A bidirectional buck-boost battery equalization method and battery management system

Abstract

This invention proposes a bidirectional buck-boost battery balancing method and battery management system. A predictive model is established for the bidirectional buck-boost battery balancing system. Then, a composite optimization objective function based on termination and energy consumption objectives is established. The optimal control sequence is obtained by solving the composite optimization objective function using quadratic programming, thereby controlling the current of each bidirectional buck-boost balancing channel for balancing control. Compared with traditional extreme value balancing strategies and fuzzy control balancing strategies, which are based on experience and rules, the model predictive control method proposed in this invention has better control performance, improves the balancing speed, and reduces balancing losses. It provides an efficient balancing strategy for bidirectional buck-boost battery balancing topologies and has significant implications.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle battery management system technology, and in particular to a bidirectional buck-boost battery balancing method and battery management system. Background Technology

[0002] Due to limitations in manufacturing processes and variations in usage environments, individual lithium-ion batteries exhibit differences, primarily in capacity, self-discharge rate, and internal resistance. Electric vehicle battery packs consist of hundreds or even thousands of lithium-ion batteries connected in series and parallel. These individual differences lead to inconsistencies within the entire battery pack, exhibiting a bottleneck effect. Inconsistencies in capacity, self-discharge efficiency, and internal resistance result in differences in the State of Charge (SOC) of the individual lithium-ion batteries within the pack. During charging, batteries with higher SOCs reach full charge first, ending the charging process while those with lower SOCs are not yet fully charged. Conversely, during discharging, batteries with lower SOCs reach the minimum discharge threshold first, ending the discharging process while those with higher SOCs still have unreleased energy. Therefore, these inconsistencies between individual lithium-ion batteries reduce the energy efficiency of the battery pack. Battery balancing technology can effectively mitigate these inconsistencies, significantly contributing to improved battery life and energy efficiency.

[0003] Currently, numerous scholars and companies have conducted extensive research on battery balancing technology. Based on energy loss methods, battery balancing can be divided into energy-consuming balancing methods and non-energy-consuming balancing methods. Energy-consuming balancing methods achieve battery balancing by consuming the power of batteries with high SOC through circuit resistance. While simple in structure, these methods suffer from high energy loss and significant heat generation. Non-energy-consuming balancing methods utilize energy storage components such as capacitors, inductors, and transformers to achieve peak-shaving and valley-filling power transfer. Capacitor-based balancing methods mainly include switched capacitors and flying capacitors, which are simple to design, but slow when the potential difference between the two batteries to be balanced is small. Inductor-based balancing methods mainly include centralized inductor and bidirectional buck-boost balancing methods. Centralized inductor balancing methods can only balance two batteries at a time; when series-connected... When there are many cells, the equalization speed is slow. The equalization method based on bidirectional buck-boost sets up a bidirectional buck-boost equalization circuit between all adjacent pairs of cells. The current can be easily controlled by adjusting PWM (Pulse Width Modulation). However, when the two cells to be equalized are located far apart, the equalization process needs to cross many buck-boost equalization modules. The equalization method based on transformers mainly includes the equalization methods of centralized flyback transformers and distributed flyback converters. Among them, the centralized transformer can only equalize two cells at a time. When there are many cells in series, the equalization speed is slow. The distributed flyback transformer requires a flyback transformer for each cell. The equalization speed is faster, but the cost is high and the size is large, making it difficult to apply to large-scale series battery packs.

[0004] Currently, the most commonly used battery balancing method in practical applications is the bidirectional buck-boost method. The ETA3000 main control switch-type balancing chip developed by Shanghai Yutai is based on the bidirectional buck-boost battery balancing method. However, it adopts an extreme value-based balancing strategy and cannot achieve variable current control. Some scholars have used fuzzy control to achieve variable current control in the bidirectional buck-boost battery balancing circuit, which has improved the balancing loss compared to the extreme value balancing strategy. However, both of these balancing methods are based on experience and rules to determine the balancing strategy. Therefore, the balancing strategy depends entirely on the designer's experience, resulting in a simple balancing strategy, slow process speed, and large loss. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a bidirectional buck-boost battery balancing method and battery management system, which solves the problems of slow speed and high losses when using experience-based balancing methods in current bidirectional buck-boost balancing topologies.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a bidirectional buck-boost battery balancing method, comprising the following steps:

