Lithium ion battery pack rapid equalizing charging method considering aging

Abstract

The invention discloses a rapid equalizing charging method for a lithium ion battery pack, and belongs to the field of battery charging. The method comprises the following steps: step 1, establishing a first-order enhanced equivalent circuit model, constructing an aging model in combination with an Arrhenius equation, and comprehensively establishing an electric heating aging fusion model; 2, constructing a battery pack model through series connection based on a single battery model of the electrothermal aging fusion model; 3, configuring a deep reinforcement learning environment on the basis of the electrothermal aging coupling model, wherein the deep reinforcement learning environment is used for simulating the real state of interaction between a deep deterministic policy gradient algorithm (DDPG) and the battery pack in deep reinforcement learning; 4, a deep reinforcement learning Actor network, a Critic network and a reward function are constructed, and the target of the reward function is defined as the shortest charging time of the battery pack, SOC (state of charge) balance and minimum aging loss so as to support operation and result evaluation of a DDPG algorithm; and 5, training the model to obtain the optimal fast equalization charging current.

Description

Technical Field

[0001] The present invention belongs to the field of battery charging, and relates to a fast balancing charging method for a lithium-ion battery pack considering aging. Background Art

[0002] Among many energy storage technologies, lithium-ion batteries have become one of the main energy storage solutions due to their outstanding advantages such as high energy density, high specific energy, relatively high working voltage, low self-discharge rate, and long service life. In the actual application of lithium-ion batteries in electric vehicles, as the energy of the battery pack increases significantly, the driving range of the vehicle increases accordingly. Therefore, during the charging process, a larger charging power is required to accelerate the recovery of battery energy to meet the extended driving demand. However, the seemingly simple charging process actually involves multiple key factors, and various aspects that may affect the integrity and safety of the battery need to be carefully considered, including fast charging, thermal management constraints, and battery balancing. Currently, the optimization of charging strategies can be mainly divided into two categories. One is the model-free method, which is widely used in actual applications. It includes constant current constant voltage (CCCV) charging, multi-stage constant current (MCC) charging, multi-stage CCCV charging, etc. The other is the model-based method. Compared with the model-free method, this method can provide better optimization performance and higher robustness, and is more suitable for the charging requirements of the entire life cycle of lithium-ion batteries. However, existing research mainly focuses on a single lithium battery model and does not fully consider the balancing problem of the battery pack during the fast charging process. Currently, the battery pack balancing strategies mainly include active balancing and passive balancing. However, the balancing time of these two methods is relatively long and it is difficult to adapt to the fast charging process. Especially for the passive balancing method, it will not only cause additional capacity loss of the battery pack, but also the unbalanced battery pack cannot achieve the maximum energy utilization rate during the fast charging process, and may even accelerate battery aging. To address the above problems, a fast balancing charging method for a lithium-ion battery pack considering aging is proposed. This method is based on the deep deterministic policy gradient (DDPG) algorithm in deep reinforcement learning to solve the optimal fast balancing charging current, that is, the charging current that realizes the shortest charging time, SOC balance, and minimum aging loss of the battery pack. This method effectively solves the problems of SOC imbalance and accelerated aging of lithium-ion battery packs during the fast charging process.

[0003] After retrieval, the application publication number is CN104065117A, a method for realizing balanced charging of a series lithium-ion battery pack. The method is as follows: First, according to the charging and discharging characteristics of lithium-ion batteries, a physical model of a single lithium-ion battery is established and its various parameters are determined; Second, according to the physical model of the single battery, during the constant-current charging process, the ampere-hour method combined with the extended Kalman filtering method is used to calculate the remaining capacity and terminal voltage of each single battery at each moment; Then, when the battery pack is charged at a constant current until the terminal voltages of each single battery are discrete, the remaining capacity and the difference in remaining capacity are obtained according to the terminal voltage at this time; Finally, the battery pack is adjusted with a small current for the entire remaining capacity until the remaining capacities of each single battery are balanced.

