A battery balancing control method and device based on MPC

Abstract

The present invention discloses a battery balancing control method and device based on MPC, the method steps include: S1. constructing a predictive control model of battery cells; S2. acquiring the SOC value and voltage value of the single cell in real time, simulating and predicting the predictive control model based on the MPC method, solving the optimal balancing threshold and balancing center point of the predictive control model, and predicting the real-time transfer power of each single cell; S3. performing balancing control on the controlled battery group according to the optimal balancing threshold, the balancing center point and the predicted real-time transfer power of each single cell solved in real time. The present invention can realize balancing between battery cells, reduce the inconsistency between battery cells, and reduce the number of invalid charging and discharging between single cells, effectively improving the balancing efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of battery pack balancing, and in particular to a battery balancing control method and device based on MPC. Background Art

[0002] New energy power generation technology is playing an increasingly important role in the current power industry. As the main power battery, lithium iron phosphate battery has the characteristics of high energy density, long service life, high temperature resistance and no memory. It has been widely used in energy storage systems, electric vehicles, DC system backup power supplies and other fields. In practical applications, lithium battery packs are composed of a certain number of single cells in series / parallel. Due to different battery production processes, operating conditions and other conditions, there will inevitably be differences between lithium battery cells. If these differences are not controlled, the differences between single cells will gradually expand over time, and will eventually cause overall failure of the battery pack, and even cause combustion and explosion accidents. Therefore, in order to reduce the impact of battery pack "inconsistency" on power supply reliability, improve battery pack efficiency, and extend battery life, it is very necessary to control the battery pack with a balancing strategy.

[0003] Among the current equalization methods, the equalization methods are mainly divided into the following categories according to the different equalization circuit topologies:

[0004] 1. Transformer type equalization

[0005] Transformer-type balancing mainly relies on energy conversion between electric and magnetic fields. It has large current and high conversion efficiency. However, its balancing circuit topology is relatively complex, the number of control components is too large, the economy is low, and it is not suitable for application in actual engineering.

[0006] 2. Sensitive balance

[0007] Inductive and capacitive balancing mainly relies on the circuit topology containing capacitors and inductors, using capacitors and inductors to store and release energy to achieve balancing between single cells with different capacities. However, this type of balancing method has a complex topology, low energy utilization, and a high price, and is also not suitable for actual engineering applications.

[0008] 3. DC / DC conversion balance

[0009] DC / DC conversion balancing is to add a high-frequency transformer to the balancing circuit to separate the primary and secondary, and transfer energy between single cells through magnetic coupling. Compared with other types of balancing technologies, the energy transfer efficiency is higher and the energy conversion speed is faster. The balancing process can be carried out simultaneously with the charging and discharging process. When this method is used, the increase in the number of batteries in series will not increase the voltage across the circuit elements. It is suitable for scenarios with a large number of single cells, such as substation DC systems.

[0010] Although the DC / DC conversion balancing method has the advantages of high energy transfer efficiency and fast energy conversion speed, in the prior art, a fixed threshold is usually set for balancing the DC / DC conversion balancing, that is, the balancing threshold is fixed and unchanged, and each single cell is controlled according to the fixed threshold. However, the state of the battery is constantly changing, and the fixed threshold method cannot adapt to the battery balancing under different states, resulting in a large number of unnecessary invalid charging and discharging between single cells, reducing the working efficiency of the balancing circuit, and thus the actual balancing control effect is not good. Summary of the invention

[0011] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides a battery balancing control method and device based on MPC, which has a simple implementation method, high control efficiency, and is safe and reliable. It can achieve balancing between battery cells, reduce inconsistency between battery cells, and reduce the number of invalid charging and discharging times between single cells, thereby effectively improving the balancing efficiency.