[0007] S1: Establish a mathematical model of the bidirectional buck-boost battery balancing system. Based on the balancing principle of the bidirectional buck-boost battery balancing system, establish a prediction model. Select the SOC value x(k) of each cell in the battery pack as the state variable, select the normalized balancing current u(k) of each balancing channel as the control variable, select the average SOC value r(k) of the battery pack as the reference trajectory for model prediction control, and establish the balancing power transfer matrix T according to the power transfer relationship of the bidirectional buck-boost battery balancing system.

[0008] S2: Establish a composite optimization objective function based on termination target and loss target;

[0009] S3: Obtain the constraints of the bidirectional buck-boost battery balancing system;

[0010] S4: Based on the mathematical model, the optimal control sequence in the future control time domain is obtained by solving the composite optimization objective function using the quadratic programming method;

[0011] S5: Select the first set of control variables in the optimal control sequence and apply them to the system to determine the equalization current of each equalization channel at the current moment;

[0012] S6: Repeat steps S4-S5 to achieve rolling optimization control until the equilibrium termination condition is met.

[0013] Further, in step S1, the prediction model is a spatial state equation, which is expressed as x(k+1)=A·x(k)+B·u(k), and the system observation equation is expressed as y(k)=x(k), where matrix A is the system matrix of the state equation, matrix B is the control matrix of the state equation, x(k) is the state variable matrix, u(k) is the control variable matrix, and y(k) is the system observation value.

[0014] Furthermore, the control matrix B of the system state equation is expressed as:

[0015] B = [Q] x -1 ·T·Q u ·Δt] (n,n-1)

[0016] Where n is the number of battery cells in the battery pack, Qx is the rated capacity matrix of each battery cell in the battery pack, Qu is the maximum equalization current matrix of each equalization channel in the battery pack, Δt is the sampling time, and T is the power transfer matrix, expressed as:

[0017]

[0018] Further, in step S2, the composite optimization objective function is expressed as:

[0019]

[0020] Where k is the current sampling period, i and j represent the distance from the current sampling period to the predicted future sampling period, x(k) is the battery SOC matrix at the current moment, U(k) is the control sequence in the control time domain, Np is the prediction time domain, Nc is the control time domain, EN(k) is the terminating target, which is the difference between the system observation y(k) and the reference trajectory r(k), Q is the weight matrix of the terminating target, and R is the weight matrix of the loss target.

[0021] U(k)=[u(k),u(k+1),u(k+2),…,u(k+N c -1)] T

[0022] Furthermore, in step S3, the constraint condition is:

[0023]

[0024] Where lb is the lower bound constraint of the independent variable U(k), ub is the upper bound constraint of the independent variable U(k), At is the coefficient matrix of the linear inequality constraint, and b is the right-hand vector of the linear inequality constraint.

[0025] Further, step S4 includes the following steps:

[0026] S4-1: Predict the system state for the next Np steps based on the mathematical model, and establish the relationship between the prediction sequence and the control sequence;

[0027] S4-2: Using the aforementioned relation, the objective function is derived into a standard quadratic programming problem;

[0028] S4-3: Solve the quadratic programming problem to obtain the optimal control sequence in the future control time domain.

[0029] Further, in step S4-1, the mathematical model is a spatial state equation, and the specific relation is:

[0030]

[0031] Where X(k) is the prediction sequence matrix and U(k) is the control sequence matrix.

[0032]

[0033] Furthermore, in step S4-2, the composite optimization objective function J can be expressed as a standard quadratic programming form using the aforementioned relation:

[0034]

[0035] Where H is the quadratic term vector of the quadratic programming, f is the linear term vector of the quadratic programming, f' is the transpose of vector f, U(k) is the control sequence in the control time domain, Np is the prediction time domain, and Nc is the control time domain.