[0004] The application publication number is CN104065117A, a method for realizing balanced charging of a series lithium-ion battery pack. When balancing the remaining capacity of the single batteries in the battery pack, separate charging and discharging operations need to be performed on the single batteries. This method has three main defects: First, the charging and discharging processes will cause energy loss; Second, the small charging current results in an extended charging time; Third, the overall balancing efficiency is low. In contrast, the present invention realizes SOC balancing by directly using the charging current, without the need for separate charging and discharging operations on the single batteries, thereby effectively avoiding the problem of energy loss, while improving the balancing efficiency and shortening the charging time. In addition, by setting the limiting conditions for battery temperature and voltage during the entire charging process, the present invention can effectively reduce the aging loss of the battery pack. Summary of the Invention

[0005] The present invention aims to solve the above problems in the prior art. A fast balanced charging method for a lithium-ion battery pack considering aging is proposed. The technical solution of the present invention is as follows:

[0006] A fast balanced charging method for a lithium-ion battery pack considering aging, which includes the following steps:

[0007] Step 1: Establish a first-order enhanced equivalent circuit model, construct an aging model in combination with the Arrhenius equation, and comprehensively establish a thermal-electrical aging fusion model;

[0008] Step 2: Based on the single-battery model of the thermal-electrical aging fusion model, construct a battery pack model by series connection;

[0009] Step 3: Configure a deep reinforcement learning environment based on the thermal-electrical aging coupling model to simulate the real state of the interaction between the deep deterministic policy gradient algorithm DDPG and the battery pack in deep reinforcement learning;

[0010] Step 4: Construct a deep reinforcement learning Actor network, a Critic network, and a reward function. The goal of the reward function is defined as: the shortest charging time of the battery pack, the balance of the SOC state of charge, and the minimum aging loss, to support the operation and result evaluation of the DDPG algorithm;

[0011] Step 5: Train the model to obtain the optimal fast balanced charging current.

[0012] Further, in Step 1: Establish a first-order enhanced equivalent circuit model, construct an aging model in combination with the Arrhenius equation, and comprehensively establish an electrothermal aging fusion model, specifically including:

[0013] The electrical model is an equivalent circuit model (ECM), defined by three basic equations: Coulomb's law equation, which determines the state of charge (1), a first-order RC model, which accounts for the diffusion voltage (2) and hysteretic kinetics (3).

[0014]

[0015] Where Q represents the total capacity of the battery cell, i(t) represents the current, the variable η(t) represents the Coulombic or charging efficiency of the battery, where η(t) equals 1 during discharge and η(t) ≤ 1 during charging. γ is a positive constant that adjusts the dynamic hysteresis of the decay rate. R1 represents the resistance and C1 represents the capacitance.

[0016] The battery terminal voltage is derived from the battery state

[0017]

[0018] Where OCV(z(t), T(t)) represents the open-circuit voltage as a function of the state of charge and temperature, the value of M determines the magnitude of the voltage generated by hysteresis, M0 represents the instantaneous hysteresis voltage, and s(t) represents the current sign.

[0019]

[0020] The thermal model considers the heat transfer between the battery core and the battery surface temperature.

[0021]

[0022] Where T c and T s represent the battery core and battery surface temperatures, C c represents the heat capacity of the battery core, and R c represents the battery core thermal conduction resistance. At the same time, the heat generated by the chemical reaction within the electrode assembly during battery operation is also considered:

[0023] Q(t) = i(t)(OCV(z(t)) - v(t)) (7)

[0024] The aging model is a capacity loss model based on the Arrhenius equation, calculating the capacity loss, the decrease in state of health (SOH), and SOH:

[0025]

[0026] B(C) = -47.836C 3 + 1215C 2 - 9418.9C + 36042 (9)

[0027] E a (c) = (31700 - 370.3·c) (10)

[0028]

[0029] N(c, T a ) = 3600Ah(c, T a ) / C n (12)

[0030]

[0031] SOH = 100 - ΔC n × 100 / 20(14)

[0032] where B is the pre - exponential factor related to the C - rate, R is the ideal gas constant, z is the power - law factor equal to 0.55, Ea is the activation energy (J / mol), Ah represents the capacity loss, N(c, T a ) represents the total number of cycles before the end of life, and ΔSOHk represents the decrease in the state of health (SOH).