[0012] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0013] A battery balancing control method based on MPC, comprising the following steps:

[0014] S1. Construct a predictive control model for battery cells;

[0015] S2. Real-time acquisition of the SOC value and voltage value of the single cell, simulation and prediction of the predictive control model based on the MPC method, solving the optimal balancing threshold and balancing center point of the predictive control model, and predicting the real-time transfer power of each single cell;

[0016] S3. Perform balancing control on the controlled battery group according to the optimal balancing threshold, the balancing center point and the predicted real-time transfer power of each single battery solved in real time.

[0017] Furthermore, the steps of step S1 include:

[0018] S101. Constructing a coupling relationship model between the SOC value of a single cell and the voltage value of the single cell, and performing local linearization to construct a dynamic linear model between the SOC value of the single cell and the voltage value of the single cell;

[0019] S102. Obtain a predictive control model of the battery cell according to the discrete equation of the dynamic linear model.

[0020] Furthermore, in step S102, in step S101, the dynamic linear model constructed is:

[0021] y k=Sx k

[0022] in i is the number value, Q i is the i-th SOC value, u i is the i-th voltage value, N is the total number of SOC values; S represents the matrix that controls the linear dynamic behavior, specifically:

[0023]

[0024] Among them, A and B are the input and output matrices of the voltage value discrete system, respectively, and C and D are the input and output matrices of the SOC value, respectively;

[0025] The discrete equation of the dynamic model in step S102 is:

[0026]

[0027] Among them, Q k is the battery capacity of the kth battery at the last moment, Q k+1 k is the battery number, m is, A m , B m is the discrete control matrix, T s is the discrete transposed matrix;

[0028] The predictive control model of the battery cell is obtained as follows:

[0029]

[0030] Wherein, M represents the number of single cells in the battery pack.

[0031] Furthermore, in step S3, the optimal equilibrium threshold of the predictive control model is solved according to the following formula in step S2:

[0032]

[0033] Among them, Q c is the equilibrium center point, Δ(Q C ) is the function constructed to solve the equilibrium center point, δ is the interval length, and f is about Q i function, η is the energy conversion rate, m is the number of high-power single-cell batteries with a power level greater than a preset threshold, and r is the number of low-power single-cell batteries with a power level less than a preset threshold;

[0034] Δ(Q C ) is used as the optimal balancing threshold.

[0035] Furthermore, in step S102, the step S2 uses a one-dimensional exploration method to solve the equilibrium point center, and the steps include:

[0036] S201. Input the initial SOC value Q of the single battery, solve f and δ in the simulation prediction model, and calculate the average SOC value Q of the single battery av and as the starting point Q C0 ;

[0037] S202. Take the next node Q according to the solution step C1 , i.e. Q C1 =Q av +h, where h is the solution step size, Q is calculated based on the determined m and r M and Q L , Q M and Q L They are the reference values ​​of the equalization center points of the corresponding high and low power single cells respectively;

[0038] S203. Comparison Q M and Q L , if ΔQ M <ΔQ L , then return to step S202 to continue solving; if ΔQ M >ΔQ L , then return to the starting point Q C0 And solve in the other direction;

[0039] S204. After traversing all the initial SOC values ​​Q of the battery cells, filter out the minimum value ΔQ min And serves as the equilibrium center point.

[0040] Furthermore, in step S102 and step S202, ΔQ1 and ΔQ0 are calculated as follows:

[0041]

[0042]

[0043] ΔQ=|Q M -Q L |

[0044] Among them, Q M Q is the balancing center point of high-capacity battery cells. L Q is the balancing center point of the low-power battery cell. i is the SOC value of the i-th battery, Q C is the equilibrium center point, Q j is the jth battery SOC value, δ is the interval length, and f is about Q i function.

[0045] Furthermore, the specific control steps of the MPC method in step S2 include:

[0046] S211. After initialization, collect the input control vector to be measured and set the target control value x set ;

[0047] S212. Input the input control vector collected at the current moment into the predictive control model to predict the state variable x in the prediction sequence of the current moment t for the next moment t+1 , which is the model prediction value;

[0048] S213. Calculate the difference between the model prediction value at the current moment and the previous moment to obtain a prediction error ΔQ;

[0049] S214. Taking the minimum prediction error ΔQ as the objective function, perform N-step prediction control on the input control vector, and obtain the real-time transfer power sequence of each single battery from each output result.