[0036] Furthermore, the state variables are obtained by collecting the current value of each battery cell through sensors and correcting it using the ampere-hour integration method and the open-circuit voltage method.

[0037] The present invention also provides a battery management system that uses the bidirectional buck-boost battery balancing method as described in any of the preceding claims for battery balancing.

[0038] The beneficial effects of this invention are:

[0039] 1. The model predictive control method described in this invention can establish a mathematical model for a bidirectional buck-boost battery balancing system. By establishing a composite objective function and converting it into a quadratic programming problem for solution, the optimized control current of each balancing channel can be obtained. Compared with traditional empirical extreme value balancing methods and fuzzy control methods, it is optimized in terms of balancing speed and balancing efficiency.

[0040] 2. The research object of the model predictive control method described in this invention is a bidirectional buck-boost battery equalization system. This equalization system has been widely used in actual battery management systems due to its advantages such as simple structure, few components, and convenient control. The equalization strategy based on model predictive control proposed in this invention can optimize the performance of this equalization topology, so that its performance can be greatly improved in practical applications, and it has practical application value.

[0041] 3. The bidirectional buck-boost battery balancing method described in this invention optimizes the balancing current and the frequent switching of the MOSFET, reducing the repeated charging and discharging of the current and the frequent switching of the MOSFET, which is of great significance for extending the service life of the battery balancing management system. Attached Figure Description

[0042] Figure 1aThis is a system block diagram of the bidirectional buck-boost battery balancing method in an embodiment of the present invention;

[0043] Figure 1b This is a visual illustration of the optimal control sequence solved by the prediction model and the quadratic programming solver in an embodiment of the present invention;

[0044] Figure 2a This is a schematic diagram of the current direction when battery B1 charges inductor L1 in an embodiment of the present invention;

[0045] Figure 2b This is a schematic diagram of the current direction when inductor L1 charges battery B2 in an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of bidirectional buck-boost battery balancing topology PWM control in an embodiment of the present invention;

[0047] Figure 4a This is a graph showing the SOC equilibrium change curve based on the extreme value algorithm in an embodiment of the present invention;

[0048] Figure 4b This is a schematic diagram of the normalized equalization current variation of each equalization channel based on the extreme value algorithm in an embodiment of the present invention.

[0049] Figure 5a This is a graph showing the SOC equilibrium change curve based on the fuzzy control algorithm in an embodiment of the present invention.

[0050] Figure 5b This is a schematic diagram of the normalized equalization current variation of each equalization channel based on the fuzzy control algorithm in an embodiment of the present invention.

[0051] Figure 6a This is a graph showing the SOC equilibrium change curve based on the model predictive control algorithm in an embodiment of the present invention.

[0052] Figure 6b This is a schematic diagram of the normalized equalization current variation of each equalization channel based on the model predictive control algorithm in an embodiment of the present invention. Detailed Implementation

[0053] To make the technical solutions and advantages of the present invention clearer, the technical solutions of the embodiments of the present invention will be fully described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0054] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0055] Based on the above technical background, this invention proposes a bidirectional buck-boost battery balancing method. By establishing a mathematical model of the bidirectional buck-boost battery balancing system, a composite optimization objective function based on termination-type and energy consumption-type objectives is proposed. The optimal control sequence is obtained by solving the composite optimization objective function of quadratic programming, thereby realizing the optimized control of the bidirectional buck-boost battery balancing system. Compared with control methods based on experience and rules, this method improves the balancing speed and reduces balancing losses, which is of great significance for the optimization of the bidirectional buck-boost balancing method.

