial numbers and the low-power cells with adjacent serial numbers are identified and clustered into a high-power battery group to be discharged and a low-power battery group to be charged respectively.

[0012] Finally, the battery group with the largest number of cells to be discharged and high power is discharged and balanced, while the battery group with the largest number of cells to be charged and low power is charged and balanced; when SOC max With SOC min The difference is less than or equal to Δ ref When the equalization circuit stops working.

[0013] Compared with the cell-to-cell balancing control method, the balancing control method of the present invention can significantly improve the balancing speed while ensuring the balancing efficiency.

[0014] Preferably, the method specifically comprises the following steps:

[0015] (1) In each sampling period, the initial SOC of each cell in the group is obtained to determine whether the balancing circuit startup conditions are met. If the conditions are met, the balancing circuit is started and step (2) is performed; otherwise, the balancing circuit does not work;

[0016] (2) Using the cell SOC as the characteristic parameter, the K-means clustering algorithm is used to classify the SOCs of all cells in the series battery pack into three categories: the first category is cells with higher SOCs, the second category is cells with moderate SOCs, and the third category is cells with lower SOCs;

[0017] (3) Cells with higher SOC are identified in ascending order of serial numbers, which are called high-capacity cells. The high-capacity cells with adjacent serial numbers are grouped into a high-capacity battery group to be discharged. Cells with lower SOC are identified in ascending order of serial numbers, which are called low-capacity cells. The low-capacity cells with adjacent serial numbers are grouped into a low-capacity battery group to be charged.

[0018] (4) discharging and balancing the battery group with the largest number of adjacent high-power cells to be discharged, and charging and balancing the battery group with the largest number of adjacent low-power cells to be charged;

[0019] (5) This process repeats until the SOC of each cell no longer meets the balancing circuit startup condition and the balancing circuit stops.

[0020] Preferably, the clustering process of monomers with higher SOC and monomers with lower SOC is as follows:

[0021] Assume that the number of cells in the series battery pack is i, and the serial numbers of each cell are B1, B2, ..., B i ; The high-power monomer clustering process is as follows:

[0022] First, determine whether cell B1 is a high-charge cell. If not, the determination stops. If so, determine whether the next-numbered cell B2 is a high-charge cell. If B2 is not a high-charge cell, cell B1 is grouped as a high-charge cell to be discharged, and the determination stops. If B2 is a high-charge cell, determine whether the next-numbered cell B3 is a high-charge cell. If B3 is not a high-charge cell, cells B1 and B2 are grouped as a high-charge cell to be discharged, and the determination stops. If B3 is a high-charge cell, determine whether the next-numbered cell B4 is a high-charge cell. If B4 is not a high-charge cell, cells B1, B2, and B3 are grouped as a high-charge cell to be discharged, and the determination stops. If B4 is a high-charge cell, cells B1, B2, B3, and B4 are grouped as a high-charge cell to be discharged, and the determination stops. Then, starting from cell B2, the process continues until cell B4. i , repeat the above steps;

[0023] The high-capacity battery group to be discharged with the largest number of high-capacity cells is discharged and balanced. If multiple high-capacity battery groups to be discharged have the largest number of cells at the same time, the high-capacity battery group to be discharged with the smallest cell number is discharged and balanced.

[0024] Preferably, the number of cells in the series battery pack is i, and the serial numbers of the cells are B1, B2, ..., B i ; The low-power monomer clustering process is as follows:

[0025] First, determine whether cell B1 is a low-battery cell. If not, the determination stops. If so, determine whether the next-numbered cell B2 is a low-battery cell. If B2 is not a low-battery cell, cell B1 is grouped as a low-battery cell to be charged, and the determination stops. If B2 is a low-battery cell, determine whether the next-numbered cell B3 is a low-battery cell. If B3 is not a low-battery cell, cells B1 and B2 are grouped as a low-battery cell to be charged, and the determination stops. If B3 is a low-battery cell, determine whether the next-numbered cell B4 is a low-battery cell. If B4 is not a low-battery cell, cells B1, B2, and B3 are grouped as a low-battery cell to be charged, and the determination stops. If B4 is a low-battery cell, cells B1, B2, B3, and B4 are grouped as a low-battery cell to be charged, and the determination stops. Then, starting from cell B2, the process continues until cell B3. i , repeat the above steps;

[0026] Perform charge balancing on the battery group with the largest number of low-battery cells to be charged. If multiple battery groups with the largest number of cells are simultaneously charged, perform charge balancing on the battery group with the largest cell number.