[0033] Furthermore, in the said step 2: Based on the single - cell model of the electro - thermal aging fusion model, a battery pack model is constructed by series connection, specifically including:

[0034] The single - cell model is the electrical model, thermal model, and aging model established in step 1. The battery pack model is obtained by series - connecting the above - mentioned electro - thermal aging fusion model.

[0035] Furthermore, in the said step 3, based on the electro - thermal aging coupling model, a deep reinforcement learning environment is configured. The state space of the environment is defined as (SOC t , U t , T c ), where SOC t represents the state of charge SOC at the current moment, U t represents the voltage at the current moment, and T c represents the temperature at the current moment; the action space of the environment is defined as I, that is, the excitation current fed back by the Actor network.

[0036] Further, in step 4, a deep reinforcement learning Actor network, a Critic network, and a reward function are established. The input of the Actor network is defined as (SOC t , U t , T c ), where SOC t represents the state of charge (SOC) at the current moment, U t represents the voltage at the current moment, and T c represents the temperature at the current moment. Here, the input is the state of the environment. The output of the Actor network is defined as the excitation current I;

[0037] The input of the Critic network is defined as (SOC t ’, U t ’, T t ’) and the excitation current I, where (SOC t ’, U t ’, T t ’) represents the state of the environment after receiving the excitation current I and being updated. The output is defined as the reward function to evaluate the excitation current. The goal of the reward function is defined as the shortest charging time of the battery pack, the SOC balance of the battery pack, and the minimum aging loss of the battery pack. The formula is:

[0038] cost = W1·C balance + W2·C soc + C decay + C heat + C volt

[0039] reward = -cost

[0040] where W1 and W2 are the weights of the battery pack balance and the shortest charging time. At the beginning of charging, the SOC difference between battery packs is large, and W1 is greater than W2. As charging progresses, the SOC difference between battery packs decreases and is less than the balance threshold, and W2 is greater than W1. cost represents the total penalty value, reward represents the reward value, C balance represents the SOC balance penalty, C soc represents the SOC deviation penalty, that is, the difference between the average SOC value of the battery pack and the target SOC, C decay represents the aging loss penalty C heat , represents the temperature penalty, C volt represents the voltage penalty. The penalty formulas are:

[0042]

[0043]

[0044] where SOC represents the average SOC of the battery pack, SOCtarget Denote the target SOC, ΔSOH i Denote the decline in state of health (SOH), T max , T min , T i Respectively denote the maximum temperature of 50 °C, the minimum temperature of 15 °C, and the current actual temperature, V max , V min , V i Respectively denote the maximum voltage of 4.2 V, the minimum voltage of 2.3 V, and the current actual voltage.

[0045] An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, it implements the rapid equalization charging method for a lithium-ion battery pack considering aging as described in any one of the above.

[0046] A non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the rapid equalization charging method for a lithium-ion battery pack considering aging as described above. The advantages and beneficial effects of the present invention are as follows:

[0047] The present invention provides a fast equalization charging method for lithium-ion battery packs considering aging. Compared with traditional equalization charging methods that usually only consider SOC or voltage equalization, the present invention introduces temperature and aging effects into the equalization charging strategy by establishing a thermoelectric aging coupling model, enabling the battery pack to take into account both equalization and life optimization during the charging process. This model can be adjusted according to the number of batteries in different battery packs and adapt to the parameters of different types of batteries, improving the versatility and adaptability. Traditional equalization methods usually involve energy conversion or discharge between batteries in the battery pack, resulting in energy loss problems. The present invention uses the charging current to achieve SOC equalization of the battery pack without involving energy conversion or discharge between battery packs, thus avoiding energy loss problems. Traditional equalization charging methods often rely on fixed rules or optimization algorithms, while the present invention adopts the Deep Deterministic Policy Gradient (DDPG) algorithm in deep reinforcement learning, enabling the agent to optimize the equalization charging strategy through interactive learning in a dynamic environment. Traditional reinforcement learning training usually starts exploration directly, but the present invention introduces a warm-up stage of the greedy strategy. In the initial stage, the agent randomly generates actions and has not yet learned, thus promoting diverse exploration and reducing the risk of early strategy convergence to a local optimum. This method can effectively improve the learning efficiency of the model, enabling the agent to have better initial experience when officially entering the learning stage. DDPG can handle continuous action spaces and is suitable for problems of continuous current regulation, unlike discrete action reinforcement learning methods that are limited by preset discrete current values. During the training process, the environment continuously transmits state information such as SOC, voltage, and temperature. The agent calculates the optimal excitation current based on the DDPG algorithm and feeds it back to the environment to form a closed-loop control. Through the reward mechanism of reinforcement learning, the agent can continuously optimize the strategy according to the goals of the shortest charging time, SOC equalization, and minimum aging loss, so that the finally obtained charging current takes into account rapidity, equalization, and life optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 FIG. is the principle of the Deep Deterministic Policy Gradient (DDPG) algorithm provided by the preferred embodiment of the present invention.

