Lithium battery state of charge determination method, apparatus, and electronic device

Abstract

The application discloses a lithium battery charging state determination method and device and electronic equipment. The method comprises the following steps: exciting a target lithium battery by using a current at a target frequency; obtaining target electrochemical impedance spectrum data and target battery voltage data of the target lithium battery under the current at the target frequency; and determining a target charging state corresponding to the target lithium battery based on the target electrochemical impedance spectrum data and the target battery voltage data, and a relationship between the electrochemical impedance spectrum data, the battery voltage data and the charging state under the current at the target frequency. The application solves the technical problem that the battery charging state cannot be accurately determined according to the battery voltage due to the flat region of the lithium iron phosphate battery voltage-battery charging state curve in the related art.

Description

Technical Field

[0001] This invention relates to the field of batteries, and more specifically, to a method, apparatus, and electronic device for determining the state of charge of a lithium battery. Background Technology

[0002] Lithium-ion batteries are widely used in various fields, especially electric vehicles, due to their high power density, reliable operating range, and long lifespan. Lithium-ion batteries often face unpredictable usage patterns, necessitating state assessment to facilitate control based on the assessment results. State of Charge (SOC) is a key indicator used to quantify the remaining usable energy in the battery. SOC is defined as the ratio of the battery's remaining capacity to its maximum capacity. Accurate SOC estimation not only reflects the battery's current state but also improves operational reliability by preventing overcharging and over-discharging. Traditional methods, such as coulomb counting and voltage-based SOC estimation, have limitations in practical applications and cannot achieve accurate SOC estimation for lithium-ion batteries.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a method, apparatus, and electronic device for determining the state of charge of a lithium battery, in order to at least solve the technical problem in the related art that the battery state of charge cannot be accurately determined based on the battery voltage due to the existence of a flat region in the voltage-battery state of charge curve of lithium iron phosphate cells.

[0005] According to one aspect of the present invention, a method for determining the state of charge of a lithium battery is provided, comprising: stimulating a target lithium battery with a current at a target frequency; acquiring target electrochemical impedance spectroscopy data and target battery voltage data of the target lithium battery at the target frequency current; and determining the target state of charge of the target lithium battery based on the target electrochemical impedance spectroscopy data and the target battery voltage data, and the relationship between the target lithium battery, the electrochemical impedance spectroscopy data, the battery voltage data and the state of charge of the target lithium battery at the target frequency current.

[0006] Optionally, before stimulating the target lithium battery with a current at the target frequency, the method further includes: charging the target lithium battery in a fully discharged state; acquiring electrochemical impedance spectroscopy (EIS) data and battery voltage data of the target lithium battery at multiple frequencies of current, respectively, when multiple states of charge are reached; determining the target frequency based on the EIS data and battery voltage data of the target lithium battery at multiple frequencies of current, respectively, when multiple states of charge are reached; and determining the relationship between the EIS data, battery voltage data and state of charge of the target lithium battery at the target frequency of current, respectively, when multiple states of charge are reached, and the multiple states of charge.

[0007] Optionally, determining the target frequency based on the electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at multiple frequencies of current when it reaches multiple states of charge includes: obtaining resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current when it reaches multiple states of charge; determining the importance of the battery voltage data, resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current when it reaches multiple states of charge; and determining the target frequency from the multiple frequencies based on the importance of the battery voltage data, resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current when it reaches multiple states of charge.

[0008] Optionally, the step of charging the target lithium battery in a fully discharged state, and acquiring electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at multiple frequencies of current when multiple charging states are reached respectively, includes: selecting the multiple charging states at predetermined intervals between a fully discharged state and a fully charged state; charging the target lithium battery in a fully discharged state, and allowing the target lithium battery to rest for a predetermined time when multiple charging states are reached respectively; stimulating the rested target lithium battery with current at the multiple frequencies respectively; and acquiring electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at the multiple frequencies of current.

[0009] Optionally, before charging the target lithium battery in a fully discharged state and acquiring electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at multiple frequencies of current when multiple charging states are reached, the method includes: performing capacity detection on the target lithium battery to determine the initial capacity of the target lithium battery, wherein the initial capacity is used to determine the multiple charging states.

[0010] Optionally, a target model is used to characterize the relationship between the electrochemical impedance spectroscopy data, battery voltage data, and state of charge of the target lithium battery at the target frequency current. The target model is trained based on a convolutional neural network using elastic network regularization.

[0011] According to another aspect of the present invention, a lithium battery charging state determination device is provided, comprising: an excitation module for exciting a target lithium battery with a current at a target frequency; a first acquisition module for acquiring target electrochemical impedance spectroscopy data and target battery voltage data of the target lithium battery at the target frequency current; and a first determination module for determining the target charging state of the target lithium battery based on the target electrochemical impedance spectroscopy data and the target battery voltage data, and the relationship between the electrochemical impedance spectroscopy data, the battery voltage data and the charging state of the target lithium battery at the target frequency current.

[0012] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is executed, it controls the device where the computer-readable storage medium is located to perform the lithium battery charging state determination method described in any one of the preceding claims.

[0013] According to another aspect of the present invention, an electronic device is provided, comprising: a memory storing an executable program; and a processor for running the program, wherein the program executes the lithium battery charging state determination method described in any one of the preceding claims.

[0014] According to another aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps of any one of the methods for determining the state of charge of a lithium battery.