[0050] Furthermore, the step S3 includes: comparing the balancing center point obtained by real-time solution with the SOC value of the single cell, performing balancing control on the battery pack whose difference is within a preset interval, and controlling the amount of power transferred between single cells according to the optimal balancing threshold obtained by solution and the predicted real-time transfer power of each single cell.

[0051] Furthermore, a power transfer matrix is ​​constructed, in which the elements are x units of power x transferred from the ith battery cell to the jth battery cell during the balancing process. ij , wherein the SOC value of the i-th battery cell is higher than that of the j-th battery cell, the sparsity S(T) of the power transfer matrix is ​​used to represent the amount of power transferred between the battery cells, and the minimum value of the sparsity S(T) of the transfer matrix is ​​solved using the real-time transfer power of each battery cell, so as to control the balancing control according to the solved minimum value.

[0052] A battery balancing control device based on MPC includes a processor and a memory, wherein the memory is used to store a computer program, the processor is used to execute the computer program, and the processor is used to execute the computer program to perform the above method.

[0053] Compared with the prior art, the advantages of the present invention are:

[0054] 1. The present invention constructs a predictive control model for battery cells, and simulates and predicts the predictive control model based on the MPC method in real time according to the SOC value and voltage value of the single battery, that is, the balancing threshold is simulated and predicted in real time, and the dynamic simulation threshold can be solved to achieve real-time dynamic balancing control between battery cells, which can reduce the number of invalid charging and discharging times between single batteries, improve the working efficiency of the balancing circuit, and effectively reduce the "inconsistency" between battery cells, thereby improving the power supply safety and reliability of the battery pack.

[0055] 2. The present invention further adopts a one-dimensional exploration method to solve the equilibrium center point, and adds a criterion for whether to solve forward in the solution process. Compared with the traditional equilibrium strategy that requires each SOC value to be judged in turn, the equilibrium efficiency can be further improved and the equilibrium time can be effectively reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Schematic diagram of the equalization circuit structure used in this embodiment.

[0057] Figure 2 It is a schematic diagram of the implementation flow of the battery balancing control method based on MPC in this embodiment.

[0058] Figure 3 It is a flow chart of solving the equilibrium center point by using the one-dimensional exploration method in this embodiment.

[0059] Figure 4 It is a schematic diagram of a specific implementation flow of the control using the MPC method in this embodiment. DETAILED DESCRIPTION

[0060] The present invention is further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0061] This embodiment implements battery balancing control based on DC / DC conversion balancing. The balancing circuit used is as follows: Figure 1 As shown, LT1~LTn are high-frequency transformers. By adding a high-frequency transformer to the balancing circuit to separate the primary and secondary, energy is transferred between the single cells in the form of transformer magnetic linkage coupling, which can effectively reduce energy loss. Based on this circuit structure, the number of series-connected batteries can be increased without increasing the voltage on the switch, and it can be applied to occasions with a large number of series-connected battery packs. By controlling the above-mentioned balancing circuit, the balancing control between the single cells in the battery pack can be achieved.

[0062] like Figure 2 As shown, the steps of the MPC-based battery balancing control method of this embodiment include:

[0063] S1. Construct a predictive control model for battery cells;

[0064] S2. Obtain the SOC value and voltage value of the single battery in real time, simulate and predict the predictive control model based on the MPC (Model Predictive Control) method, and solve the optimal balancing threshold and balancing center point Q of the predictive control model. c , and predict the real-time transfer power of each single battery. The above-mentioned balancing threshold is the battery power threshold determined by a fixed range or ratio with the balancing center point as the center. The balancing center point is the starting point in the balancing solution process;

[0065] S3. Perform balancing control on the controlled battery group according to the optimal balancing threshold, balancing center point and the predicted real-time transfer power of each single battery solved in real time.

