Large-scale energy storage battery pack equalization method based on EMD (Empirical Mode Decomposition) optimal planning

Abstract

The invention discloses a large-scale energy storage battery pack equalization method based on EMD optimal planning, which comprises the steps that a terminal communicates with a server through a network, a data storage system is used for storing data needing to be processed by the server, the data storage system is integrated on the server, and the terminal is personal networking equipment. According to the large-scale energy storage battery pack equalization method based on EMD optimal planning, a novel structural design is adopted, an EMD intelligent optimization algorithm is introduced, and based on comprehensive consideration of electric quantity difference, energy loss and equalization time among a plurality of batteries, electric quantity distribution among the batteries can be more accurately optimized in each equalization process, so that the equalization efficiency is improved. According to the method, the improvement of the equalization efficiency is ensured, the phenomenon of over-equalization is avoided, that is, the target voltage / SOC can be more efficiently reached in each equalization, the condition of incomplete equalization is reduced, and an EMD optimal planning algorithm is adopted.

Description

Technical Field

[0001] The present invention relates to the technical field of active balancing of large-scale energy storage battery groups, and in particular to a large-scale energy storage battery group balancing method based on EMD optimal planning. Background Art

[0002] Smart active balancing of energy storage battery packs is an intelligent battery pack management technology used to improve the overall performance and life of the battery pack. In the battery management system (BMS), active balancing technology reduces the imbalance caused by capacity differences between battery cells by monitoring and adjusting the power of each battery cell.

[0003] In the prior art, a Chinese patent with publication number CN118473042A discloses an active equalization method, device, electronic device and storage medium for energy storage batteries. The above patent is the closest prior art at present. As shown in the above patent: a data-driven method is used to determine whether the energy storage battery system needs to be equalized. This type of method establishes an active equalization judgment matrix based on the current state of the i-th battery cell; based on the judgment matrix, a memory correction factor of the i-th battery cell is determined; wherein the memory correction factor includes terminal voltage, state of charge, terminal voltage change rate, state of charge change rate, terminal voltage change rate average value and state of charge change rate average value; the memory correction factor is used to comprehensively evaluate the state of the i-th battery cell and actively equalize it.

[0004] However, in actual application, the above existing patents have certain technical problems, such as:

[0005] 1. Only applicable to battery packs with a small number of cells

[0006] 2. Calculate each cell one by one. When the total number of cells increases, the amount of calculation of the active balancing judgment matrix increases exponentially, and the calculation cost and judgment time are too long.

[0007] 3. The calculation factors of the memory correction factor are not comprehensive enough to meet the needs of reducing costs and increasing profits. Summary of the invention

[0008] The object of the present invention is to provide a large-scale energy storage battery group balancing method based on EMD optimal planning to solve the problem of small application scope proposed in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solutions: a large-scale energy storage battery group balancing method based on EMD optimal planning, including a terminal communicating with a server through a network, a data storage system for storing data to be processed by the server, the data storage system being integrated on the server, and the terminal being a personal networked device;

[0010] The large-scale energy storage battery group balancing method based on EMD optimal planning includes the following steps:

[0011] S201: Evaluate battery SOC and SOH. By real-time monitoring of battery voltage, current and temperature data, combined with existing SOC and SOH estimation models, calculate the current SOC value of each battery. SOC reflects the current state of charge of the battery, while SOH describes the health of the battery, which is evaluated by the number of charge and discharge cycles, internal impedance and battery capacity degradation;

[0012] S202: Establishing a balancing objective function, calculating energy loss and balancing speed. In this stage, according to the SOC and SOH evaluation results of the battery, the balancing objective function is established, and the relevant energy loss and balancing speed are calculated. The energy loss includes MOSFET switching loss, copper loss and iron loss of the inductor. These losses will affect the efficiency of the balancing system.

[0013] S203: Optimizing the balancing strategy, using the Q matrix to optimize the balancing strategy, the Q matrix is ​​a matrix representing the discharge and charge amounts between batteries, wherein each element represents the amount of electricity transferred from the i-th battery to the j-th battery, and using an optimization algorithm to solve the Q matrix, so that the difference in electricity during the balancing process is minimized, the energy loss is minimized, and the balancing efficiency can be maximized;

[0014] S204: Calculating the power difference between the batteries. The power difference between the batteries is obtained by calculating the SOC value of each battery. The greater the power difference, the stronger the balancing requirement between the batteries and the more urgent the balancing process. The power difference is calculated based on the SOC value of each battery and the total capacity of the battery pack.

[0015] S205: adjusting the battery discharge and charging strategies through EMD optimal planning, and using the EMD algorithm for optimal planning to determine the battery discharge and charging strategies. The EMD algorithm is a distance measurement method for measuring the difference between two distributions, and can select the optimal energy transfer path between batteries in battery balancing;

[0016] S206: Perform active balancing and check the balancing results. This stage is the execution stage of the whole process. The main task is to perform actual battery balancing operations by controlling the battery management system according to the previously optimized balancing strategy. According to the optimized Q matrix, the balancing process will start the energy transfer between batteries. During the balancing process, the system will monitor the power and status of each battery in real time. At the same time, the system will check the balancing results and compare them with the set target SOC to determine whether the balancing is completed.