[0027] Furthermore, the method of the present invention is applicable to a balancing topology in which balancing energy can be transferred between cells, cells and battery modules, and battery modules and battery modules.

[0028] The beneficial effect of the present invention is that: the present invention uses SOC as the balancing indicator and introduces the clustering concept, changing the traditional charge / discharge balancing method of a single battery to a balancing control method between battery groups composed of multiple adjacent high / low power cells with similar SOCs, thereby achieving a faster balancing speed while ensuring balancing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the principles of the present invention and the technical solutions in its implementation, the technical solutions involved in the present invention will be further introduced using the accompanying drawings. The following drawings are only some implementation examples of the present invention. For those skilled in the art, other technical solutions can be obtained based on the following drawings without any creative work.

[0030] Figure 1 This is a control strategy flow chart of Example 2 of the present invention;

[0031] Figure 2 This is a structural diagram of a 12-cell series-connected battery pack balancing circuit according to Example 2 of the present invention;

[0032] Figure 3 This is a flow chart of clustering groups of high-capacity batteries to be discharged according to embodiment 2 of the present invention;

[0033] Figure 4 This is a flow chart of clustering low-battery groups to be charged according to embodiment 2 of the present invention;

[0034] Figure 5 The SOC distribution diagram of the cells before and after balancing for the cell-to-cell balancing control method and the group balancing control method of Example 2 when the power of the 12 cells in a 12-cell series battery pack is high in the middle and low on both sides;

[0035] Figure 6 A diagram showing the SOC range variation of all cells during the balancing process of the cell-to-cell balancing control method and the group balancing control method of Example 2, when the cells in a 12-cell series battery pack have a high charge in the middle and low charges on both sides;

[0036] Figure 7 The SOC distribution diagram of the cells before and after balancing of the cell-to-cell balancing control method and the group balancing control method of Example 2 when the power of the 12 cells in a 12-cell series battery pack is high on both sides and low in the middle;

[0037] Figure 8 A graph showing the SOC range variation of all cells during the balancing process of the cell-to-cell balancing control method and the group balancing control method of Example 2, when the 12 cells in a 12-cell series battery pack have high power on both sides and low power in the middle;

[0038] Figure 9 The cell SOC distribution diagram before and after balancing of the cell-to-cell balancing control method and the group balancing control method of Example 2 when the 12 cells of a 12-cell series battery pack have uniform power distribution;

[0039] Figure 10 The figure shows the SOC range variation of all cells during the balancing process of the cell-to-cell balancing control method and the group balancing control method of Example 2, when the power distribution of the 12 cells in a 12-cell series battery pack is uniform. DETAILED DESCRIPTION

[0040] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. These illustrative embodiments of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0041] Example 1

[0042] A K-means clustering-based balancing control method for a series battery pack includes the following steps:

[0043] The SOC of each cell is used as the balancing variable, and the cell SOC range difference greater than the balancing threshold is used as the balancing circuit working condition. The K-means clustering algorithm is used to cluster and group each cell, and the duty cycle of the switch control signal is set according to the number of cells in the high-capacity battery group to be discharged.

[0044] Furthermore, the following are included:

[0045] (1) Obtain the initial SOC and set the equalization circuit startup conditions

[0046] In each sampling period, the initial SOC of each cell in the group is obtained to determine whether the balancing circuit startup conditions are met. If the conditions are met, the balancing circuit is started and step (2) is performed; otherwise, the balancing circuit does not work.

[0047] First, the initial SOC of each cell in the series battery pack is obtained.

[0048] Secondly, determine the SOC of the single cell with the highest power in the series battery pack as SOC max , the lowest power cell SOC is SOC min , set the threshold of the equalization circuit to Δ ref .

[0049] When satisfied:

[0050] SOC max -SOC min >Δ ref (1)

[0051] The balancing circuit meets the working conditions and starts.