[0049] Figure 2 FIG. is a diagram of the interaction between the agent and the environment.

[0050] Figure 3 FIG. is the SOC of the battery pack.

[0051] Figure 4 FIG. is the voltage of the battery pack.

[0052] Figure 5 FIG. is the temperature of the battery pack.

[0053] Figure 6 FIG. is the aging of the battery pack. Specific Embodiments

[0054] Next, the technical solutions in the embodiments of the present invention will be clearly and detailedly described in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention.

[0055] The technical solution for the present invention to solve the above technical problems is as follows:

[0056] A fast equalization charging method for a lithium-ion battery pack considering aging, including the following steps:

[0057] Step 1: Establish a first-order enhanced equivalent circuit model, construct an aging model in combination with the Arrhenius equation, and comprehensively establish an electrothermal aging fusion model;

[0058] Step 2: Based on the single-cell model of the electrothermal aging fusion model, construct a battery pack model by series connection;

[0059] Step 3: Configure a deep reinforcement learning environment based on the electrothermal aging coupling model to simulate the real state of the interaction between the deep deterministic policy gradient algorithm (DDPG) and the battery pack in deep reinforcement learning;

[0060] Step 4: Construct a deep reinforcement learning Actor network, Critic network, and reward function. The goal of the reward function is defined as: the shortest charging time of the battery pack, SOC (state of charge) balance, and minimum aging loss, to support the operation and result evaluation of the DDPG algorithm;

[0061] Step 5: Train the model to obtain the optimal fast equalization charging current.

[0062] Furthermore, configure a deep reinforcement learning environment based on the electrothermal aging coupling model. The state space of the environment is defined as (SOC t , U t , T c ), where SOC t represents the state of charge (SOC) at the current moment, U t represents the voltage at the current moment, and T c represents the temperature at the current moment; the action space of the environment is defined as I, that is, the excitation current fed back by the Actor network;

[0063] Furthermore, establish a deep reinforcement learning Actor network, Critic network, and reward function. The input of the Actor network is defined as (SOC t , U t , T c ), where SOC t represents the state of charge (SOC) at the current moment, U t represents the voltage at the current moment, and Tc represents the temperature at the current moment, and the output of the Actor network is defined as the excitation current I;

[0064] The input of the Critic network is defined as (SOC t ’, U t ’, T t ’) and the excitation current I, where (SOC t ’, U t ’, T t ’) represents the state of the environment after receiving the excitation current I and being updated. The output is defined as the reward function to evaluate the excitation current. The goal of the reward function is defined as the shortest charging time of the battery pack, the SOC balance of the battery pack, and the minimum aging loss of the battery pack. The formula is:

[0065] cost = W1·C balance +W2·C soc +C decay +C heat +C volt

[0066] reward = -cost

[0067] where W1 and W2 are the weights of the battery pack balance and the shortest charging time. At the initial stage of charging, the SOC difference between battery packs is large, W1 is greater than W2. As the charging progresses, the SOC difference between battery packs decreases and is less than the balance threshold, and W2 is greater than W1. The penalty formula is:

[0068]

[0069] The present invention uses a first-order enhanced equivalent circuit, a heat transfer module, and an aging module to build an electro-thermal aging coupling model based on the parameters of a 25 Ah automotive battery in Python language on PyCharm; finally, based on the built electro-thermal aging coupling model, a learning environment for deep reinforcement learning is configured, and the inputs and outputs of the Actor network and the Critic network of the deep deterministic gradient (DDPG) algorithm are determined.