[0015] In this embodiment of the invention, a target lithium battery is excited using a current at a target frequency; target electrochemical impedance spectroscopy (EIS) data and target battery voltage data of the target lithium battery at the target frequency current are acquired; based on the target EIS data and target battery voltage data, and the relationship between the EIS data, battery voltage data, and state of charge of the target lithium battery at the target frequency current, the target state of charge of the target lithium battery is determined. This achieves the goal of determining the state of charge of the lithium battery based on its EIS data and battery voltage data, thereby achieving a more accurate determination of the state of charge of the lithium battery. Furthermore, it solves the technical problem in related technologies where the voltage-state of charge curve of lithium iron phosphate cells has a flat region, making it impossible to accurately determine the state of charge of the battery based on the battery voltage. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0017] Figure 1 This is a flowchart of a method for determining the charging state of a lithium battery according to an embodiment of the present invention;

[0018] Figure 2 This is a schematic diagram of an EIS-based SOC algorithm framework according to an optional embodiment of the present invention;

[0019] Figure 3 This is a schematic diagram of cell internal resistance decomposition based on EIS according to an optional embodiment of the present invention;

[0020] Figure 4 This is a schematic diagram of the importance analysis of features based on the random forest algorithm according to an optional embodiment of the present invention;

[0021] Figure 5 This is a structural block diagram of a lithium battery charging state determination device according to an embodiment of the present invention. Detailed Implementation

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] First, some nouns or terms that appear in the description of the embodiments of this application shall be interpreted as follows:

[0025] State of Charge (SOC), also known as the state of charge, is a crucial parameter in battery technology. It quantifies the remaining usable energy in a battery. SOC is defined as the ratio of the battery's remaining capacity to its maximum capacity, and can be expressed as a percentage (0% to 100%). Accurately estimating SOC not only reflects the battery's current state but also improves battery reliability by preventing overcharging and over-discharging, making it essential for ensuring efficient and safe battery operation.

[0026] The Coulomb Counting Method, also known as the ampere-hour (AH) method, is a commonly used method in battery management technology for estimating the state of charge (SOC) of a battery. Its basic principle is to measure the current during charging and discharging, integrate these current values ​​to calculate the amount of charge already added or released, and then estimate the remaining battery capacity. The Coulomb Counting Method requires a known initial SOC value for real-time SOC calculation. Based on this initial SOC, the accumulated coulomb count is calculated by continuously monitoring the battery's charging and discharging process, thus updating the SOC value. Therefore, its accuracy depends on determining the initial SOC and accurately measuring the charging and discharging currents. Its effectiveness is limited when the initial SOC is difficult to determine. Furthermore, the cumulative error of the current sensor during long-term use can also affect the accuracy of the SOC estimation.

[0027] Due to the limitations of the coulomb counting method in practical applications, related technologies estimate the state of charge (SOC) of lithium batteries based on measurement data, such as battery voltage data obtained through various experiments. However, battery voltage is significantly affected by the chemical processes of lithium battery charging and discharging. To obtain a relatively stable battery voltage, the lithium battery needs to be allowed to rest. During the resting process, the lithium battery is in a state where it has not been used or connected to an external circuit after the charging and discharging process has ended. During this period, the internal chemical reaction of the lithium battery does not stop but continues until a chemical equilibrium is reached. The result of this process is that the battery voltage gradually changes and tends to a stable value—the open circuit voltage (OCV), which is the voltage measured after the lithium battery has been rested and without external current. There is a non-linear relationship between OCV and SOC, which is often used to construct an OCV-SOC curve to estimate the SOC of the lithium battery. However, this change in battery voltage during the resting process increases the complexity of estimating the SOC of a lithium battery based on battery voltage data.

[0028] In addition, battery voltage is also affected by other factors. For example, different charge / discharge rates lead to different battery voltage change behaviors; as the battery ages, the charge / discharge characteristics of lithium batteries change, thus affecting the battery voltage; temperature fluctuations also significantly affect battery voltage behavior. For instance, in cold environments, lithium batteries reach the charging cutoff voltage faster than at normal temperatures. The influence of these other factors on battery voltage makes it difficult to accurately estimate the State of Charge (SOC) based solely on battery voltage data.

[0029] When estimating the SOC of a lithium battery using battery voltage data based on the OCV-SOC curve, the SOC can be estimated by obtaining the battery voltage data after resting and then looking up a table. For lithium iron phosphate (LiFePO4, LFP) batteries, LiFePO4 is used as the cathode material. Since the lithium-ion insertion and extraction in the LiFePO4 cathode occurs through a two-phase reaction process, a two-phase equilibrium is formed at a specific SOC level. This allows the LFP battery to maintain a stable OCV change over a wide SOC range, resulting in a flat plateau in the OCV-SOC curve. This, in turn, limits the application of methods for estimating lithium battery SOC based on battery voltage data.

[0030] To overcome the limitations of OCV-SOC curves, this invention proposes a method for determining the state of charge (SOC) of lithium-ion batteries by combining electrochemical impedance spectroscopy (EIS) technology. EIS is a testing technique used to analyze the impedance changes of electrochemical systems, such as batteries, electrodes, and electrolytes, with frequency. By applying a small-amplitude AC signal of known frequency and amplitude and measuring the amplitude and phase of the system response, EIS can determine the impedance characteristics of the electrochemical system at different frequencies. EIS technology provides rich, non-destructive information over a wide frequency range, reflecting the internal electrochemical processes of the battery in more detail and providing a more comprehensive understanding of battery behavior and chemical reactions. Impedance here refers to the resistance to AC signals, including both resistance (the amount that impedes the flow of current) and reactance (the resistance to the flow of current, caused by inductance and capacitance). Unlike measurements such as voltage and current, the internal impedance of a battery is relatively consistent in a steady state, exhibiting small changes, which makes EIS data more reliably reflect changes in the battery's SOC.