[0056] like Figure 1a The diagram shows a system block diagram of the bidirectional buck-boost battery balancing method according to an embodiment of the present invention. The A1 control object of the model predictive control is the series battery pack and its bidirectional buck-boost battery balancing system. The SOC value of each cell in the battery pack is selected as the system state variable, and the balancing current of each balancing channel of the bidirectional buck-boost battery balancing system is selected as the control variable. The spatial state equation of the bidirectional buck-boost battery balancing system is established as the A5 predictive model. Real-time parameters of voltage, current, and temperature of each cell in the battery pack are obtained through A2 data acquisition. A3 state estimation uses the ampere-hour integration method to estimate the SOC value of each cell. The SOC value at the current moment is used to predict the system state for the next Np steps using the prediction model. This predicted state is then used to derive the A7 composite optimization objective function based on termination and loss objectives, resulting in a quadratic programming problem. The optimal control sequence for the next Nc steps is calculated using the A8 quadratic programming solver, which optimizes the composite optimization objective function. The first control variable u(k) in the optimal control sequence for the next Nc steps is used as the current control quantity and applied to the bidirectional buck-boost battery balancing system in the A1 controlled object to determine the magnitude and direction of the balancing current in each balancing channel at the current moment. This iterative optimization of the model predictive control method improves the balancing speed and efficiency of the bidirectional buck-boost battery balancing method.

[0057] like Figure 1b A6 and A9 are visualizations of the optimal control sequences obtained by the A5 prediction model and the A8 quadratic programming solver, respectively.

[0058] like Figure 2a , Figure 2b This is a schematic diagram of a bidirectional buck-boost battery balancing topology. A bidirectional buck-boost battery balancing module is placed between every two adjacent batteries. Each buck-boost battery balancing module consists of one inductor, one NMOS transistor, one PMOS transistor, and two diodes. By controlling the field-effect transistors, the flow of current in different directions can be balanced. This buck-boost battery balancing module is very easily expandable. For a battery pack consisting of n lithium-ion batteries connected in series, (n-1) buck-boost battery balancing modules are needed, comprising a total of (n-1) inductors, (n-1) NMOS transistors, (n-1) PMOS transistors, and 2n diodes. The working principle of the bidirectional buck-boost battery balancing topology is as follows: Figure 2a , Figure 2b As shown, assuming battery B1 has a higher charge than battery B2, NMOS transistor G1 is first closed, forming a closed loop with battery B1, inductor L1, and NMOS transistor G1. The current direction is as follows. Figure 2a As shown by the arrow, battery B1 charges inductor L1; then NMOS transistor G1 is disconnected. At this time, the current in inductor L1 cannot change abruptly. Battery B2, inductor L1, the parasitic diode of PMOS transistor G2, and diode D2 form a closed loop, with the current direction as shown. Figure 2b As shown, inductor L1 charges battery B2; repeating the above process allows for the continuous transfer of charge from battery B1 to battery B2. The buck-boost battery balancing module can also transfer charge from battery B2 to battery B1, thus enabling bidirectional current flow.

[0059] like Figure 3 The diagram shows a bidirectional buck-boost battery balancing topology PWM control, where iL is the inductor current in the buck-boost circuit; idis is the discharge current of the battery when it discharges and charges the inductor in the buck-boost circuit; and ichg is the charging current of the battery when it is charged by the inductor in the buck-boost circuit.

[0060] Figure 3 In diagram (a), the gate PWM drive signal of the MOSFET switch has an effective time of T1. During this time, the equalization current increases linearly from zero, as shown in the diagram. Figure 3 As shown in (c), the MOSFET is disconnected after time T1 is reached. At this time, the balancing current linearly increases and decreases to zero, as shown in (c). Figure 3 As shown in (d), the current in the buck-boost inductor L1 is as follows: Figure 3 As shown in (b).

[0061] The battery pack in this embodiment is composed of n lithium-ion batteries connected in series. The state variable matrix x(k) is composed of the SOC of the n battery cells, and the SOC range is [0,1].

[0062] x(k)=[x1(k),x2(k),x3(k),…,x n (k)] T 0≤x i (k)≤1

[0063] A battery pack consisting of n batteries connected in series requires (n-1) bidirectional buck-boost battery equalization modules. The maximum equalization current of each equalization module is Imax, and the maximum equalization current matrix of the system can be represented as Qu.