[0017] Preferably, evaluating the battery SOC and SOH in S201 is a prerequisite for balancing. SOC describes the current charge of the battery, while SOH describes the health status of the battery, including the degree of aging and remaining service life of the battery. SOC is calculated through the voltage, current and temperature parameters of the battery. The estimation method used is the Antoine model, and its calculation formula is.

[0018]

[0019] Where SOC(0) is the initial SOC, I(t) is the current, C nominal is the nominal capacity of the battery. SOH is estimated by the internal resistance and capacity decay of the battery. SOH is estimated by the open circuit voltage of the battery and the voltage change during charging. The SOH evaluation formula is:

[0020]

[0021] Preferably, in S202, the energy loss and balancing time in the battery balancing process are minimized by establishing an objective function to improve the balancing efficiency. The goal is to minimize the total energy loss and balancing time in the battery balancing process while ensuring that the battery power difference is minimized. The objective function is expressed as:

[0022] min(QuantityDiverse+EnergyLoss+RealBalanceTime)

[0023] Among them, QuantityDiverse is the difference in electricity, represented by the standard deviation of the electricity in the group; the total energy loss EnergyLoss includes the switching loss and inductance loss of the mosfet components:

[0024] EnergyLoss=(2*balance_rounds)*SwitchLoss+CopperLoss+CoreLoss*TotalBalanceTime

[0025] The single switching loss is a constant value, and the total switching loss is proportional to the number of switching times.

[0026] TotalBalanceTime is the total balance time, which is the algebraic sum of the time for each balance:

[0027] TotalBalanceTime=∑BalanceTime i

[0028] Inductor losses include losses associated with the inductor winding, commonly known as copper losses.

[0029]

[0030] R 0 is the DC internal resistance of the inductor, I RMS is the current value of a single balanced inductor, the formula is as follows

[0031]

[0032] RealBalanceTime is the actual total balancing time. The actual balancing time takes into account parallel balancing and the simultaneous operation of multiple inductors to avoid repeated calculation of time. If there are multiple inductors and multiple balancing occurs at the same time, RealBalanceTime does not repeat the calculation time, and TotalBalanceTime repeats the calculation time. RealBalanceTime = t balanceend -t balancestart .

[0033] Preferably, in S203, the Q matrix is ​​optimized to adjust the discharge and charging strategies between each battery, maximize the balancing efficiency, and minimize the energy loss. The formula of the Q matrix is ​​expressed as:

[0034]

[0035] The Q matrix is ​​an N*N matrix, where N is the number of cells, representing the discharge and charge between batteries, Qij represents the amount of power discharged from the i-th battery to the j-th battery, and the Q matrix satisfies the following constraints:

[0036] qii=0, indicating that the battery cannot discharge itself

[0037] 0≤q ij max , the total power of a single balancing does not exceed the threshold

[0038] q ij <min[q 可放 ,q 可充 ],in

[0039] Through an optimization algorithm (such as linear programming or heuristic algorithm), the power difference in the Q matrix is ​​minimized, and the maximum discharge and charge capabilities of the battery are taken into account.

[0040] Preferably, in the step S204, the power difference between the batteries is calculated, and the power difference between the batteries is calculated using the SOC of the batteries:

[0041] QuantityDiverse=StdDev(SOC 1 ,SOC 2 ,…,SOC N )​

[0042] Among them, StdDev represents the standard deviation, which is used to measure the difference in power between batteries.

[0043] Preferably, in step S205, the shortest energy transfer path between batteries is found by using an EMD algorithm to optimize the charge and discharge strategy. The EMD algorithm is used to measure the energy transfer difference between two battery packs and select the shortest path. Based on the EMD calculation result, the discharge and charge strategies between batteries are adjusted to minimize the energy difference and maximize the energy transfer efficiency. The EMD is calculated in the following manner:

[0044]

[0045] Among them, γ ij Transport flow, d ij is the distance from the ith battery to the jth battery.

[0046] Preferably, in step S206, the balancing operation is performed, and the balancing effect is checked after each balancing to ensure that the battery reaches an ideal SOC state. According to the optimization result of the Q matrix, the charge and discharge operation between each battery is performed. The balancing operation should be performed under the condition of satisfying the battery discharge and charge constraints. After each balancing, it is checked whether the SOC value of the battery reaches a predetermined target. If the difference still exists, the balancing operation is continued. The balancing end condition is:

[0047] max(SOC 1 ,SOC 2 ,…,SOC N )-min(SOC 1 ,SOC 2 ,…,SOC N )<∈

[0048] Where ∈ is the minimum allowed SOC difference, indicating that equalization is complete.