[0031] According to an embodiment of the present invention, an embodiment of a method for determining the state of charge of a lithium battery is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0032] Figure 1 This is a flowchart of a method for determining the state of charge of a lithium battery according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes the following steps:

[0033] Step S102: Excite the target lithium battery with a current of the target frequency.

[0034] As an optional embodiment, the execution entity of this method can be a battery management system or a smart device for determining the state of charge of a lithium battery. The battery management system is responsible for monitoring key parameters such as the health status, state of charge, and temperature of the battery pack. It can directly integrate the lithium battery state of charge determination method to more accurately monitor the battery status (e.g., state of charge) and optimize battery usage strategies and extend battery life accordingly. The smart device can have a built-in processor and sensors, capable of performing electrochemical impedance spectroscopy measurements and processing data locally or in the cloud to determine the state of charge of the target lithium battery.

[0035] As an optional embodiment, the target lithium battery can be a lithium battery with a state of charge to be determined, and its type can be an LFP battery. The current at the target frequency can be an AC signal at the target frequency, which can be an AC current or an AC voltage, and its purpose is to externally stimulate the target lithium battery. By stimulating the target lithium battery with current, that is, applying a small-amplitude AC signal to the two ends of the target lithium battery, the response of the target lithium battery to the stimulation can be obtained, that is, the electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at that current can be obtained. The response is related to the frequency of stimulation; different frequencies of current can reveal the electrochemical behavior of different regions of the target lithium battery. When stimulating the target lithium battery with a current at the target frequency, a specific response of the target lithium battery to the stimulation at the target frequency can be obtained, thereby determining the target state of charge of the target lithium battery based on this specific response.

[0036] Step S104: Obtain the target electrochemical impedance spectroscopy data and target battery voltage data of the target lithium battery at the target frequency and current.

[0037] As an optional embodiment, the target electrochemical impedance spectroscopy (EIS) data can be EIS data collected by applying a current at a target frequency across the target lithium battery, and the EIS data can be in complex form. The target battery voltage data can be battery voltage data collected by applying a current at a target frequency across the target lithium battery. Specifically, when the target lithium battery is excited using a current at a target frequency, the target EIS data of the target lithium battery at the target frequency current can be calculated by measuring the response generated by the target lithium battery (e.g., the generated AC voltage), or the target battery voltage data of the target lithium battery at the target frequency current can be directly measured, thereby obtaining the target EIS data and target battery voltage data of the target lithium battery at the target frequency current.

[0038] Step S106: Based on the target electrochemical impedance spectroscopy data and target battery voltage data, as well as the relationship between the target lithium battery's electrochemical impedance spectroscopy data, battery voltage data, and state of charge at the target frequency current, determine the target state of charge corresponding to the target lithium battery.

[0039] As an optional embodiment, various methods can be used to determine the target state of charge (SOC) of a target lithium battery based on target electrochemical impedance spectroscopy (EIS) data and target battery voltage data. For example, the target SOC can be determined based on the target EIS data, target battery voltage data, and the relationship between the EIS data, battery voltage data, and SOC at a target frequency and current. The relationship between the EIS data, battery voltage data, and SOC at a target frequency and current can be determined in various ways. For instance, it can be established using historical or experimental data. This method of determining the target SOC overcomes the limitations of the OCV-SOC curve. In the flat region of the OCV-SOC curve, where the SOC cannot be determined solely based on battery voltage data, the target EIS data and target battery voltage data can be used to find the corresponding target SOC based on the relationship between the EIS data, battery voltage data, and SOC at a target frequency and current. Since electrochemical impedance spectroscopy data can provide more detailed information about the internal electrochemical processes of a battery and can show smaller fluctuations, combining electrochemical impedance spectroscopy data with battery voltage data can more accurately determine the target state of charge of a target lithium battery.

[0040] Determining the state of charge (SOC) of a battery by combining electrochemical impedance spectroscopy (EIS) data and battery voltage data avoids the problem of inaccurate SOC determination based solely on battery voltage, compared to determining it from open-circuit voltage data. This is because EIS data corresponds to different values ​​under different SOC conditions, meaning the EIS-SOC curve has a relatively pronounced slope. Therefore, the above steps achieve the goal of determining the SOC of a lithium battery based on its EIS and voltage data, resulting in a more accurate determination of the lithium battery's SOC.

[0041] As an optional embodiment, before exciting the target lithium battery with a current at the target frequency, the target frequency can be determined in various ways, and the relationship between electrochemical impedance spectroscopy (EIS) data, battery voltage data, and state of charge (SOC) can be determined. For example, the target lithium battery in a fully discharged state can be charged, and EIS data and battery voltage data at multiple frequencies can be obtained when the target lithium battery reaches multiple SOCs. Based on the EIS data and battery voltage data at multiple frequencies when the target lithium battery reaches multiple SOCs, the target frequency can be determined. Based on the EIS data and battery voltage data at the target frequency when the target lithium battery reaches multiple SOCs, and the multiple SOCs, the relationship between the EIS data, battery voltage data, and SOC at the target frequency current can be determined.