[0052] (2) Using K-means clustering algorithm

[0053] Taking the cell SOC as the characteristic parameter, all cells in the series battery pack are divided into three categories through the K-means clustering algorithm: the first category is cells with higher SOC, the second category is cells with moderate SOC, and the third category is cells with lower SOC.

[0054] ① Take the SOC of each monomer as the data set X, and randomly select 3 elements from the data set X as the center points of the 3 clusters, i.e., the centroid;

[0055] ② According to the similarity criterion, the SOC of each monomer is assigned to the closest centroid to form different clusters;

[0056] ③ Use the Euclidean calculation formula to recalculate the centers of the three clusters and update the centroid of each cluster;

[0057] ④ Repeat ② and ③ until the centroids no longer change, and output the results;

[0058] Using the K-means clustering algorithm, three centroids K1, K2, and K3 are generated, satisfying K1>K2>K3. All cells in the series battery pack are divided into the following three categories: the first category (centroid K1) is cells with a higher SOC, known as high-charge cells, which require discharge balancing; the second category (centroid K2) is cells with a moderate SOC, which do not require balancing; the third category (centroid K3) is cells with a lower SOC, known as low-charge cells, which require charge balancing. Based on the clustering results, the corresponding cell numbers are determined;

[0059] (3) Identify the high-capacity battery group to be discharged and the low-capacity battery group to be charged

[0060] From monomer B1 to monomer B N (N is the total number of battery cells) and make judgments in sequence. For cell B m (m=1,2,……,N), first, if it does not satisfy:

[0061]

[0062] Then the judgment stops; if it is satisfied, then monomer B m It is a high-power monomer, and the judgment continues;

[0063] Then, if not satisfied:

[0064]

[0065] Then monomer B m It is classified as a high-capacity battery group to be discharged, and the judgment stops; if it meets the requirements, then the single cell B m+1 It is a high-power monomer, and the judgment continues;

[0066] Then, if not satisfied:

[0067]

[0068] Then monomer B m 、B m+1 Classified as a high-capacity battery group to be discharged, the judgment stops; if satisfied, then single B m+2 It is a high-power monomer, and the judgment continues;

[0069] Finally, if not satisfied:

[0070]

[0071] Then monomer B m 、B m+1 、B m+2 Classified as a high-capacity battery group to be discharged, the judgment stops; if satisfied, single cell B m+3 For high power monomer, monomer B m 、Bm+1 、B m+2 、B m+3 It is classified as a high-capacity battery group to be discharged, and the judgment is stopped;

[0072] Similarly, for monomer B n (n=1,2,……,N), first, if it does not satisfy:

[0073]

[0074] Then the judgment stops; if it is satisfied, then monomer B n It is a low-battery cell, and the judgment continues;

[0075] Then, if not satisfied:

[0076]

[0077] Then monomer B n It is classified as a low-power battery group to be charged, and the judgment stops; if it meets the requirements, then the single cell B n+1 It is a low-battery cell, and the judgment continues;

[0078] Then, if not satisfied:

[0079]

[0080] Then monomer B n 、B n+1 Classified as a low-power battery group to be charged, the judgment stops; if satisfied, then the cell B n+2 It is a low-battery cell, and the judgment continues;

[0081] Finally, if not satisfied:

[0082]

[0083] Then monomer B n 、B n+1 、B n+2 Classified as a low-power battery group to be charged, the judgment stops; if satisfied, single B m+3 For low power monomer, monomer B n 、B n+1 、B n+2 、B n+3 It is classified as a low-battery group to be charged, and the judgment is stopped.

[0084] (4) Setting the control signal duty cycle

[0085] To prevent the energy storage element from saturating, it must operate in discontinuous current mode, that is, the current must drop to zero within one switching cycle. Therefore, the appropriate control signal duty cycle can be determined based on the number of cells involved in discharge balancing. The control signal duty cycle D1 for balancing a group of high-capacity batteries to be discharged should meet the following requirements:

[0086]

[0087] The duty cycle D2 of the control signal for balancing the low-battery group to be charged is:

[0088] D2=1-D1 (11)

[0089] (5) Group balance

[0090] Discharge balancing is performed on the group with the largest number of high-capacity cells in each group of high-capacity batteries to be discharged. If the group with the largest number of high-capacity cells is not unique, discharge balancing is performed on the group containing the cell with the smallest sequence number.