[0049] Preferably, the steps of the EMD optimization algorithm are:

[0050] S301. It is necessary to define optimization objectives and constraints, sorted by importance, and the optimization objectives are: to ensure that the SOC of the batteries in the battery pack is consistent by minimizing the difference in power between batteries; to reduce the energy loss caused by power transmission during the balancing process; and to shorten the time required for the balancing operation as much as possible. The objective function is written as:

[0051] ObjectiveFunction=α·QuantityDiverse+β·EnergyLoss+γ·RealBalanceTime

[0052] Wherein α, β, γ are weight coefficients used to balance the relative importance of different objectives, and the amount of power transferred between batteries must meet the discharge and charge capacity limits of the batteries described in S203;

[0053] S302, input data to the heuristic algorithm: Through the collected data described in S301, the particle swarm optimization algorithm is input. The particle swarm optimization algorithm is a heuristic algorithm that simulates group cooperation behavior and is applicable to optimization problems in multi-dimensional space. It simulates the movement of particle groups to find the optimal solution. In PSO, each particle represents a possible solution, that is, a Q matrix. The speed of the particle represents the adjustment range of the power transfer between batteries. PSO uses the optimization objective function as the fitness function:

[0054] Fitness=-(α·QuantityDiverse+β·EnergyLoss+γ·RealBalanceTime)

[0055] S303, perform EMD particle update optimization: the particle update formula is:

[0056] v i (t+1)=w·v i (t)+c 1 ·r 1 ·(p i -x i (t))+c 2 ·r 2 ·(g best -x i (t))x i (t+1)=x i (t)+v i (t+1)

[0057] In the above formula, vi(t) is the velocity of particle i at time t; xi(t) is the position of particle i at time t; pi is the historical optimal position of particle i; gbest is the historical optimal position of all particles; w is the inertia weight, which controls the influence of particle velocity; c1, c2 are learning factors, which control the degree of dependence of particles on personal experience and global experience; r1, r2 are random numbers in the range of [0, 1]. When the maximum number of iterations is reached or the global optimal solution no longer changes, the optimization process is stopped;

[0058] S304, providing the EMD optimal balancing operation: outputting the result to the server according to the EMD result output format described in S205, and performing the charging and discharging operation described in S206 based on the optimization result.

[0059] Compared with the prior art, the beneficial effects of the present invention are as follows: the large-scale energy storage battery group balancing method based on EMD optimal planning adopts a new structural design, and its specific contents are as follows:

[0060] 1. The present invention introduces the EMD intelligent optimization algorithm, and based on the comprehensive consideration of the power difference, energy loss and balancing time between multiple batteries, can more accurately optimize the power distribution between batteries in each balancing process. This method ensures the improvement of balancing efficiency and avoids the "over-balancing" phenomenon, that is, each balancing can more efficiently reach the target voltage / SOC, reducing the situation of incomplete balancing. The EMD optimal planning algorithm is adopted, and the Earth Mover's Distance (EMD) is used instead of the traditional Euclidean distance calculation. The battery's health state (SOH) is considered to optimize the power distribution and balancing speed between batteries, while reducing unnecessary energy loss and heat. This method ensures that the balancing process is both fast and safe, effectively avoids problems such as overcharging and over-discharging of batteries, and prolongs the battery life.

[0061] 2. The present invention proposes a multi-objective optimization model. The optimization objectives include minimum energy loss, fastest balancing speed, lowest cost and minimum heat generation. Through these comprehensive objectives, the overall optimization of the balancing process can be achieved. The EMD optimization algorithm has the advantages of strong global search capability, fast convergence speed, and convenient parallel computing when the number of parameters is large. Its randomness and group collaboration can effectively avoid the local optimal dilemma. It is suitable for high-dimensional complex optimization problems brought about by large-scale energy storage battery groups, and is easy to combine with other algorithms to improve performance. It is an ideal choice for large-scale optimization problems, reducing system power consumption and cost, while improving battery balancing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is an application environment diagram of the large-scale energy storage battery group intelligent active balancing method based on EMD optimal planning of the present invention;

[0063] Figure 2 It is a schematic diagram of the process of intelligent active balancing of a large-scale energy storage battery group based on EMD optimal planning of the present invention;

[0064] Figure 3 A schematic diagram of the process of solving the battery balancing strategy by EMD optimal planning of the present invention;

[0065] Figure 4 It is a topological diagram of a battery active balancing device according to the method for solving a battery balancing strategy based on EMD optimal planning of the present invention.

[0066] In the figure: 101, data storage system; 102, terminal; 103, communication network; 104, server. DETAILED DESCRIPTION

[0067] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] See also Figure 1-Figure 4 The present invention provides the following technical solutions: A large-scale energy storage battery group balancing method based on EMD optimal planning, in such Figure 1 In the application environment shown, the terminal 102 communicates with the server 104 via the network 103. The data storage system 101 can be used to store data that needs to be processed by the server 104, and can be integrated on the server 104, or placed on the cloud or other network servers. The terminal 102 can be various personal computers, laptops, smart phones, tablet computers, and Internet of Things devices. The server 104 can be an independent server or a server cluster composed of multiple servers.