[0042] By charging a target lithium battery in a fully discharged state, and then exciting the battery with currents at multiple frequencies each time the target lithium battery reaches one of a predetermined charging state, electrochemical impedance spectroscopy (EIS) data and battery voltage data are acquired at these multiple frequencies. This allows for comprehensive coverage of EIS and voltage data across different charging states and excitation frequencies. Using this comprehensive EIS and voltage data, the excitation frequency that best describes the charging state can be determined, thus identifying the target frequency. Subsequently, based on the electrochemical impedance spectroscopy (EIS) data and battery voltage data of the target lithium battery at the target frequency current under multiple charging states, and the multiple charging states, the relationship between the EIS data, battery voltage data and charging state of the target lithium battery at the target frequency current is determined. Therefore, in determining the relationship and subsequently using the relationship to determine the charging state, only the EIS data and battery voltage data at the target frequency current need to be acquired. Compared with acquiring EIS data and battery voltage data at multiple frequencies current, this can significantly reduce the data acquisition time and effectively reduce the relationship establishment time without affecting the accuracy of the charging state determination, thereby improving the speed of charging state determination.

[0043] As an optional embodiment, various methods can be used to determine the target frequency based on electrochemical impedance spectroscopy data and battery voltage data at multiple frequencies of current when the target lithium battery reaches multiple states of charge. For example, resistance, reactance, amplitude, and phase angle data of the target lithium battery at multiple frequencies of current when the target lithium battery reaches multiple states of charge can be obtained based on the electrochemical impedance spectroscopy data of the target lithium battery at multiple frequencies of current when the target lithium battery reaches multiple states of charge; the importance of the battery voltage, resistance, reactance, amplitude, and phase angle data of the target lithium battery at multiple frequencies of current when the target lithium battery reaches multiple states of charge can be determined; and the target frequency can be determined from multiple frequencies based on the importance of the battery voltage, resistance, reactance, amplitude, and phase angle data of the target lithium battery at multiple frequencies of current when the target lithium battery reaches multiple states of charge. In electrochemical impedance spectroscopy (EIS) data, the real part characterizes resistance, and the imaginary part characterizes reactance. EIS data also possesses amplitude and phase angle characteristics. Therefore, corresponding resistance, reactance, amplitude, and phase angle data can be obtained, allowing for the representation of more information across multiple dimensions. The importance of battery voltage, resistance, reactance, amplitude, and phase angle data at multiple current frequencies when the target lithium battery reaches multiple states of charge can be used to assess their contribution to accurately determining the state of charge. For example, a random forest algorithm can be used to determine this importance. By analyzing the importance of data at multiple current frequencies, it can be determined which frequency(s) are most critical for determining the state of charge, thus making the accuracy of determining the state of charge based on data at a specific target current frequency comparable to the accuracy of determining the state of charge based on data at multiple current frequencies.

[0044] For example, taking 61 frequencies as the research object, representative characteristic parameters (such as resistance data, reactance data, amplitude data, and phase angle data) are selected at each frequency to analyze their changing trends under different charging states. Through regression models and feature importance assessment methods, the characteristic parameters that contribute most to the prediction of the charging state at different frequencies and their corresponding frequencies are identified. Based on this result, key frequencies and characteristic parameter combinations are further selected as input variables to construct a coupled correlation model between electrochemical impedance spectroscopy data and the charging state. This helps to reveal the intrinsic relationship between electrochemical impedance spectroscopy data and the charging state in the frequency domain, improving the accuracy and physical interpretability of the data-driven model in determining the charging state. For instance, for 61 frequencies, 20 charging states, and the aforementioned 5 characteristic data, the contribution of each of the 5 characteristic data to the 20 charging states under the 61 frequency excitations is determined. Therefore, a 61*20*5 contribution matrix can be obtained. The elements in this contribution matrix are sorted, and the frequency corresponding to the element with the largest contribution is selected as the target frequency.

[0045] As an optional embodiment, when charging a target lithium battery in a fully discharged state and obtaining electrochemical impedance spectroscopy (EIS) data and battery voltage data at multiple frequencies of current, after reaching multiple states of charge, various methods can be employed. For example, multiple states of charge can be selected at predetermined intervals between a fully discharged state and a fully charged state; the target lithium battery in a fully discharged state can be charged, and after reaching multiple states of charge, it can be left to rest for a predetermined time; the rested target lithium battery can be stimulated with currents at multiple frequencies; and the EIS and battery voltage data at multiple frequencies of current can be obtained. Here, the fully discharged state can be 0%, the fully charged state can be 100%, the predetermined interval can be 5%, and the predetermined time can be 1 hour. By selecting multiple states of charge at predetermined intervals between the fully discharged state and the fully charged state, the range of battery states of charge can be comprehensively covered, thereby ensuring that the experiment can capture the behavioral characteristics of the battery at different states of charge levels. When charging the target lithium battery in a fully discharged state, a constant current / constant voltage charging method can be used to charge it to the selected multiple states of charge. Once one of the selected states of charge is reached, allowing the battery to stand still can reduce transient effects and stabilize the internal chemical reactions, thereby obtaining more accurate electrochemical impedance spectroscopy data and battery voltage data that reflect the static electrochemical characteristics of the battery.

[0046] As an optional embodiment, before charging a target lithium battery in a fully discharged state and acquiring electrochemical impedance spectroscopy data and battery voltage data at multiple frequencies of current, and considering multiple states of charge, the state of charge can be determined in various ways. For example, capacity testing can be performed on the target lithium battery to determine its initial capacity, which is used to determine multiple states of charge. Since the state of charge can be expressed as the ratio of the remaining battery capacity to the maximum battery capacity, the remaining battery capacity can be obtained by charging the target lithium battery in a fully discharged state, while the maximum battery capacity can be determined by performing capacity testing on the target lithium battery to determine its initial capacity. Determining the initial capacity of the target lithium battery through capacity testing provides a more accurate basis for calculating the state of charge.