[0064]

[0065] The bidirectional buck-boost battery balancing module can achieve variable current control by adjusting the PWM. Therefore, the control variable matrix u(k) is set to normalized current, with a value range of [-1, 1]. The actual balancing current value of each balancing channel is the product of the normalized current and the maximum current of the balancing channel. u(k) > 0 indicates that the balancing module transfers power from the high-position battery to the low-position battery, and u(k) < 0 indicates that the balancing module transfers power from the low-position battery to the high-position battery.

[0066] u(k)=[u1(k),u2(k),u3(k),…,u n-1 (k)] T -1≤u i (k)≤1

[0067] (1) Establish a mathematical model of the bidirectional buck-boost battery balancing system.

[0068] A bidirectional buck-boost battery balancing system consists of a bidirectional buck-boost battery balancing module between every two adjacent battery cells. The battery pack is composed of n lithium-ion batteries connected in series, requiring (n-1) bidirectional buck-boost battery balancing modules. Each balancing module can achieve power transfer in any direction between two adjacent batteries, and each balancing module can work independently and simultaneously. Based on the characteristics of the above balancing topology, the SOC of each battery cell in the battery pack is used as the state variable, and the balancing current of each bidirectional buck-boost battery balancing module is used as the control variable to establish a mathematical model of the bidirectional buck-boost battery balancing system. Then, the power transfer matrix of (n-1) bidirectional buck-boost battery balancing modules can be represented as matrix T.

[0069]

[0070] The rated capacity of each cell in the battery pack can be represented as a matrix Qx, where Ci refers to the rated capacity of the i-th cell.

[0071]

[0072] The sampling time of the discrete system is Δt. The current SOC of the lithium-ion battery is equal to the SOC value of the previous time plus the change in SOC during the sampling period. The system matrix of the state equation is A, and the control matrix of the state equation is B.

[0073]

[0074] B = [Q] x -1 ·T·Q u ·Δt] (n,n-1)

[0075] Where n is the number of battery cells in the battery pack.

[0076] The state equation of the system is expressed as:

[0077] x(k+1) = A·x(k) + B·u(k)

[0078] The system's observation equations are expressed as follows:

[0079] y(k)=x(k)

[0080] (2) Establish the composite objective function of the bidirectional buck-boost battery balancing system.

[0081] Battery equalization management objectives can be measured in two ways: equalization speed and equalization losses. Based on this, this invention proposes a composite objective function for a bidirectional buck-boost battery equalization system, including a termination objective and a loss objective. The termination objective is that the SOC values ​​of all individual cells in the battery pack should approach the average SOC of the battery pack. Since the equalization process involves factors such as the on-resistance and switching losses of the MOSFETs, and the charge transfer losses of the buck-boost circuit, a poor equalization strategy can lead to repeated charging and discharging and frequent switching of the MOSFETs, resulting in increased equalization losses. Therefore, the loss objective is to reduce repeated charging and frequent switching of the MOSFETs during the equalization process.

[0082] The reference trajectory for the model predictive control algorithm is the average SOC value of the battery pack, represented as the r(k) matrix.

[0083]

[0084] The terminating target EN(k) can be expressed as the difference between the observed value of the observation equation and the reference trajectory.

[0085] E N (k)=y(k)-r(k)=T c ·x(k)

[0086] The Tc matrix is ​​represented as follows:

[0087]

[0088] The prediction time domain of model predictive control is Np, and the control time domain is Nc, where Nc ≤ Np. The prediction sequence of the state variables is represented by matrix X(k), and the control sequence of the control variables is represented by matrix U(k).

[0089] X(k)=[x(k+1|k),x(k+2|k),x(k+3|k),…,x(k+N p |k)] T

[0090] U(k)=[u(k),u(k+1),u(k+2),…,u(k+N c -1)] T

[0091] The combined optimization objective function of the termination objective and the loss objective can be expressed as function J.

[0092]

[0093] Where k is the current sampling period, i and j represent the distance from the current sampling period to the predicted future sampling period, Q is the weight matrix of the terminating target, and R is the weight matrix of the loss target.

[0094]

[0095] Where q is the weight coefficient of the termination objective and r is the weight coefficient of the loss objective. By modifying the weight coefficients of q and r, the proportions of the Q and R matrices are changed to achieve different optimization effects.