[0069] In one embodiment, Figure 2 As shown in the figure, a large-scale energy storage battery group intelligent active balancing method based on EMD similarity metric optimal planning is provided. Figure 1 The server in the example is used to illustrate the following steps:

[0070] S201: Evaluate battery SOC and SOH: In this step, the current SOC value of each battery is calculated by real-time monitoring of battery voltage, current, temperature and other data combined with existing SOC and SOH estimation models. SOC reflects the current state of charge of the battery, while SOH describes the health of the battery, which is usually estimated by the number of charge and discharge cycles, internal impedance and battery capacity degradation.

[0071] S202: Establish a balancing objective function, calculate energy loss and balancing speed: In this stage, according to the SOC and SOH evaluation results of the battery, establish a balancing objective function, and start calculating the relevant energy loss and balancing speed. The objective function usually considers multiple factors, such as the quantity difference between batteries (QuantityDiverse), energy loss (EnergyLoss) and the time required for balancing (RealBalanceTime). The energy loss includes MOSFET switching loss, copper loss and iron loss of inductor, etc. These losses will affect the efficiency of the balancing system.

[0072] S203: Optimize balancing strategy (Q matrix): The Q matrix is ​​a matrix that represents the discharge and charge amounts between batteries, where each element represents the amount of electricity transferred from the ith battery to the jth battery. At this stage, an optimization algorithm (such as the EMD optimization method) is used to solve the Q matrix so that the difference in electricity during the balancing process is minimized, the energy loss is minimized, and the balancing efficiency is maximized. The optimization of the Q matrix must not only satisfy the balance of electricity between batteries, but also need to consider constraints such as the battery's charge and discharge capabilities, maximum current limit, SOC and SOH.

[0073] S204: Calculate the quantity difference between batteries: In this step, the quantity difference (QuantityDiverse) between batteries is obtained by calculating the SOC value of each battery. The larger the quantity difference, the stronger the need for balancing between batteries and the more urgent the balancing process. The calculation of the quantity difference is based on the SOC value of each battery and the total capacity of the battery pack.

[0074] S205: Adjust battery discharge and charging strategies through EMD optimal planning: In this stage, the EarthMover's Distance (EMD) algorithm is used for optimal planning to determine the battery discharge and charging strategies. The EMD algorithm is a distance measurement method that measures the difference between two distributions. It can select the optimal energy transfer path between batteries in battery balancing. Different from the traditional Euclidean distance, EMD can better consider the global information of the power difference between batteries, thereby finding the shortest path and realizing efficient energy transfer between batteries.

[0075] S206: Perform active balancing and check the balancing results: This stage is the execution stage of the entire process. The main task is to perform actual battery balancing operations by controlling the battery management system (BMS) according to the previously optimized balancing strategy. According to the optimized Q matrix, the balancing process will start the energy transfer between batteries, that is, charging or discharging operations. During the balancing process, the system will monitor the power and status of each battery in real time to ensure that the battery operates within a safe range and avoid problems such as overcharging, over-discharging or overheating. At the same time, the system will check the balancing results and compare them with the set target SOC to determine whether the balancing is completed. If the balancing does not reach the predetermined target, the strategy will continue to be optimized and the balancing operation will be performed again until the SOC differences of all batteries are eliminated. Through this process, the entire battery pack will reach a balanced state, thereby improving the overall efficiency and stability of the system.

[0076] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0077] In the step described in S201, evaluating the battery SOC and SOH is a prerequisite for balancing. SOC describes the current charge of the battery, while SOH describes the health status of the battery, including the degree of aging and remaining service life of the battery. SOC is calculated through battery parameters such as voltage, current, and temperature. The estimation method used is the Antoine model, and its calculation formula is:

[0078]

[0079] Where SOC(0) is the initial SOC, I(t) is the current, C nominal It is the nominal capacity of the battery. SOH is usually estimated by the internal resistance and capacity decay of the battery. SOH can be estimated by the open circuit voltage (OCV) of the battery and the voltage change during charging. The SOH evaluation formula is:

[0080]

[0081] In the step S202, by establishing an objective function, the energy loss and balancing time in the battery balancing process are minimized to improve the balancing efficiency. The goal is to minimize the total energy loss and balancing time in the battery balancing process while ensuring that the battery power difference is minimized. The objective function can be expressed as:

[0082] min(QuantityDiverse+EnergyLoss+RealBalanceTime)

[0083] Among them, QuantityDiverse is the difference in electricity, represented by the standard deviation of the electricity in the group; the total energy loss EnergyLoss includes the switching loss and inductance loss of the mosfet components:

[0084] EnergyLoss=(2*balance_rounds)*SwitchLoss+CopperLoss+CoreLoss*TotalBalanceTime

[0085] The single switching loss (SwitchLoss) is a constant value, and the total switching loss is proportional to the number of switches.