[0066] This embodiment first constructs a predictive control model for the battery cell, and simulates and predicts the predictive control model based on the MPC method in real time according to the SOC value and voltage value of the single battery, that is, performs real-time simulation prediction of the balancing threshold, and can solve the dynamic simulation threshold to achieve real-time dynamic balancing control between battery cells. It can reduce the number of invalid charging and discharging times between single cells, improve the working efficiency of the balancing circuit, and effectively reduce the "inconsistency" between battery cells, thereby improving the safety and reliability of battery pack power supply.

[0067] In order to optimize the direction and amount of power transfer, a corresponding mathematical model must be established to solve the equilibrium center point Q c , and the balance threshold, the balance threshold is specifically set to: Q c ±[(1+f)δ], where f is the value of Q i The function can be adjusted in combination with the battery operating characteristics. The δ value is the interval length and can be set according to the balancing accuracy.

[0068] In this embodiment, the specific steps of constructing the predictive control model of the battery cell in step S1 include:

[0069] S101. Constructing a coupling relationship model between the SOC (battery state of charge) value of a single cell and the voltage value of the single cell, and performing local linearization to construct a dynamic linear model between the SOC value of the single cell and the voltage value of the single cell;

[0070] S102. Obtain a predictive control model for the battery cell based on the discrete equation of the dynamic linear model.

[0071] Since the battery pack charging process is a dynamic nonlinear process, the relationship between the battery SOC value and the single cell voltage is difficult to express clearly. In this embodiment, an optimization algorithm is used to locally linearize the coupling relationship between the battery SOC value and the single cell voltage:

[0072] [X=x1x2…x N ],[Y=y1y2…y N ] (1)

[0073]

[0074] in i is the number value, Q i is the i-th SOC value, u i is the i-th voltage value, N is the total number of SOC values;

[0075] Then the dynamic linear model can be expressed as:

[0076] y k =Sx k (3)

[0077] Where S represents the matrix that controls the linear dynamic behavior:

[0078]

[0079] Among them, A and B are the input and output matrices of the voltage value discrete system, C and D are the input and output matrices of the SOC value, respectively;

[0080] The matrix S can be solved by the least squares method, that is:

[0081]

[0082] Where: Represents the pseudoinverse of the matrix X.

[0083] Then according to equations (4) and (5), the forward Euler method is applied, according to the formula The two discrete state quantities (x i ,y i ) are connected through the state vector derivative to obtain the discrete equation of the dynamic linear model:

[0084]

[0085] Among them, Q k is the battery capacity of the last moment, k is the battery number, m is the number of high-capacity batteries, A m , B m is the discrete control matrix, T s is the discrete transposed matrix;

[0086] After finishing, we get:

[0087] Q k+1 =A i Q k +B iu k (7)

[0088] Where: A i =A m T s +Ω is the discrete system matrix, B i =B m T s is a discrete input matrix.

[0089] Then the predictive control model of the battery cell can be obtained as:

[0090]

[0091] In the formula, is the predicted value of the battery capacity of the ith single cell, A i , B i is the input and output matrix of the ith discrete system, is the electromagnetic voltage value of the ith single cell, and M represents the number of single cells in the battery pack.

[0092] That is, in step S101 of this embodiment, a dynamic linear model is first constructed according to formula (3), and a matrix S is constructed according to formula (4). S is solved by the least squares method, and then the forward Euler method is applied according to formulas (4) and (5) to obtain the discrete equation of the dynamic model such as formula (6), and finally the predictive control model of the battery cell such as formula (8) is derived.

[0093] In step S2 of this embodiment, the optimal equilibrium threshold value of the predictive control model is solved according to the following formula:

[0094]

[0095] Among them, Q c is the equilibrium center point, Δ(Q C ) is the function constructed to solve the equilibrium center point, δ is the interval length, and f is about Q i function, η is the energy conversion rate, m is the number of high-power single cells, and r is the number of low-power single cells;

[0096] Δ(Q C ) is taken as the optimal equilibrium threshold.