[0070] Embodiments of the present invention are based on the established deep learning environment and DDPG algorithm, and train deep reinforcement learning agents. The main parameters of deep reinforcement learning are shown in Table 1, and finally the optimal fast equalization charging current is obtained. First, the agent uses the greedy strategy to enter the warm-up stage. In this stage, the agent randomly generates actions and does not start learning. As the number of warm-up steps increases, it gradually transitions to the agent learning stage. Throughout the process, the environment transmits the current battery pack status (SOC, voltage, and temperature) to the agent. The agent obtains the excitation current through the Deep Deterministic Policy Gradient (DDPG) algorithm and feeds it back to the environment. The environment calculates the battery pack status under the current excitation current and transmits it back to the agent. At the same time, the agent will reward the excitation current, store the battery pack status, excitation current, and reward value in the buffer, and judge the quality of the action based on the reward value. This process is repeated, and the agent obtains the optimal fast equalization charging current in the interaction with the environment, that is, the charging current with the shortest charging time of the battery pack, SOC balance, and minimum aging loss. The battery pack is assumed to have 6 batteries, and the initial SOCs are 0.05, 0.15, 0.2, 0.25, 0.3, and 0.35 respectively. The SOC during the entire charging process is as Figure 3 shown. As can be seen from the figure, the battery pack reaches equilibrium at 245 seconds after the start of charging, and the SOC of all batteries reaches the target at 1449 seconds, completing the charging of the battery pack. During the charging process, after the battery pack reaches SOC equilibrium, the maximum SOC difference does not exceed 0.5%, and the SOC equalization effect is good. During the charging process, the voltage also always remains within the safe voltage of 4.2V, as Figure 4 shown. The temperature during the charging process always remains within 15 - 50°C, as Figure 5 shown. The SOH drops by a maximum of 0.085%. The aging change of the electronic group during the charging process is as Figure 6 shown.

[0071] Table 1 Main parameters of deep reinforcement learning

[0072]

[0073] The systems, devices, modules, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0074] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0075] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0076] The above embodiments should be understood to be only used to illustrate the present invention and not to limit the protection scope of the present invention. After reading the contents of the present invention, technicians can make various changes or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A rapid equalization charging method for a lithium-ion battery pack considering aging, characterized in that, Including the following steps: Step 1: Establish a first-order enhanced equivalent circuit model, construct an aging model by combining with the Arrhenius equation, and comprehensively establish an electro-thermal aging fusion model; Step 2: Based on the single-cell model of the electro-thermal aging fusion model, construct a battery pack model by series connection; Step 3: Configure a deep reinforcement learning environment based on the electro-thermal aging coupling model to simulate the real state of the interaction between the deep deterministic policy gradient algorithm DDPG and the battery pack in deep reinforcement learning; Step 4: Construct a deep reinforcement learning Actor network, Critic network, and reward function. The goal of the reward function is defined as: the shortest charging time of the battery pack, the balance of the SOC state of charge, and the minimum aging loss, to support the operation and result evaluation of the DDPG algorithm; Step 5: Train the model to obtain the optimal fast equilibrium charging current.