[0047] As an optional embodiment, the relationship between the electrochemical impedance spectroscopy data, battery voltage data, and state of charge of the target lithium battery at a target frequency and current can be characterized in various ways, such as using a mathematical model or a neural network model. Due to the complexity of the above relationship, using a neural network model can fit the relationship more accurately and conveniently. For example, a target model can be used to characterize the relationship between the electrochemical impedance spectroscopy data, battery voltage data, and state of charge of the target lithium battery at a target frequency and current. This target model is trained based on a convolutional neural network using elastic network regularization. Convolutional neural networks can automatically learn and extract complex features from the input electrochemical impedance spectroscopy data and battery voltage data through convolutional layers. Elastic network regularization combines the advantages of L1 and L2 regularization. L1 regularization achieves feature sparsity by compressing weights to zero, helping to eliminate irrelevant or redundant features, while L2 regularization prevents individual weights from becoming too large through weight shrinkage, ensuring a smoother distribution of model parameters. Therefore, elastic network regularization can help the target model avoid overfitting during training, while performing feature selection to improve the predictive and generalization abilities of the target model.

[0048] Through the above steps, the target lithium battery can be excited using a current at the target frequency; the target electrochemical impedance spectroscopy (EIS) data and target battery voltage data of the target lithium battery at the target frequency current can be obtained; based on the target EIS data and target battery voltage data, and the relationship between the EIS data, battery voltage data, and state of charge of the target lithium battery at the target frequency current, the target state of charge of the target lithium battery can be determined. Since the EIS data and battery voltage data of the lithium battery can more accurately reflect the state of charge of the lithium battery than the open-circuit voltage of the battery, and there is no situation where the open-circuit voltage remains flat while the state of charge of the battery changes, the purpose of determining the state of charge of the lithium battery based on the EIS data and battery voltage data is achieved. This achieves the technical effect of more accurately determining the state of charge of the lithium battery, and solves the technical problem in related technologies where the battery voltage-state of charge curve has a flat region, making it impossible to accurately determine the state of charge of the battery based on the battery voltage.

[0049] Based on the above embodiments and optional embodiments, an optional implementation method is provided. In this optional implementation method, a solution for the flat region of the OCV-SOC curve in lithium iron phosphate (LFP) cells based on EIS is proposed. This solution explores the dependency between SOC and various components of EIS (e.g., resistance, reactance, amplitude, and phase angle) through EIS-SOC experiments on LFP cells. EIS measurement data is collected every 5% within the 0% to 100% SOC range. Subsequently, the collected data is processed and analyzed to identify the correlation between EIS and battery SOC, aiming to solve the problem of the flat region of the OCV-SOC curve in LFP cells. This solution establishes the relationship between EIS features and battery SOC through a Convolutional Neural Network (CNN) combined with Elastic Net (EN) regularization (i.e., a CNN-EN model), without relying on explicit battery mathematical modeling. Using this solution, in this optional implementation method, an EIS experiment for SOC correlation analysis is designed and implemented. Figure 2 This is a schematic diagram of an EIS-based SOC algorithm framework according to an optional embodiment of the present invention, as shown below. Figure 2 As shown, the experimental steps are as follows.

[0050] S1, select three high-capacity LFP batteries (i.e., the aforementioned target lithium batteries).

[0051] It should be noted that this invention investigates the relationship between battery internal impedance and state of charge, particularly for large-capacity batteries with low internal impedance. Subsequent low-C-rate pulse charging of LFP batteries will be performed to analyze their OCV-SOC curves, especially the flat region within the medium SOC range. Experimental data will demonstrate and visualize the ability of EIS measurements to reflect SOC changes within the medium SOC range. Here, low C-rate refers to C / 10 or lower. Low C-rate charging facilitates the battery reaching steady-state conditions, resulting in more accurate EIS measurements and more precise capture of OCV changes, especially in the flat plateau region where SOC changes are subtle. The capacity of the LFP battery can be 106 Ah.

[0052] S2 performs capacity testing on three high-capacity LFP batteries.

[0053] First, each battery was placed in a constant-temperature chamber set to 25 degrees Celsius and left to stand for 2 hours to reach a stable temperature. Then, the batteries were fully charged using a constant-current-constant-voltage (CC-CV) charging method and left to stand for 1 hour. Next, the batteries were fully discharged to their minimum cutoff voltage. This charge-discharge cycle was repeated three times to activate the new batteries. The battery capacity was calculated by integrating the discharge current of the final cycle, and this capacity was defined as the initial capacity of the battery at 25 degrees Celsius. This initial capacity can be used for calculating the state of charge in subsequent experimental steps.

[0054] During capacity testing, after the battery is fully discharged, it is allowed to rest for 1 hour to reduce the impact of the subsequent battery resting period on EIS measurements. Considering the practical applications of EIS measurements, choosing a 1-hour resting time balances the impact of the battery resting period on EIS measurements with the flexibility of practical applications.

[0055] S3 performs EIS measurements on three high-capacity LFP cells.