[0096] The system state at Np steps in the time domain can be derived using the prediction model:

[0097]

[0098] make:

[0099]

[0100]

[0101] The relationship between the predicted sequence X(k) and the control sequence U(k) can be obtained:

[0102]

[0103] The composite optimization objective function J can be further derived using the relationship between the prediction sequence X(k) and the control sequence U(k):

[0104]

[0105] The Qp and Rp matrices are:

[0106]

[0107]

[0108] because The solution is independent of the control sequence U(k), therefore J(x(k),U(k),N) is obtained. p N c This part can be ignored when there are extreme values. Let and The composite optimization objective function J can then be expressed in the form of a standard quadratic programming problem:

[0109]

[0110] Where H is the quadratic term vector of the quadratic programming, f is the linear term vector of the quadratic programming, f' is the transpose of vector f, U(k) is the control sequence in the control time domain, Np is the prediction time domain, and Nc is the control time domain.

[0111] (3) Constraints for the bidirectional buck-boost battery balancing system are proposed.

[0112] The mathematical model of the bidirectional buck-boost battery balancing system uses the SOC of each battery as the state variable and the balancing current of each bidirectional buck-boost balancing module as the control variable. The constraint condition of SOC is [0,1], and the constraint condition of balancing current is the maximum current of each channel [-Imax,Imax]. The sign of the current is related to the direction of charge transfer between adjacent batteries.

[0113] Since the normalized current ranges from [-1, 1], the constraint condition for the control sequence U(k) is lb ≤ U(k) ≤ ub, where lb and ub take the following values:

[0114]

[0115] Furthermore, the battery state of charge (SOC) ranges from [0,1], so the constraints for the predicted sequence X(k) are:

[0116]

[0117] Based on the relationship between the predicted sequence X(k) and the control sequence U(k) The constraints on the predicted sequence X(k) can be transformed into the following equation:

[0118]

[0119] By phase shifting and rearranging, the second constraint condition for the control sequence U(k) can be obtained as At·U(k)≤b, where At and b take the following values:

[0120]

[0121] (4) Solving the composite objective function.

[0122] Based on the mathematical model, the SOC value of each battery cell can be estimated from the sampled battery voltage, current, and temperature data at the current moment. The SOC state variable of the future prediction time domain Np steps can be predicted by the mathematical model using the SOC matrix of each battery cell in the current battery pack. The composite objective function can be derived as a quadratic programming problem. By solving the quadratic programming problem, the optimal control sequence of the future control time domain Nc steps can be obtained. By applying the first step of the control sequence of the control time domain Nc steps as the equalization current to the system, the battery pack can be continuously optimized through rolling optimization, resulting in significant optimization in both equalization speed and equalization loss.

[0123] Through the derivation of steps (1) to (3), the composite optimization objective function J can be transformed into a standard quadratic programming problem with constraints, and its optimization objective and constraints are as follows:

[0124]

[0125]

[0126] Solving a standard quadratic programming problem with constraints can be done using the quadprog function in MATLAB. By specifying the matrix H, matrix f, constraints lb, ub, At, b, and the initial values ​​for the problem to be solved (which can be determined using random numbers), the quadprog function can obtain the optimal control sequence U(k) of the composite optimization objective function J. The first control variable u1(k) in the optimal control sequence U(k) is selected as the control quantity at the current moment. The PWM of the MOSFET in the buck-boost battery equalization module is adjusted to generate the corresponding control current to achieve optimal control.

[0127] The present invention also proposes a battery management system that uses the above-described bidirectional buck-boost battery balancing method for battery balancing.