[0091] Charge balancing is performed on the group with the largest number of low-battery cells in each group of low-battery batteries to be charged. If the group with the largest number of low-battery cells is not unique, charge balancing is performed on the group containing the cell with the largest sequence number.

[0092] (6) End of equilibrium

[0093] At the end of each sampling period, if the SOC of each monomer meets the working conditions of the balancing circuit, the balancing circuit continues to work; if it does not meet the working conditions of the balancing circuit, the balancing circuit stops and the balancing ends.

[0094] Example 2

[0095] like Figure 1 FIG. 1 is a flow chart of the control strategy of Example 2 of the present invention.

[0096] like Figure 2 As shown in the figure, it is a structural diagram of a 12-cell series battery pack balancing circuit according to embodiment 2 of the present invention. The balancing circuit includes 12 cells, 26 MOS tubes, 22 diodes, and 1 inductor L. Each cell is marked as B1, B2, ..., B 12 MOS tubes connected to the monomer are marked as S0, S1, ..., S 25 . Monomers B2, B3, ..., B 12 The left and right bridge arms of the positive electrode are connected to the MOS tube and diode in series; monomers B1, B2, ..., B 11 The left and right bridge arms of the negative electrode are connected to the MOS tube and the diode in series; the left bridge arm of the positive electrode of the monomer B1 is connected to the MOS tube S1, and the right bridge arm of the positive electrode of the monomer B1 is connected to the MOS tube S0; the monomer B 12 The left bridge arm of the negative pole and the MOS tube S 25Connection, monomer B 12 The right bridge arm of the negative pole and the MOS tube S 24 Connection: The source of MOS tube S0 is connected to the drain of MOS tube with an even number. 25 The drain of is connected to the source of the odd-numbered MOS tube; the two ends of the inductor L are connected to the MOS tube S 25 、MOS tube S 24 are connected.

[0097] The aforementioned 12-cell series battery pack balancing circuit is used as an example to demonstrate the advantages of the present invention's K-means clustering-based group balancing control method for series battery packs over conventional single-cell balancing control methods. The cells used are 18650-type ternary lithium-ion batteries with a nominal voltage of 3.7V and a capacity of 1800mAh. To closely simulate actual operating conditions, three sets of initial battery SOC values ​​are set in this embodiment: Experiment 1, Experiment 2, and Experiment 3. Experiment 1 describes a 12-cell series battery pack with a high charge distribution in the middle and low charges on both sides; Experiment 2 describes a 12-cell series battery pack with a high charge distribution on both sides and low charges in the middle; and Experiment 3 describes a 12-cell series battery pack with a uniform charge distribution across all 12 cells.

[0098] Specifically, the following steps are included:

[0099] (1) Obtain the initial SOC and set the equalization circuit startup conditions

[0100] Determine the SOC of the single cell with the highest power in the series battery pack as SOC max , the lowest power cell SOC is SOC min , set the threshold value Δ of the equalization circuit ref =1%.

[0101] When satisfied:

[0102] SOC max -SOC min >1% (1)

[0103] The balancing circuit meets the working conditions and starts.

[0104] (2) Using K-means clustering algorithm

[0105] ① Take the SOC of each monomer as the data set X, and randomly select 3 elements from the data set X as the center points of the 3 clusters, i.e., the centroid;

[0106] ② According to the similarity criterion, the SOC of each monomer is assigned to the nearest centroid to form different clusters;

[0107] ③ Use the Euclidean calculation formula to recalculate the centers of the three clusters and update the centroid of each cluster;

[0108] ④ Repeat ② and ③ until the centroids no longer change, and output the results;

[0109] Using the K-means clustering algorithm, three centroids K1, K2, and K3 are generated, satisfying K1>K2>K3. All cells in the series battery pack are divided into the following three categories: the first category (centroid K1) is cells with a higher SOC, known as high-charge cells, which require discharge balancing; the second category (centroid K2) is cells with a moderate SOC, which do not require balancing; the third category (centroid K3) is cells with a lower SOC, known as low-charge cells, which require charge balancing. Based on the clustering results, the corresponding cell numbers are determined;

[0110] (3) Identify the high-capacity battery group to be discharged and the low-capacity battery group to be charged

[0111] From monomer B1 to monomer B 12 Judge in sequence, for monomer B m (m=1,2,……,12), first, if it does not satisfy:

[0112]

[0113] Then the judgment stops; if satisfied, then monomer B m It is a high-power cell, so the judgment continues;

[0114] Then, if not satisfied:

[0115]

[0116] Then monomer B m It is classified as a high-capacity battery group to be discharged, and the judgment stops; if it meets the requirements, then the single cell B m+1 It is a high-power cell, so the judgment continues;

[0117] Then, if not satisfied:

[0118]

[0119] Then monomer B m 、B m+1 Classified as a high-capacity battery group to be discharged, the judgment stops; if satisfied, then single B m+2 It is a high-power cell, so the judgment continues;

[0120] Finally, if not satisfied:

[0121]

[0122] Then monomer B m、B m+1 、B m+2 Classified as a high-capacity battery group to be discharged, the judgment stops; if satisfied, single cell B m+3 For high power monomer, monomer B m 、B m+1 、B m+2 、B m+3 It is classified as a high-capacity battery group to be discharged, and the judgment is stopped;

[0123] Similarly, for monomer B n (n=1,2,……,i), first, if it does not satisfy:

[0124]

[0125] Then the judgment stops; if it is satisfied, then monomer B n It is a low-battery cell, and the judgment continues;

[0126] Then, if not satisfied:

[0127]

[0128] Then monomer B n It is classified as a low-power battery group to be charged, and the judgment stops; if it meets the requirements, then the single cell B n+1 It is a low-battery cell, and the judgment continues;

[0129] Then, if not satisfied:

[0130]

[0131] Then monomer B n 、B n+1 Classified as a low-power battery group to be charged, the judgment stops; if satisfied, then the cell B n+2 It is a low-battery cell, and the judgment continues;

[0132] Finally, if not satisfied:

[0133]

[0134] Then monomer B n 、B n+1 、B n+2 Classified as a low-power battery group to be charged, the judgment stops; if satisfied, single B m+3 For low power monomer, monomer B n 、B n+1 、B n+2 、B n+3 It is classified as a low-battery group to be charged, and the judgment stops;

[0135] The identification process of high-capacity battery groups to be discharged and low-capacity battery groups to be charged is as follows: Figure 3 、 Figure 4 shown.

[0136] (4) Setting the control signal duty cycle

[0137] To prevent the energy storage element from saturating, it must operate in discontinuous current mode, that is, the current must drop to zero within one switching cycle. Therefore, the appropriate control signal duty cycle can be determined based on the number of cells involved in discharge balancing. The control signal duty cycle D1 for balancing a group of high-capacity batteries to be discharged should meet the following requirements:

[0138]

[0139] The duty cycle D2 of the control signal for balancing the low-battery group to be charged is:

[0140] D2=1-D1 (11)

[0141] From step (3), it can be seen that the minimum number of cells to be discharged is 1 and the maximum is 4; the minimum number of cells to be charged is 1 and the maximum is 4; to meet the conditions, it is set as follows: when the number of cells to be discharged is 1, D1 = 50%, D2 = 50%; when the number of cells to be discharged is 2, D1 = 33%, D2 = 67%; when the number of cells to be discharged is 3, D1 = 25%, D2 = 75%; when the number of cells to be discharged is 4, D1 = 20%, D2 = 80%.

[0142] (5) Group balance

[0143] Discharge balancing is performed on the group with the largest number of high-capacity cells in each group of high-capacity batteries to be discharged. If the group with the largest number of high-capacity cells is not unique, discharge balancing is performed on the group containing the cell with the smallest sequence number.

[0144] Charge balancing is performed on the group with the largest number of low-battery cells in each group of low-battery batteries to be charged. If the group with the largest number of low-battery cells is not unique, charge balancing is performed on the group containing the cell with the largest sequence number.

[0145] (6) End of equilibrium

[0146] At the end of each sampling period, if the SOC of each cell meets the balancing circuit operating conditions, the balancing circuit continues to operate. If not, the balancing circuit stops operating, and balancing ends. Taking the balancing of a series battery pack in a standby state as an example, a simulation model of Example 2 was built in Matlab / Simulink, with model parameters shown in Table 1.