[0097] In the above formula (9), δ is generally taken as 2%Q C (You can also take other values ​​according to actual needs), then the corresponding threshold is [1±2% (1+f)]Q C , η is the energy conversion rate. Analyzing formula (9), we know that the function Δ(Q C )| is a concave function. It is not difficult to deduce that the function Δ(Q C)|There is only one extreme point, which is the minimum point. Then the real-time optimal threshold under the predictive control model can be obtained by the threshold corresponding to the minimum point.

[0098] After solving the optimal threshold, we need to further solve the center of the equilibrium point. Considering that the SOC value of the single battery is usually distributed in the SOC mean value Q av In this embodiment, the mean value Q is set to av As the starting point for solving the equilibrium point center, the equilibrium point center is quickly solved through the one-dimensional exploration method. Figure 2 As shown, the steps of using the one-dimensional exploration method to solve the equilibrium point center in step S2 specifically include:

[0099] S201. Input the initial SOC value Q of the single battery, solve f and δ in the simulation prediction model, and calculate the average SOC value Q of the single battery av and as the starting point Q C0 ;

[0100] S202. Take the next node Q according to the solution step C1 , i.e. Q C1 =Q av +h, where h is the solution step size, Q is calculated based on the determined m and r M and Q L , Q M and Q L They are the reference values ​​of the equalization center points of the corresponding high and low power single cells respectively;

[0101] S203. Comparison Q M and Q L , if ΔQ M <ΔQ L , then return to step S202 to continue solving; if ΔQ M >ΔQ L , then return to the starting point Q C0 And solve in the other direction;

[0102] S204. After traversing all the initial SOC values ​​Q of the battery cells, filter out the minimum value ΔQ min and serves as the equilibrium center point.

[0103] This embodiment uses a one-dimensional exploration method to explore the equilibrium center point, and adds a criterion for whether to forward solve in the solution process. The exploration and solution process starts from one side of the equilibrium center point and explores all SOC values ​​that meet the conditions in this direction. If the conditions are met, the exploration continues. If the conditions are not met, the exploration moves to the other side. Since the SOC value changes in real time, the exploration efficiency can be improved through the above method. Compared with the traditional equilibrium strategy that requires judging each SOC value in turn, the equilibrium efficiency can be further improved and the equilibrium time can be effectively reduced.

[0104] In the above step S202, ΔQ1 and ΔQ0 are calculated specifically according to the following formula:

[0105]

[0106]

[0107] ΔQ=|Q M -Q L | (12)

[0108] Among them, Q M Q is the center point of high-capacity battery cell balance, L Q is the center point of the low-power battery cell balance. i is the SOC value of the i-th battery, Q C is the equilibrium center point, Q j is the jth battery SOC value, δ is the interval length, and f is about Q i function.

[0109] like Figure 3 As shown, the specific control steps of the MPC method in step S2 of this embodiment include:

[0110] S211. After initialization, collect the input control vector to be measured and set the target control value x set ;

[0111] S212. Input the input control vector collected at the current moment into the predictive control model to predict the state variable x in the prediction sequence of the current moment t for the next moment t+1 , which is the model prediction value;

[0112] S213. Calculate the difference between the model prediction value at the current moment and the previous moment to obtain a prediction error ΔQ;

[0113] S214. Taking the minimum prediction error ΔQ as the objective function, perform N-step prediction control on the input control vector, and obtain the real-time transfer power sequence of each single battery from each output result.

[0114] In this embodiment, step S3 includes: comparing the balancing center point obtained by real-time solution with the SOC value of the single cell, performing balancing control on the battery pack whose difference is within a preset interval, and controlling the amount of power transferred between the single cells according to the optimal balancing threshold obtained by solution and the predicted real-time transfer power of each single cell.