[0128] To verify the feasibility of the embodiments of the present invention, a comparison is made below using a six-cell series-connected lithium-ion battery as an example. The model predictive control algorithm described in the embodiments of the present invention is compared with traditional empirical extremum algorithms and fuzzy control algorithms. The prediction time domain Np of the model predictive control algorithm is set to 5, and the control time domain Nc is also set to 5. The initial SOC values ​​of the six cells are selected as [80, 45, 60, 70, 65, 50], and the equilibrium termination condition is selected as the difference between the battery SOC value and the average SOC value of the battery pack being less than 0.001. Figure 4a The figure shows the battery SOC equalization curve based on the extreme value algorithm, as shown in the figure. Figure 4b The figure shows the normalized equalization current variation curves for each equalization channel based on the extreme value algorithm. As can be seen from the curves, the normalized equalization current of each channel exhibits repeated jumps over a long period. Furthermore, the current based on the extreme value algorithm can only be set to the maximum value and cannot achieve variable current control. Therefore, there will be many repeated charging and discharging cycles and repeated changes in the direction of the equalization current during the equalization process, resulting in significant losses. Figure 5a The figure shows the battery SOC equalization curve based on the fuzzy control algorithm, as shown in the figure. Figure 5b The figure shows the normalized equalization current variation curves for each equalization channel based on the fuzzy control algorithm. From these curves, it can be seen that the fuzzy control-based normalized current achieves variable current control. The magnitude of the normalized equalization current in each equalization channel gradually decreases as the SOC difference between adjacent cells decreases. Finally, when the SOC of all cells is equalized, the normalized current value gradually approaches 0. The fuzzy control-based equalization method offers significant optimization compared to the extreme value-based equalization method in terms of repeated charging and discharging and frequent changes in the equalization current direction. However, from... Figure 5a It can be seen that the equalization time based on the fuzzy control algorithm is relatively long. This is because after the SOC difference between adjacent batteries decreases in the later stages of equalization, the equalization current also decreases, leading to a slower equalization speed. Figure 6a The figure shows the battery SOC equalization curve based on the model predictive control algorithm, as shown in the figure. Figure 6bThe figure shows the normalized equalization current variation curves of each equalization channel based on the model predictive control algorithm. It can be seen from the normalized equalization current variation curves that the equalization current does not exhibit repeated charging and discharging or frequent changes in the direction of the equalization current. A larger equalization current is used for equalization in the early stage. In the later stage of equalization, compared with fuzzy control, the equalization current does not decrease as the SOC difference between adjacent batteries decreases. Instead, the optimized normalized control current is obtained by solving the composite optimization objective function, which makes the equalization speed very fast in the later stage of equalization.

[0129] Table 1 shows a comparison of SOC and time before and after equalization for the three equalization methods. Compared with the extreme value equalization method, the model predictive control method reduces the equalization time by 4.8%, the average SOC loss before and after equalization is 3.15% for the extreme value equalization method, and 2.1% for the model predictive control method, resulting in an average SOC loss improvement of 1.05%. Compared with the fuzzy control method, the model predictive control method reduces the equalization time by 71.5%, the average SOC loss before and after equalization is 1.78% for the fuzzy control method, and the average SOC loss of the model predictive control method is 0.32% higher than that of the fuzzy control method. In summary, the model predictive control method is superior to the extreme value equalization method in both equalization speed and equalization loss. Compared with fuzzy control, the model predictive control method has a slightly higher equalization loss of 0.32%, but reduces the equalization time by 71.5%. Therefore, the overall performance of the model predictive control method is superior to that of the fuzzy control method. Experiments also verify that the established composite optimization objective function based on the termination target and the loss target can comprehensively improve the equalization speed and reduce the equalization loss of the buck-boost battery equalization system.

[0130] The state variables are obtained by collecting the current value of each battery cell through sensors and correcting it using the ampere-hour integration method and the open-circuit voltage method.

[0131] Table 1 Comparison of SOC and time before and after equilibrium using three equilibrium methods

[0132]

[0133] The above describes the working principle, workflow, and application prospects of the bidirectional buck-boost battery balancing method of the present invention. In this specification, the terms "an embodiment" and "example," etc., refer to specific features, structures, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to corresponding embodiments or examples in a suitable manner.