[0086] TotalBalanceTime is the total balance time, which is the algebraic sum of the time for each balance:

[0087] TotalBalanceTime=∑BalanceTime i

[0088] Inductor losses include two aspects:

[0089] Core loss CoreLoss, loss related to the magnetic core, is generally obtained by checking the manufacturer's table and can be ignored;

[0090] The losses associated with the inductor windings, commonly known as copper losses,

[0091]

[0092] R 0 is the DC internal resistance of the inductor, I RMS is the current value of a single balanced current passing through the inductor, and the formula is as follows

[0093]

[0094] RealBalanceTime is the actual total balancing time. The actual balancing time takes into account parallel balancing and the simultaneous operation of multiple inductors to avoid repeated calculation of time. The difference from TotalBalanceTime is that if there are multiple inductors and multiple balancing occurs at the same time, RealBalanceTime does not repeat the calculation of time, while TotalBalanceTime does.

[0095] RealBalanceTime = t balanceend -t balancestart

[0096] In the step described in S203, by optimizing the Q matrix, the discharge and charging strategies between each battery are adjusted to maximize the balancing efficiency and minimize the energy loss. The formula of the Q matrix is ​​expressed as:

[0097]

[0098] The Q matrix is ​​an N*N matrix, where N is the number of cells, representing the discharge and charge between batteries, Qij represents the amount of power discharged from the i-th battery to the j-th battery, and the Q matrix satisfies the following constraints:

[0099] qii=0 (indicates that the battery cannot discharge itself)

[0100] 0≤q ij max (The total power of a single balancing does not exceed the threshold)

[0101] q ij <min[q 可放 ,q 可充 ])in

[0102] Through an optimization algorithm (such as linear programming or heuristic algorithm), the charge difference in the Q matrix is ​​minimized, and the maximum discharge / charge capacity of the battery is taken into account.

[0103] In step S204, the power difference between the batteries is calculated to determine which batteries need to be charged or discharged. The power difference between the batteries is calculated using the SOC of the batteries:

[0104] QuantityDiverse=StdDev(SOC 1 ,SOC 2 ,…,SOC N )

[0105] Among them, StdDev represents the standard deviation, which is used to measure the difference in power between batteries.

[0106] In the step described in S205, the shortest energy transfer path between batteries is found through the EMD (Earth Mover's Distance) algorithm to optimize the charging and discharging strategy. The EMD algorithm is used to measure the energy transfer difference between two battery groups and select the shortest path. Based on the EMD calculation result, the discharge and charging strategies between batteries are adjusted to minimize the energy difference and maximize the energy transfer efficiency. The EMD is calculated in the following way:

[0107]

[0108] Among them, γ ij Transport flow, d ij is the distance (charge difference) from the ith battery to the jth battery.

[0109] ​In the step S206, the balancing operation is performed, and the balancing effect is checked after each balancing to ensure that the battery reaches the ideal SOC state. According to the optimization result of the Q matrix, the charge and discharge operation between each battery is performed. The balancing operation should be performed under the condition of satisfying the battery discharge / charge constraint. After each balancing, check whether the SOC value of the battery reaches the predetermined target. If the difference still exists, continue to perform the balancing operation. The balancing end condition is:

[0110] max(SOC 1 ,SOC 2 ,…,SOC N )-min(SOC 1 ,SOC 2 ,…,SOC N )<∈

[0111] Where ∈ is the minimum allowed SOC difference, indicating that equalization is complete.

[0112] The EMD optimization algorithm uses the global search and dynamic adjustment capabilities of particle swarm optimization (PSO) to effectively solve the multi-objective and multi-constrained optimization problems in battery balancing. Through the construction of objective functions, data input, particle update and balancing operation output, the entire algorithm realizes closed-loop control from model establishment to actual operation, providing an efficient and robust optimization solution for battery balancing. In energy storage systems, there are often differences in power (SOC) between batteries. In order to extend battery life and improve system efficiency, it is necessary to make the power between batteries as balanced as possible. At the same time, in order to minimize the loss generated during energy transmission during the balancing process and complete the balancing operation as quickly as possible, this algorithm simultaneously considers the following three goals:

[0113] Battery charge balancing: Minimize the charge difference between all batteries.

[0114] Reduced energy loss: Reduces energy loss caused by transferring power between batteries.

[0115] Shorter operation time: Equalization operations are completed as quickly as possible.

[0116] These three objectives are combined with certain weights to form a comprehensive EMD objective function. Collect the real-time data of each battery (such as voltage, SOC, temperature, etc.), and combine the connection relationship between batteries to form a data set that describes the state of the entire system. We need to find an optimal solution in this multidimensional data space: that is, determine an electric power transfer matrix Q. This matrix specifies in detail how to transfer electricity between each battery to meet the balance requirements. Specifically, this embodiment adopts an optimization method called "particle swarm optimization algorithm". This method imitates the group cooperation behavior of biological groups to find the optimal solution. Many intelligent agents (particles) are constructed in this algorithm. Each intelligent agent represents a possible power transfer scheme, that is, a matrix Q for transferring electricity between batteries. By designing an objective function, it is used to evaluate the quality of each candidate scheme. The lower the value of the objective function, the better the scheme performs in terms of power balance, energy loss and balance time. During the search process, each intelligent agent will update its own scheme based on two pieces of information:

[0117] The best performing program in your history;

[0118] The best solution currently found in the entire particle swarm.