[0056] The battery was charged from a fully discharged state, and EIS measurements were collected every 5% of the state of charge (SOC) from 0% to 100%. To characterize the battery impedance changes caused by the AC excitation current at different frequencies, the complete impedance spectrum corresponding to the AC excitation current was measured for each SOC, ranging from 0.01 Hz to 10 kHz. The impedance changes of the battery at these frequencies were visualized using Nyquist plots. Figure 3 This is a schematic diagram of cell internal resistance decomposition based on EIS according to an optional embodiment of the present invention, as shown below. Figure 3As shown in the diagram. Impedance comprises both resistance and reactance. In the Nyquist plot, the frequency of the AC excitation current decreases from left to right from 10 kHz to 0.01 Hz. The high-frequency region represents the inductance of the battery and measurement circuit; the ohmic resistance is represented by the intersection of the zero point and the real axis in the plot, reflecting the electrolyte resistance of the battery and the contact resistance of the measurement device. The mid-frequency region captures the electrochemical reactions of the battery, including solid electrolyte interface (SEI) degradation and charge transfer processes. Since SEI degradation is relatively minor, the charge transfer resistance in this range is used to approximate the battery reaction. In the low-frequency region, lithium-ion diffusion in the electrodes dominates; the real axis portion of the Nyquist plot reflects the diffusion resistance of the battery. From the EIS-SOC experiment, not only EIS measurement data but also corresponding battery voltage data were collected and correlated with OCV, which was measured one hour after pulse charging.

[0057] S4, Feature Analysis.

[0058] It should be noted that feature analysis includes consecutive steps of feature extraction and selection in order to extract and reconstruct effective features from the raw data.

[0059] Feature extraction expands the raw EIS data, which contains only real and imaginary parts, into features for the real part (resistance), imaginary part (reactance), magnitude, and phase angle. This process enhances the representational power of the raw data by providing more useful features. Feature extraction also extracts the OCV at different EIS measurement frequencies to characterize SOC variations. Finally, the raw EIS measurement data and voltage data are combined into a new dataset containing five sets of features, each consisting of 61 feature points, forming a 5x61 feature set. This dataset is organized in matrix form.

[0060] To reduce the size of the original feature set (a 5x61 feature set for each SOC sample), feature selection is used to reduce feature complexity while retaining the necessary information describing the relationship between battery internal impedance, battery voltage, and SOC. Feature selection is based on feature analysis using a random forest regression algorithm, which identifies features most strongly correlated with the target variable. In random forest regression, feature importance is measured by each feature's contribution to reducing the variance of the dataset. Feature importance is determined by calculating the total reduction in mean squared error by that feature across the entire forest. Features with a greater impact on splitting the data and reducing the variance of the target variable are considered more important, while features with lower importance scores are considered relatively irrelevant.

[0061] Through feature selection, the feature size of each sample was reduced from 5x61 to 5x1, and irrelevant frequency points were removed. The importance of each feature was calculated and compared using the random forest algorithm, and the visualization results of the most significant features are shown below. Figure 4 As shown, Figure 4 This is a schematic diagram of the importance analysis of features based on the random forest algorithm according to an optional embodiment of the present invention, wherein the EIS phase angle at 0.0125Hz (i.e., the aforementioned target frequency) is identified as the most important feature. Furthermore, a graph showing the relationship between the phase angle and SOC for the same battery is plotted to illustrate the reliability of the random forest results. In the low-frequency range, the phase angle exhibits a near-linear relationship with SOC, and similar patterns were observed in other batteries in the experiment. When the SOC increases from 0% to 100%, the phase angle increases by approximately 10 degrees, a relatively large change compared to the changes in resistance and reactance.

[0062] S5, model building.

[0063] The CNN-EN model combines convolutional neural networks (CNNs) with elastic network regularization to extract patterns from a feature set. The extracted patterns are then flattened and passed to fully connected dense layers, transforming SOC estimation into a regression problem to generate estimated SOC values. The CNN model, consisting of convolutional layers, pooling layers, flattening layers, and fully connected dense layers, excels at automatically learning hierarchical feature representations from raw data through convolutional layers, making it well-suited for extracting spatial or temporal patterns. However, it is prone to overfitting when dealing with complex datasets or limited training data. To mitigate overfitting and enhance feature selection, elastic network regularization, combining L1 and L2 regularization, is introduced into the CNN model. Elastic network regularization achieves feature selection within the CNN model by pruning unnecessary weights during training. L1 regularization introduces sparsity by driving some weights to zero, causing some filters (or neuron connections) to become inactive when the corresponding weights are zero. This selective pruning is equivalent to feature selection, allowing the model to focus on the most important filters and ignore irrelevant or noisy features. Meanwhile, L2 regularization prevents individual weights from becoming too large by shrinking the weight values, thus promoting a balanced weight distribution. Although L2 does not directly eliminate features, it complements L1 by ensuring smooth and well-distributed feature representations.

[0064] The training data for the CNN-EN model accounts for 80% of the entire dataset, with 10% further allocated as validation data to monitor the model's performance on unseen data during training.

[0065] The Adam optimizer is used for optimization through backpropagation and adaptive learning rate. The Adam optimizer starts with an initial learning rate of 0.001 and gradually adjusts the learning rate with a decay rate of 0.1 as the training epochs increase, gradually minimizing the difference between the predicted value and the target value.

[0066] The optimized hyperparameters of the CNN-EN model are shown in Table 1. Using this method, the CNN-EN model can not only extract the complex relationship between EIS features and SOC, but also effectively alleviate overfitting, while optimizing model training efficiency and performance.

[0067] Table 1. Hyperparameter selection of the CNN-EN model

[0068]

[0069] S6, Correlation analysis between EIS and SOC.

[0070] Based on the data obtained from the designed EIS-SOC experiment, the correlation between EIS and SOC was analyzed. For example, in the feature analysis step, features such as resistance, reactance, amplitude, and phase angle were extracted from the raw EIS data to characterize battery behavior at different SOC levels. The correlation between each EIS feature and SOC was analyzed to explore the potential relationship between EIS parameters and SOC.