[0134] The above provides a detailed description of the bidirectional buck-boost battery balancing method provided by the present invention, and further elaborates the invention with specific embodiments. It should be noted that the above description of the embodiments is not intended to limit the invention but only to help understand the core idea of ​​the invention. For those skilled in the art, any improvements to the invention and equivalent alternatives made to the invention without departing from the principle of the invention are also within the scope of protection of the claims of the present invention.

Claims

1. A bidirectional buck-boost battery balancing method, characterized in that, Includes the following steps: S1: Establish a mathematical model for the bidirectional buck-boost battery balancing system, that is, establish a prediction model based on the balancing principle of the bidirectional buck-boost battery balancing system, select the SOC value x(k) of each cell in the battery pack as the state variable, select the normalized balancing current u(k) of each balancing channel as the control variable, select the average SOC value r(k) of the battery pack as the reference trajectory for model prediction control, and establish a balancing power transfer matrix according to the power transfer relationship of the bidirectional buck-boost battery balancing system. S2: Establish a composite optimization objective function based on termination target and loss target; S3: Obtain the constraints of the bidirectional buck-boost battery balancing system; S4: Based on the prediction model, the optimal control sequence in the future control time domain is obtained by solving the composite optimization objective function using a quadratic programming method; S5: Select the first set of control variables in the optimal control sequence and apply them to the system to determine the equalization current of each equalization channel at the current moment; S6: Repeat steps S4-S5 until the equilibrium termination condition is met. In step S1, the prediction model is a spatial state equation, and its state equation is expressed as follows: The system's observation equation is expressed as , where matrix A is the system matrix of the state equation, matrix B is the control matrix of the state equation, x(k) is the state variable matrix, representing the battery SOC at the current moment, u(k) is the control variable matrix, and y(k) is the system observation value; The control matrix B of the system state equation is expressed as: Where n is the number of battery cells in the battery pack, Qx is the rated capacity matrix of each battery cell in the battery pack, and Qu is the maximum equalization current matrix of each equalization channel in the battery pack. Let T be the sampling time, and T be the energy transfer matrix, expressed as: ； In step S2, the composite optimization objective function is expressed as: Where k is the current sampling period, i and j represent the distance from the current sampling period to the predicted future sampling period, x(k) is the battery SOC matrix at the current moment, U(k) is the control sequence in the control time domain, Np is the prediction time domain, Nc is the control time domain, J is the composite optimization objective function, T is the equalization charge transfer matrix, B is the control matrix of the system state equation, EN(k) is the termination target, which is the difference between the system observation value y(k) and the reference trajectory r(k), Q is the weight matrix of the termination target, and R is the weight matrix of the loss target. Step S4 includes the following steps: S4-1: Based on the prediction model, predict the system state of the prediction time domain Np and establish the relationship between the prediction sequence and the control sequence; S4-2: Using the aforementioned relation, the objective function is derived into a standard quadratic programming problem; S4-3: Solve the quadratic programming problem to obtain the optimal control sequence in the future control time domain.

2. The method according to claim 1, characterized in that, In step S3, the constraint condition is: in lb ub is the lower bound constraint of the independent variable U(k), ub is the upper bound constraint of the independent variable U(k), At is the coefficient matrix of the linear inequality constraint, and b is the right-hand vector of the linear inequality constraint.

3. The method according to claim 1, characterized in that, In step S4-1, the prediction model is a spatial state equation, and the specific relation is as follows: ， Where X(k) is the prediction sequence matrix and U(k) is the control sequence matrix. 。 4. The method according to claim 3, characterized in that, In step S4-2, the composite optimization objective function J is expressed as a standard quadratic programming form using the aforementioned relation: Where H is the quadratic term vector of the quadratic programming, f is the linear term vector of the quadratic programming, f' is the transpose of vector f, U(k) is the control sequence in the control time domain, Np is the prediction time domain, and Nc is the control time domain.

5. The method according to claim 1, characterized in that, The state variables are obtained by collecting the current value of each battery cell through sensors and correcting it using the ampere-hour integration method and the open-circuit voltage method.

6. A battery management system, characterized in that, Battery balancing is performed using the bidirectional buck-boost battery balancing method according to any one of claims 1-5.

Images