[0119] Specifically, the intelligent agent will refer to these two plans and, combined with a certain degree of randomness, adjust its own "speed" and "position", that is, adjust the current plan to make it evolve in a better direction. All intelligent agents continuously update and iterate. When the predetermined number of iterations is reached or the optimal solution of the entire group is no longer significantly improved, the algorithm stops. At this time, the optimal plan is the final power transfer matrix Q.

[0120] Specifically, the steps of the EMD optimization algorithm used in this embodiment are:

[0121] First, S301, we need to define the optimization objectives and constraints. In order of importance, the optimization objectives are: to ensure that the SOC of the batteries in the battery pack is consistent by minimizing the difference in power between the batteries; to reduce the energy loss caused by power transfer during the balancing process; and to minimize the time required for the balancing operation. The objective function is written as:

[0122] Objective Function=α·QuantityDiverse+β·EnergyLoss+γ·RealBalanceTime

[0123] Wherein α, β, γ are weight coefficients used to balance the relative importance of different objectives, and the amount of power transferred between batteries must meet the discharge and charge capacity limits of the batteries described in S203.

[0124] S302 inputs data to the heuristic algorithm: Through the collected data described in S301, the particle swarm optimization algorithm is input. The particle swarm optimization algorithm is a heuristic algorithm that simulates group cooperation behavior and is applicable to optimization problems in multidimensional space. It simulates the movement of particle groups to find the optimal solution. In PSO, each particle represents a possible solution, that is, a Q matrix (battery power transfer matrix). The speed of the particle represents the adjustment range of the power transfer between batteries. PSO uses the optimization objective function as the fitness function:

[0125] Fitness=-(α·QuantityDiverse+β·EnergyLoss+γ·RealBalanceTime)

[0126] S303 performs EMD particle update optimization: the particle update formula is:

[0127] v i (t+1)=w·v i (t)+c 1 ·r 1 ·(p i -x i (t))+c 2 ·r 2 ·(g best -x i (t))x i (t+1)=x i (t)+v i (t+1)

[0128] In the above formula, vi(t) is the velocity of particle i at time t; xi(t) is the position of particle i at time t (i.e., the charge transfer matrix Q); pi is the historical optimal position of particle i; gbest is the historical optimal position of all particles; w is the inertia weight, which controls the influence of particle velocity; c1 and c2 are learning factors, which control the degree of dependence of particles on personal experience and global experience; r1 and r2 are random numbers in the range of [0,1]. When the maximum number of iterations is reached or the global optimal solution no longer changes, the optimization process is stopped.

[0129] S304 provides the EMD optimal balancing operation: outputs the result to the server 104 according to the EMD result output format described in S205, and performs the charging and discharging operation described in S206 based on the optimization result.

[0130] In this embodiment, the evaluated least squares result is input into the equalization judgment step described in S205. In this embodiment, the equalization judgment matrix Q result calculated by the model in S303 is less than the set equalization judgment threshold matrix, so the equalization is not triggered.

[0131] In other possible embodiments, if the balance judgment matrix Q calculated by S205 is greater than the set balance judgment threshold moment, balance will be triggered, and the balance execution judgment and balance strategy information will be transmitted to the terminal 102 through the communication network shown in 103 for display to the user, and at the same time, Figure 4 The battery active balancing device shown performs a balancing operation.

[0132] Select the balanced battery pair, and use the algorithm to determine the battery pair that needs to be balanced. For example, suppose Figure 4 The voltage of cell 1 shown in the figure is too high. Figure 4 The voltage of cell 2 is lower than that of cell 2. Figure 4 The cell 1 shown in Figure 4 The cell 2 shown in FIG. 2 transfers a certain amount of electricity to achieve balance, so it is necessary to connect the battery and the capacitor through a switch accordingly. Turn on the switch S 1 , Figure 4 The cell 1 shown in FIG. is connected to a capacitor to form Figure 4 The path from cell 1 to the capacitor shown in FIG. 1 is opened by switching S 2 , Figure 4 The cell 2 shown in FIG. 1 is connected to a capacitor to form a capacitor to Figure 4 The path of the battery cell 2 is shown in FIG.

[0133] because Figure 4 The voltage of cell 1 is higher, and the charge of cell 1 is transferred through S 1 In this process, the capacitor acts as a temporary storage medium to carry Figure 4 The amount of electricity released by cell 1 shown in the figure is as follows. When enough electricity is accumulated in the capacitor, the system turns off S 1 And open S 2 , transferring the charge of the capacitor to Figure 4 The battery cell 2 shown in FIG. Figure 4 The charge of the cell 2 shown in FIG. 1 gradually increases until its voltage is equal to Figure 4 The battery cell 1 shown in FIG.