[0115] In this embodiment, a power transfer matrix is ​​constructed, in which the elements are x units of power x transferred from the ith battery cell to the jth battery cell during the balancing process. ij, where the SOC value of the i-th battery cell is higher than that of the j-th battery cell, the sparsity S(T) of the power transfer matrix is ​​used to represent the amount of power transferred between the single cells, and the minimum value of the sparsity S(T) of the transfer matrix is ​​solved using the real-time transfer power of each single cell, so as to control the balancing control according to the solved minimum value. By obtaining the power transfer equation according to the balancing threshold, that is, obtaining the optimal power transfer scheme between single cells, the number of invalid charge and discharge times between single cells can be reduced, and the working efficiency of the balancing circuit can be improved.

[0116] Assume x ij In the balancing process, the high SOC value battery cell i transfers x units of electricity to the low SOC value battery cell j. Based on this, the electricity transfer matrix T can be obtained. m×r :

[0117]

[0118] The sparsity S(T) of the transfer matrix above represents the number of power transfers between battery cells. In order to maximize the balancing efficiency of the entire battery pack, the number of transfers should be minimized. Therefore, the problem of solving the balancing strategy can be transformed into solving the problem of minimizing the sparsity S(T) of the matrix.

[0119] For optimization problems with constraints, the following formula (14) can be used to solve them:

[0120]

[0121] The minimum value of the matrix sparsity S(T) can be solved according to formula (14) to obtain the optimal number of power transfers between battery cells.

[0122] Let the transposed matrix x T =[x 11 ,x 12 ,…x 1r ,x 21 …x mr ] 1×mr is a solution vector of formula (14), then x ij The required transfer time is t ij , then:

[0123]

[0124] Wherein, T is the pulse signal period generated by the controller, and ΔQ is the amount of electricity transferred between the two single cells during the period.

[0125] Transfer time t ij It can reflect the balanced control performance. According to the above calculation, the transfer time t ijAfter that, it can be determined whether the current balance control performance meets the preset requirements. If not, the balance control is re-executed until the balance control performance meets the preset requirements.

[0126] The present invention dynamically solves the optimal balancing threshold and balancing center point in real time based on the real-time status (voltage, SOC value) of the single battery. Compared with the traditional balancing strategy of setting a single threshold method, it can effectively reduce the number of invalid threshold solutions and improve the efficiency of the balancing process.

[0127] The MPC-based battery balancing control device of this embodiment includes a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to execute the above method.

[0128] The present invention can be applied to lithium iron phosphate batteries to perform balancing control between single cells in a battery pack, and can certainly be applied to other types of batteries to perform balancing control between single cells in a battery pack according to actual needs.