[0071] The performance of the proposed CNN-EN model is evaluated using two different feature sets selected through random forest. The effectiveness and efficiency of each feature set are comprehensively investigated to demonstrate that EIS measurements can address the issue of flat regions in the OCV-SOC curve of LFP batteries, particularly within the medium SOC range, thus enabling accurate SOC estimation in practical applications. Experimental results show that the selected EIS features at specific frequencies accurately reflect changes in battery SOC, not only improving the accuracy of SOC estimation but also significantly reducing the time required for data acquisition.

[0072] To investigate the variation of battery internal resistance with state of charge (SOC) at different frequencies, four frequencies from high to low were selected to represent the change in battery resistance at different SOC values. First, the results show that in the relatively high-frequency region around 251 Hz, resistance reflects changes in SOC, and this frequency is mainly related to ohmic resistance. In the high-frequency region (also known as the inductive region), resistance is mainly affected by the conductivity of the electrolyte, separator, and wires. At 251 Hz, the resistance increases with increasing SOC, with a total increase of 0.02 megaohms (mOhm) for the three batteries in the experiment.

[0073] Furthermore, in the frequency range below 1 Hz (including contributions from ohmic resistance, charge transfer resistance, and diffusion resistance), the resistance exhibits a decreasing trend with increasing SOC. At 0.3981 Hz, this trend is almost linear, with a total decrease in resistance of 0.02 mOhm with increasing SOC. At 0.0125 Hz, the resistance increases between approximately 25% and 60% SOC, and decreases between 60% and 100%. This pattern is consistent across the three cells tested.

[0074] On the other hand, at high frequencies (such as 2000 Hz), the resistance failed to reflect changes in battery SOC because the experimental data did not show a consistent pattern. At these frequencies, the capacitance of the battery cells and the measurement circuitry affects the observed resistance, making the pattern undetectable or unusable for SOC estimation.

[0075] These observations indicate that the resistive component of EIS can effectively reflect changes in SOC, especially in the medium SOC range (20% to 80%) where OCV changes relatively constantly with SOC.

[0076] S7, Feature Engineering and CNN-EN Model Research.

[0077] After a comprehensive analysis of the features extracted from the EIS measurement data, the selected features were used to train the proposed CNN-EN regression model, aiming to establish a mapping relationship between EIS measurement data and SOC without explicit battery modeling. The features used for training included normalized EIS data at selected effective frequency points and data collected in the frequency range of 10kHz to 0.01Hz. The performance of the model and the effectiveness of the features were evaluated using various regression metrics to verify the robustness and reliability of the EIS features and the CNN-EN model. These regression metrics included Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Coefficient of Determination (R-squared). 2 ).

[0078] First, the CNN-EN model was trained using a full-band EIS dataset (containing resistance, reactance, amplitude, and phase angle) ranging from 10 kHz to 0.01 Hz. By utilizing the detailed information in the EIS data, the CNN-EN model achieved SOC estimation, obtaining results of MSE 0.0009, MAE 0.0227, R-squared 0.9870, and MAPE 0.0592. Although the model can accurately estimate SOC using full-band EIS data, the data collection process took 730 seconds because it required measuring the impedance at each frequency by dividing the voltage response by the excitation AC current.

[0079] To optimize data collection time and make the process more suitable for practical applications, the aforementioned feature selection method was adopted. Experimental results and random forest evaluation show that the EIS measurement at 0.0125Hz is strongly correlated with SOC in different batteries. Therefore, EIS measurements at 0.0125Hz were extracted from a total of 61 frequency points to characterize battery impedance behavior at different SOC levels. For comparison, the CNN-EN model was retrained using the same data partition (training, validation, and test data). When modeling using the CNN-EN framework, the selected 0.0125Hz EIS feature showed good performance in SOC estimation. The model achieved estimation performance of MSE 0.0012, MAE 0.02753, R-squared 0.9820, and MAPE 0.0607.

[0080] The performance of data-driven SOC estimators is highly dependent on feature quality to capture the relationship between input and target values. Although a single frequency point provides fewer features than full-band EIS data, the estimation accuracy remains essentially unchanged. Specifically, R-squared and MAPE remain almost constant, while the decreases in MSE and MAE are minimal. Table 2 shows a performance comparison between full-band EIS and single-frequency 0.0125Hz EIS in the CNN-EN model. The smaller number of features provided by a single frequency point compared to full-band EIS data results in significant savings in data acquisition and model training time. Due to the simplified structure of the EIS feature set, model training time is reduced by 6%. More importantly, the data acquisition burden is significantly reduced, requiring only the extraction of a single frequency point from the full-band EIS data, reducing data acquisition time by approximately 85%. Table 3 shows an efficiency comparison between full-band EIS and single-frequency 0.0125Hz EIS in the CNN-EN model.

[0081] This optimization of EIS measurement not only improves model training efficiency but also enhances EIS data acquisition efficiency, providing feasibility for implementing EIS measurement in practical applications and thereby strengthening the ability to estimate SOC in real-world scenarios.

[0082] Table 2. Performance comparison of full-band EIS and single-frequency point 0.0125Hz EIS in CNN-EN models.