2. A rapid equalization charging method for a lithium-ion battery pack considering aging according to claim 1, characterized in that, The said Step 1: Establish a first-order enhanced equivalent circuit model, construct an aging model by combining with the Arrhenius equation, and comprehensively establish an electro-thermal aging fusion model, specifically including: The electrical model is an equivalent circuit model (ECM), defined by three basic equations: Coulomb's law equation, which determines the state of charge (1), a first-order RC model, which illustrates the diffusion voltage (2) and hysteretic dynamics (3). Where Q represents the total capacity of the battery cell, i(t) represents the current, the variable η(t) represents the Coulomb efficiency or charging efficiency of the battery, where η(t) is equal to 1 during discharge, and η(t) ≤ 1 during charging. γ is a positive constant that adjusts the dynamic hysteresis of the decay rate. R1 represents the resistance, and C1 represents the capacitance. The battery terminal voltage is derived according to the battery state Where OCV(z(t),T(t)) represents the open-circuit voltage as a function of the state of charge and temperature, the value of M determines the magnitude of the voltage generated by hysteresis, M0 represents the instantaneous hysteresis voltage, and s(t) represents the current sign. The thermal model considers the heat transfer between the battery core and the battery surface temperature. Among them, T c and T s represent the cell core and the battery surface temperature, C c represents the heat capacity of the cell core, and R c represents the heat conduction resistance of the cell core. At the same time, the heat generated by the chemical reaction in the electrode assembly during battery operation is also considered: Q(t) = i(t)(OCV(z(t)) - v(t)) (7) The aging model is a capacity loss model based on the Arrhenius equation, which calculates the capacity loss, the decrease in the state of health (SOH), and the SOH: B(C) = -47.836C 3 + 1215C 2 -9418.9C + 36042(9) E a (c) = (31700 - 370.3·c) (10) N(c,T a ) = 3600Ah(c,T a ) / C n (12) SOH = 100 - ΔC n × 100 / 20(14) where B is the exponential pre-factor related to the C rate, R is the ideal gas constant, z is the power-law factor equal to 0.55, Ea is the activation energy (J / mol), Ah represents the capacity loss, N(c,T a ) represents the total number of cycles before end of life, and ΔSOHk represents the decrease in state of health (SOH).

3. A rapid equalization charging method for a lithium-ion battery pack considering aging according to claim 1, characterized in that, The said Step 2: Based on the single-cell model of the electro-thermal aging fusion model, construct a battery pack model by series connection, specifically including: The single-cell model refers to the electrical model, thermal model, and aging model established in Step 1; the battery pack model is the series connection of the above electro-thermal aging fusion models.

4. A rapid equalization charging method for a lithium-ion battery pack considering aging according to claim 1, characterized in that, In step 3, a deep reinforcement learning environment is configured based on the electrothermal aging coupling model. The state space of the environment is defined as (SOC t , U t , T c ), where SOC t represents the state of charge SOC at the current moment, U t represents the voltage at the current moment, and T c represents the temperature at the current moment. The action space of the environment is defined as I, that is, the excitation current fed back by the Actor network.

5. A rapid equalization charging method for a lithium-ion battery pack considering aging according to claim 1, characterized in that In step 4, a deep reinforcement learning Actor network, a Critic network, and a reward function are established. The input of the Actor network is defined as (SOC t , U t , T c ), where SOC t represents the state of charge (SOC) at the current moment, U t represents the voltage at the current moment, and T c represents the temperature at the current moment. Here, the input is the state space of the environment. The output of the Actor network is defined as the excitation current I; The input of the Critic network is defined as (SOC t ’, U t ’, T t ’) and the excitation current I, where (SOC t ’, U t ’, T t ’) represents the state of the environment after receiving the updated excitation current I. The output is defined as a reward function to evaluate the excitation current. The goal of the reward function is defined as the shortest charging time of the battery pack, the SOC balance of the battery pack, and the minimum aging loss of the battery pack. The formula is as follows: cost = W1·C balance + W2·C soc + C decay + C heat + C volt reward = -cost Where W1 and W2 are the weights for battery pack balancing and the shortest charging time. At the initial stage of charging, the SOC difference between battery packs is large, and W1 is greater than W2. As charging progresses, the SOC difference between battery packs decreases and is less than the balancing threshold, and W2 is greater than W1. cost represents the total penalty value, reward represents the reward value, C balance represents the SOC balancing penalty, C soc represents the SOC deviation penalty, that is, the difference between the average SOC value of the battery pack and the target SOC, C decay represents the aging loss penalty C heat represents the temperature penalty, C volt represents the voltage penalty. The penalty formulas are as follows: Among them, SOC represents the average SOC of the battery pack, and SOC target represents the target SOC, and ΔSOH i represents the decrease in the state of health (SOH), T max and T min and T i respectively represent the maximum temperature of 50 °C, the minimum temperature of 15 °C, and the current actual temperature, V max and V min and V i respectively represent the maximum voltage of 4.2 V, the minimum voltage of 2.3 V, and the current actual voltage.

6. An electronic device, characterized in that, Including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the method for fast equilibrium charging of a lithium-ion battery pack considering aging according to any one of claims 1 to 5.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for fast equilibrium charging of a lithium-ion battery pack considering aging according to any one of claims 1 to 5.

Images