[0129] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. A battery balancing control method based on MPC, characterized in that the steps include: S1. Construct a predictive control model for battery cells; S2. Real-time acquisition of the SOC value and voltage value of the single cell, simulation and prediction of the predictive control model based on the MPC method, solving the optimal balancing threshold and balancing center point of the predictive control model, and predicting the real-time transfer power of each single cell; S3. Performing balanced control on the controlled battery pack according to the optimal balanced threshold, the balanced center point and the predicted real-time transfer power of each single battery solved in real time; The predictive control model of the battery cell is: Where M represents the number of cells in the battery pack, Q k is the last moment's battery capacity of the kth battery, k is the battery number, A i is the discrete system input matrix, B i is the discrete system output matrix, u i k is the electromagnetic voltage value of the i-th monomer; In step S2, the optimal equilibrium threshold of the predictive control model is solved according to the following formula: Among them, Q i is the i-th SOC value, Q j is the jth battery SOC value, Q c is the equilibrium center point, Δ(Q C ) is the function constructed to solve the equilibrium center point, δ is the interval length, and f is about Q i function, η is the energy conversion rate, m is the number of high-power single-cell batteries with a power level greater than a preset threshold, and r is the number of low-power single-cell batteries with a power level less than a preset threshold; Δ(Q C ) as the optimal balance threshold; In step S2, a one-dimensional exploration method is used to solve the equilibrium center point, and the steps include: S201. Input the initial SOC value Q of the single battery, solve f and δ in the simulation prediction model, and calculate the average SOC value Q of the single battery av and as the starting point Q C0 ; S202. Take the next node Q according to the solution step C1 , i.e. Q C1 =Q av +h, where h is the solution step size, Q is calculated based on the determined m and r M and Q L , Q M and Q L They are the reference values ​​of the equalization center points of the corresponding high and low power single cells respectively; S203. Comparison Q M and Q L , if ΔQ M <ΔQ L , then return to step S202 to continue solving; if ΔQ M >ΔQ L , then return to the starting point Q C0 And solve in the other direction; S204. After traversing all the initial SOC values ​​Q of the battery cells, filter out the minimum value ΔQ min and serves as the equilibrium center point; In step S202, ΔQ1 and ΔQ0 are calculated as follows: ΔQ=|Q M -Q L | Among them, Q M Q is the balancing center point of high-capacity battery cells. L Q is the balancing center point of the low-power battery cell. i is the SOC value of the i-th battery, Q C is the equilibrium center point, Q j is the jth battery SOC value, δ is the interval length, and f is about Q i Function of The step S3 includes: comparing the balancing center point obtained by real-time solution with the SOC value of the single cell, performing balancing control on the battery pack whose difference is within a preset interval, and controlling the amount of power transferred between single cells according to the optimal balancing threshold obtained by solution and the predicted real-time transfer power of each single cell.

2. The MPC-based battery balancing control method according to claim 1, characterized in that: The steps of step S1 include: S101. Constructing a coupling relationship model between the SOC value of a single cell and the voltage value of the single cell, and performing local linearization to construct a dynamic linear model between the SOC value of the single cell and the voltage value of the single cell; S102. Obtain a predictive control model of the battery cell according to the discrete equation of the dynamic linear model.

3. The MPC-based battery balancing control method according to claim 2, characterized in that: In step S102, in step S101, the dynamic linear model constructed is: and k =Sx k in i is the number value, Q i is the i-th SOC value, u i is the i-th voltage value, N is the total number of SOC values; S represents the matrix that controls the linear dynamic behavior, specifically: Among them, A and B are the input and output matrices of the voltage value discrete system, respectively, and C and D are the input and output matrices of the SOC value, respectively; The discrete equation of the dynamic linear model in step S102 is: Among them, A m , B m is the discrete control matrix, T s is the discrete transposed matrix; The predictive control model of the battery cell is obtained:

4. The MPC-based battery balancing control method according to any one of claims 1 to 3, characterized in that: The specific control steps of the MPC method in step S2 include: S211. After initialization, collect the input control vector to be measured and set the target control value x set ; S212. Input the input control vector collected at the current moment into the predictive control model to predict the state variable x in the prediction sequence of the current moment t for the next moment t+1 , which is the model prediction value; S213. Calculate the difference between the model prediction value at the current moment and the previous moment to obtain a prediction error ΔQ; S214. Taking the minimum prediction error ΔQ as the objective function, perform N-step prediction control on the input control vector, and obtain the real-time transfer power sequence of each single battery from each output result.

5. The MPC-based battery balancing control method according to claim 1, characterized in that: Construct a power transfer matrix, where the elements in the power transfer matrix are x units of power x transferred from the ith battery cell to the jth battery cell during the balancing process ij , wherein the SOC value of the i-th battery cell is higher than that of the j-th battery cell, the sparsity S(T) of the power transfer matrix is ​​used to represent the amount of power transferred between the battery cells, and the minimum value of the sparsity S(T) of the transfer matrix is ​​solved using the real-time transfer power of each battery cell, so as to control the balancing control according to the solved minimum value.

6. A battery balancing control device based on MPC, comprising a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program, characterized in that: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 5.

Images