[0147] Table 1 Equilibrium model simulation parameters

[0148]

[0149]

[0150] The balancing results of the cell-to-cell balancing control in Experiments 1, 2, and 3 and the group balancing control method of Example 2 (i.e., the group balancing control method for series battery packs based on K-means clustering, hereinafter referred to as the group balancing control method) are shown in Tables 2, 3, and 4.

[0151] Table 2 Equalization simulation results of Experiment 1

[0152]

[0153] Table 3 Equalization simulation results of Experiment 2

[0154]

[0155] Table 4 Equalization simulation results of Experiment 3

[0156]

[0157]

[0158] Figure 5 、 Figure 6 They are respectively the SOC distribution and SOC range changes of all cells before and after balancing in Experiment 1 (the case where the power of the 12 cells in a 12-cell series battery pack is high in the middle and low on both sides), the cell-to-cell balancing control method and the group balancing control method of Example 2.

[0159] Depend on Figure 5 As can be seen from Table 2, the group balancing control method of Example 2 is more effective in improving the inconsistency of the series battery pack. After the balancing is completed, the SOC dispersion of the cells corresponding to the group balancing control is smaller than the control method based on cell-to-cell balancing. Figure 6 It can be seen that after a 131-second balancing process, the SOC range is reduced from 2% to 1% based on the group balancing control of Example 2. Compared with the 317-second balancing time of the cell-to-cell balancing control method, the balancing speed is increased by 58.7%.

[0160] Figure 7 、 Figure 8 They are respectively the SOC distribution and SOC range changes of all cells before and after balancing in the cell-to-cell balancing control method and the group balancing control method of Example 2 in Experiment 2 (the case where the 12 cells in a 12-cell series battery pack have high power on both sides and low in the middle).

[0161] Depend on Figure 7As can be seen from Table 3, the group balancing control method of Example 2 is more effective in improving the inconsistency of the series battery pack. After the balancing is completed, the SOC dispersion of the cells corresponding to the group balancing control is smaller than the control method based on cell-to-cell balancing. Figure 8 It can be seen that after a 180-second balancing process, the SOC range is reduced from 2% to 1% based on the group balancing control of Example 2. Compared with the balancing time of 410 seconds of the cell-to-cell balancing control method, the balancing speed is increased by 56.1%.

[0162] Figure 9 、 Figure 10 They are respectively the SOC distribution and SOC range changes of all cells before and after balancing using the cell-to-cell balancing control method and the group balancing control method of the present invention in Experiment 3 (the case where the power distribution of the 12 cells in a 12-cell series battery pack is uniform).

[0163] Depend on Figure 9 As can be seen from Table 4, the group balancing control method of Example 2 is more effective in improving the inconsistency of the series battery pack. After the balancing is completed, the SOC dispersion of the cells corresponding to the group balancing control is smaller than the control method based on cell-to-cell balancing. Figure 10 It can be seen that after a 147-second balancing process, the SOC range is reduced from 2% to 1% based on the group balancing control of Example 2. Compared with the balancing time of 227 seconds of the cell-to-cell balancing control method, the balancing speed is increased by 35.2%.

[0164] Based on the above analysis, based on the established balancing topology, for different distributions of initial SOC differences, the group balancing control method of the present invention significantly improves the balancing speed compared with the cell-to-cell balancing control method while maintaining similar balancing efficiency.

[0165] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiment. It should be noted that simple modifications, improvements, substitutions, simplifications, and combinations of the present invention made within the scope of this technical field without departing from the technical concept of the present invention are all included in the scope of protection of the present invention.