[0134] During the energy transmission process, the system continuously monitors the voltage, current and other parameters of the battery and capacitor. The switch status is adjusted according to the monitoring results to avoid problems such as overcharging and over-discharging. Figure 4 The cells 1 and Figure 4 When the voltage of cell 2 shown in FIG. is equalized, turn off S 2 ,disconnect Figure 4 The connections of cell 2 and capacitor are shown in Figure 2. 1 ,disconnect Figure 4 The connection between the battery cell 1 and the capacitor is shown in FIG.

[0135] If there are multiple battery pairs in the system that need to be balanced, repeat the above steps and determine the next battery pair to be balanced according to the algorithm.

[0136] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A large-scale energy storage battery group balancing method based on EMD optimal planning, characterized in that: It includes a terminal communicating with a server through a network, a data storage system for storing data to be processed by the server, the data storage system is integrated on the server, and the terminal is a personal networked device; The large-scale energy storage battery group balancing method based on EMD optimal planning includes the following steps: S201: Evaluate battery SOC and SOH. By real-time monitoring of battery voltage, current and temperature data, combined with existing SOC and SOH estimation models, calculate the current SOC value of each battery. SOC reflects the current state of charge of the battery, while SOH describes the health of the battery, which is evaluated by the number of charge and discharge cycles, internal impedance and battery capacity degradation; S202: Establishing a balancing objective function, calculating energy loss and balancing speed. In this stage, according to the SOC and SOH evaluation results of the battery, the balancing objective function is established, and the relevant energy loss and balancing speed are calculated. The energy loss includes MOSFET switching loss, copper loss and iron loss of the inductor. These losses will affect the efficiency of the balancing system. S203: Optimizing the balancing strategy, using the Q matrix to optimize the balancing strategy, the Q matrix is ​​a matrix representing the discharge and charge amounts between batteries, wherein each element represents the amount of electricity transferred from the i-th battery to the j-th battery, and using an optimization algorithm to solve the Q matrix, so that the difference in electricity during the balancing process is minimized, the energy loss is minimized, and the balancing efficiency can be maximized; S204: Calculating the power difference between the batteries. The power difference between the batteries is obtained by calculating the SOC value of each battery. The greater the power difference, the stronger the balancing requirement between the batteries and the more urgent the balancing process. The power difference is calculated based on the SOC value of each battery and the total capacity of the battery pack. S205: adjusting the battery discharge and charging strategies through EMD optimal planning, and using the EMD algorithm for optimal planning to determine the battery discharge and charging strategies. The EMD algorithm is a distance measurement method for measuring the difference between two distributions, and can select the optimal energy transfer path between batteries in battery balancing; S206: Perform active balancing and check the balancing results. This stage is the execution stage of the whole process. The main task is to perform actual battery balancing operations by controlling the battery management system according to the previously optimized balancing strategy. According to the optimized Q matrix, the balancing process will start the energy transfer between batteries. During the balancing process, the system will monitor the power and status of each battery in real time. At the same time, the system will check the balancing results and compare them with the set target SOC to determine whether the balancing is completed.

2. A large-scale energy storage battery group balancing method based on EMD optimal planning according to claim 1, characterized in that: The evaluation of the battery SOC and SOH in S201 is a prerequisite for balancing. SOC describes the current power of the battery, while SOH describes the health status of the battery, including the degree of aging and remaining service life of the battery. SOC is calculated by the voltage, current and temperature parameters of the battery. The estimation method used is the Antoine model, and its calculation formula is: Where SOC(0) is the initial SOC, I(t) is the current, C nominal is the nominal capacity of the battery. SOH is estimated by the internal resistance and capacity decay of the battery. SOH is estimated by the open circuit voltage of the battery and the voltage change during charging. The SOH evaluation formula is:

3. The large-scale energy storage battery group balancing method based on EMD optimal planning according to claim 1 is characterized in that: In S202, the energy loss and balancing time in the battery balancing process are minimized and the balancing efficiency is improved by establishing an objective function. The goal is to minimize the total energy loss and balancing time in the battery balancing process and ensure that the battery power difference is minimized. The objective function is expressed as: min(QuantityDiverse+EnergyLoss+RealBalanceTime) Among them, QuantityDiverse is the difference in electricity, represented by the standard deviation of the electricity in the group; the total energy loss EnergyLoss includes the switching loss and inductance loss of the mosfet components: EnergyLoss=(2*balance_rounds)*SwitchLoss+CopperLoss+CoreLoss*TotalBalanceTime The single switching loss is a constant value, and the total switching loss is proportional to the number of switching times. TotalBalanceTime is the total balance time, which is the algebraic sum of the time for each balance: TotalBalanceTime=∑BalanceTime i Inductor losses include losses associated with the inductor winding, commonly known as copper losses. R0 is the DC internal resistance of the inductor, I RMS is the current value of a single balanced current passing through the inductor, and the formula is as follows RealBalanceTime is the actual total balancing time. The actual balancing time takes into account parallel balancing and the simultaneous operation of multiple inductors to avoid repeated calculation of time. If there are multiple inductors and multiple balancing occurs at the same time, RealBalanceTime does not repeat the calculation time, while TotalBalanceTime repeats the calculation time. RealBalanceTime=t balanceend -t balancestart 。 4. The large-scale energy storage battery group balancing method based on EMD optimal planning according to claim 1 is characterized in that: In S203, the Q matrix is ​​optimized to adjust the discharge and charge strategies between each battery, maximize the balancing efficiency, and minimize the energy loss. The formula of the Q matrix is ​​expressed as: The Q matrix is ​​an N*N matrix, where N is the number of cells, representing the discharge and charge between batteries, Qij represents the amount of power discharged from the i-th battery to the j-th battery, and the Q matrix satisfies the following constraints: qii=0, indicating that the battery cannot discharge itself 0≤q ij max , the total power of a single balancing does not exceed the threshold​ q ij <min[q 可放 ,q 可充 ],in Through an optimization algorithm (such as linear programming or heuristic algorithm), the power difference in the Q matrix is ​​minimized, and the maximum discharge and charge capabilities of the battery are taken into account.