Claims

1. A K-means clustering-based balancing control method for series battery groups; characterized in that: The following steps are involved: Assume that the working threshold of the equalization circuit is Δ ref The maximum single cell capacity of the series battery pack is SOC max , the minimum single cell power is SOC min , if SOC max With SOC min The difference is greater than Δ ref , the balancing circuit starts; First, based on SOC, all cells in the series battery pack are divided into three groups using the K-means clustering algorithm: cells with higher SOC, cells with moderate SOC, and cells with lower SOC. Secondly, the high-power cells with adjacent serial numbers and the low-power cells with adjacent serial numbers are identified and clustered into a high-power battery group to be discharged and a low-power battery group to be charged respectively. Finally, the battery group with the largest number of cells to be discharged and high power is discharged and balanced, while the battery group with the largest number of cells to be charged and low power is charged and balanced; when SOC max With SOC min The difference is less than or equal to Δ ref When , the equalization circuit stops working; The specific steps include: (1) In each sampling period, the initial SOC of each cell in the group is obtained to determine whether the balancing circuit startup conditions are met. If the conditions are met, the balancing circuit is started and step (2) is performed; otherwise, the balancing circuit does not work; (2) Using the cell SOC as the characteristic parameter, the K-means clustering algorithm is used to classify the SOCs of all cells in the series battery pack into three categories: the first category is cells with higher SOCs, the second category is cells with moderate SOCs, and the third category is cells with lower SOCs; (3) Cells with higher SOC are identified in ascending order of serial numbers, which are called high-capacity cells. The high-capacity cells with adjacent serial numbers are grouped into a high-capacity battery group to be discharged. Cells with lower SOC are identified in ascending order of serial numbers, which are called low-capacity cells. The low-capacity cells with adjacent serial numbers are grouped into a low-capacity battery group to be charged. (4) discharging and balancing the battery group with the largest number of adjacent high-power cells to be discharged, and charging and balancing the battery group with the largest number of adjacent low-power cells to be charged; (5) This process repeats until the SOC of each cell no longer meets the balancing circuit startup condition and the balancing circuit stops.

2. The K-means clustering-based balancing control method for series battery groups according to claim 1, characterized in that: Assume that the number of cells in the series battery pack is i, and the serial numbers of each cell are B1, B2, ..., B i ; The high-power monomer clustering process is as follows: First, determine whether cell B1 is a high-charge cell. If not, the determination stops. If so, determine whether the next-numbered cell B2 is a high-charge cell. If B2 is not a high-charge cell, cell B1 is grouped as a high-charge cell to be discharged, and the determination stops. If B2 is a high-charge cell, determine whether the next-numbered cell B3 is a high-charge cell. If B3 is not a high-charge cell, cells B1 and B2 are grouped as a high-charge cell to be discharged, and the determination stops. If B3 is a high-charge cell, determine whether the next-numbered cell B4 is a high-charge cell. If B4 is not a high-charge cell, cells B1, B2, and B3 are grouped as a high-charge cell to be discharged, and the determination stops. If B4 is a high-charge cell, cells B1, B2, B3, and B4 are grouped as a high-charge cell to be discharged, and the determination stops. Then, starting from cell B2, the process continues until cell B4. i , repeat the above steps; The high-capacity battery group to be discharged with the largest number of high-capacity cells is discharged and balanced. If multiple high-capacity battery groups to be discharged have the largest number of cells at the same time, the high-capacity battery group to be discharged with the smallest cell number is discharged and balanced.

3. The K-means clustering-based balancing control method for series battery packs according to claim 1, characterized in that: Assume that the number of cells in the series battery pack is i, and the serial numbers of each cell are B1, B2, ..., B i ; The low-power monomer clustering process is as follows: First, determine whether cell B1 is a low-battery cell. If not, the determination stops. If so, determine whether the next-numbered cell B2 is a low-battery cell. If B2 is not a low-battery cell, cell B1 is grouped as a low-battery cell to be charged, and the determination stops. If B2 is a low-battery cell, determine whether the next-numbered cell B3 is a low-battery cell. If B3 is not a low-battery cell, cells B1 and B2 are grouped as a low-battery cell to be charged, and the determination stops. If B3 is a low-battery cell, determine whether the next-numbered cell B4 is a low-battery cell. If B4 is not a low-battery cell, cells B1, B2, and B3 are grouped as a low-battery cell to be charged, and the determination stops. If B4 is a low-battery cell, cells B1, B2, B3, and B4 are grouped as a low-battery cell to be charged, and the determination stops. Then, starting from cell B2, the process continues until cell B3. i , repeat the above steps; Perform charge balancing on the battery group with the largest number of low-battery cells to be charged. If multiple battery groups with the largest number of cells are simultaneously charged, perform charge balancing on the battery group with the largest cell number.

4. The K-means clustering-based balancing control method for a series battery pack group according to any one of claims 1 to 3, characterized in that: The method is applicable to a balancing topology in which balancing energy can be transferred between cells, cells and battery modules, and battery modules and battery modules.

Images