5. The large-scale energy storage battery group balancing method based on EMD optimal planning according to claim 1 is characterized in that: In the step S204, the power difference between the batteries is calculated, and the power difference between the batteries is calculated using the SOC of the batteries: QuantityDiverse=StdDev(SOC1,SOC2,…,SOC N ) Among them, StdDev represents the standard deviation, which is used to measure the difference in power between batteries.

6. The large-scale energy storage battery group balancing method based on EMD optimal planning according to claim 1 is characterized in that: In the step S205, the shortest energy transfer path between batteries is found through the EMD algorithm to optimize the charge and discharge strategy. The EMD algorithm is used to measure the energy transfer difference between two battery groups and select the shortest path. Based on the EMD calculation result, the discharge and charge strategies between batteries are adjusted to minimize the energy difference and maximize the energy transfer efficiency. The EMD is calculated in the following way: Among them, γ ij Transport flow, d ij is the distance from the ith battery to the jth battery.

7. The large-scale energy storage battery group balancing method based on EMD optimal planning according to claim 1 is characterized in that: In the step S206, the balancing operation is performed, and the balancing effect is checked after each balancing to ensure that the battery reaches the ideal SOC state. According to the optimization result of the Q matrix, the charge and discharge operation between each battery is performed. The balancing operation should be performed under the condition of satisfying the battery discharge and charge constraints. After each balancing, it is checked whether the SOC value of the battery reaches the predetermined target. If the difference still exists, the balancing operation is continued. The balancing end condition is: max(SOC1,SOC2,…,SOC N )-min(SOC1,SOC2,…,SOC N )< Where ∈ is the minimum allowed SOC difference, indicating that equalization is complete.

8. The large-scale energy storage battery group balancing method based on EMD optimal planning according to claim 1 is characterized in that: The steps of the EMD optimization algorithm are: S301. It is necessary to define optimization objectives and constraints, and sort them by importance. The optimization objectives are: to ensure that the SOC of the batteries in the battery pack is consistent by minimizing the difference in power between the batteries; to reduce the energy loss caused by power transmission during the balancing process; and to shorten the time required for the balancing operation as much as possible. The objective function is written as: Objective Function=α·QuantityDiverse+β·EnergyLoss+γ·RealBalanceTime Wherein α, β, γ are weight coefficients used to balance the relative importance of different objectives, and the amount of power transferred between batteries must meet the discharge and charge capacity limits of the batteries described in S203; S302, input data to the heuristic algorithm: collect data through S301 and input it into the particle swarm optimization algorithm. The particle swarm optimization algorithm is a heuristic algorithm that simulates group cooperation behavior and is applicable to optimization problems in multidimensional space. It simulates the movement of particle groups to find the optimal solution. In PSO, each particle represents a possible solution, that is, a Q matrix. The speed of the particle represents the adjustment range of the power transfer between batteries. PSO uses the optimization objective function as the fitness function: Fitness=-(α·QuantityDiverse+β·EnergyLoss+γ·RealBalanceTime) S303, perform EMD particle update optimization: the particle update formula is: v i (t+1)=w·v i (t)+c1·r1·(p i -x i (t))+c2·r2·(g best -x i (t))x i (t+1)=x i (t)+v i (t+1) In the above formula, vi(t) is the velocity of particle i at time t; xi(t) is the position of particle i at time t; pi is the historical optimal position of particle i; gbest is the historical optimal position of all particles; w is the inertia weight, which controls the influence of particle velocity; c1, c2 are learning factors, which control the degree of dependence of particles on personal experience and global experience; r1, r2 are random numbers in the range of [0, 1]. When the maximum number of iterations is reached or the global optimal solution no longer changes, the optimization process is stopped; S304, providing the EMD optimal balancing operation: outputting the result to the server according to the EMD result output format described in S205, and performing the charging and discharging operation described in S206 based on the optimization result.

Images