[0083]

[0084] The above experiments verified the effectiveness of the proposed EIS-based solution for the OCV-SOC flat region of lithium iron phosphate cells in this optional embodiment. This method employs non-destructive electrochemical impedance spectroscopy to characterize the battery behavior under different SOCs. Extensive EIS-SOC experiments were conducted on LFP cells to explore the correlation between EIS and SOC. EIS measurement data were represented in various forms, and feature selection was performed using a random forest algorithm. A convolutional neural network model incorporating elastic networks was proposed for SOC estimation based on EIS features. The results of feature analysis and the random forest regression model show that 0.0125Hz EIS measurement data can effectively characterize SOC variations within the range of 0% to 100%. Furthermore, the performance of the CNN-EN model using EIS features for SOC estimation was evaluated. The results show that by utilizing only 0.0125Hz EIS measurement data, the data collection time can be reduced by approximately 85% compared to EIS measurements across the entire frequency range from 10kHz to 0.01Hz, without a significant decrease in SOC estimation accuracy. Furthermore, the results show that even using only a single frequency point, the CNN-EN model can still accurately estimate the SOC, achieving high performance with MSE of 0.0012, MAE of 0.02753, R-squared of 0.9820, and MAPE of 0.0607. This demonstrates that our method can effectively address the following issue: due to the chemical composition characteristics of LFP batteries, the battery maintains a relatively stable voltage change over a wide SOC range, which limits the effectiveness of the OCV curve in SOC estimation, especially in the medium SOC range.

[0085] According to an embodiment of the present invention, a lithium battery charging state determination device is provided. Figure 5 This is a structural block diagram of a lithium battery charging state determination device according to an embodiment of the present invention, such as... Figure 5 As shown, the device includes: an excitation module 502, a first acquisition module 504, and a first determination module 506. The device will be described below.

[0086] The excitation module 502 is used to excite the target lithium battery with a current at the target frequency; the first acquisition module 504 is connected to the excitation module 502 and is used to acquire the target electrochemical impedance spectroscopy data and target battery voltage data of the target lithium battery at the target frequency current; the first determination module 506 is connected to the first acquisition module 504 and is used to determine the target charging state of the target lithium battery based on the target electrochemical impedance spectroscopy data and target battery voltage data, and the relationship between the electrochemical impedance spectroscopy data, battery voltage data and charging state of the target lithium battery at the target frequency current.

[0087] It should be noted that the above-mentioned incentive module 502, first acquisition module 504 and first determination module 506 correspond to steps S102 to S106 in the embodiment. The multiple modules and the corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiment.

[0088] As an optional embodiment, the above-described apparatus further includes: a second acquisition module, a second determination module, and a third determination module. The second acquisition module is used to charge the target lithium battery in a fully discharged state, and, upon reaching multiple states of charge, acquire electrochemical impedance spectroscopy (EIS) data and battery voltage data of the target lithium battery at multiple frequencies of current. The second determination module, connected to the second acquisition module, is used to determine a target frequency based on the EIS data and battery voltage data of the target lithium battery at multiple frequencies of current, upon reaching multiple states of charge. The third determination module, connected to the second determination module, is used to determine the relationship between the EIS data, battery voltage data, and states of charge of the target lithium battery at a target frequency of current, based on the EIS data and battery voltage data of the target lithium battery at a target frequency of current, upon reaching multiple states of charge, and the multiple states of charge.

[0089] As an optional embodiment, the second determining module includes: a first acquiring unit, a first determining unit, and a second determining unit. The first acquiring unit is used to acquire resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current under multiple states of charge, based on electrochemical impedance spectroscopy data of the target lithium battery at multiple states of charge and at multiple frequencies of current. The first determining unit, connected to the first acquiring unit, is used to determine the importance of the battery voltage data, resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple states of charge and at multiple frequencies of current. The second determining unit, connected to the first determining unit, is used to determine a target frequency from multiple frequencies based on the importance of the battery voltage data, resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple states of charge and at multiple frequencies of current.

[0090] As an optional embodiment, the second acquisition module includes: a selection unit, a resting unit, an excitation unit, and a second acquisition unit. The selection unit is used to select multiple charging states at predetermined intervals between a fully discharged state and a fully charged state; the resting unit, connected to the selection unit, is used to charge the target lithium battery in the fully discharged state, and, after reaching multiple charging states, to rest the target lithium battery for a predetermined time; the excitation unit, connected to the resting unit, is used to excite the rested target lithium battery with currents of multiple frequencies; the second acquisition unit, connected to the excitation unit, is used to acquire electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery under currents of multiple frequencies.

[0091] As an optional embodiment, the second acquisition module further includes a detection unit. The detection unit is used to perform capacity detection on the target lithium battery to determine the initial capacity of the target lithium battery, wherein the initial capacity is used to determine multiple states of charge.

[0092] As an optional embodiment, the third determining module mentioned above includes a characterization unit. The characterization unit is used to characterize the relationship between the electrochemical impedance spectroscopy data, battery voltage data, and state of charge of the target lithium battery at a target frequency current using a target model. The target model is trained based on a convolutional neural network using elastic network regularization.

[0093] According to an embodiment of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is executed, it controls the device where the computer-readable storage medium is located to execute the lithium battery charging state determination method described in any one of the above-mentioned methods.

[0094] According to an embodiment of the present invention, an electronic device is provided, comprising: a memory storing an executable program; and a processor for running the program, wherein the program executes the lithium battery charging state determination method described in any one of the preceding embodiments.

[0095] According to an embodiment of the present invention, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of any of the methods described above.

[0096] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0097] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0098] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0099] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0100] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0101] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0102] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for determining the state of charge of a lithium battery, characterized in that, include: The target lithium battery is excited using a current at the target frequency; Obtain the target electrochemical impedance spectroscopy data and target battery voltage data of the target lithium battery at the target frequency current; Based on the target electrochemical impedance spectroscopy data and the target battery voltage data, as well as the relationship between the target lithium battery's electrochemical impedance spectroscopy data, battery voltage data, and state of charge at the target frequency current, the target state of charge corresponding to the target lithium battery is determined. The target lithium battery is a lithium iron phosphate battery. The method further includes: charging the target lithium battery in a fully discharged state before determining the target frequency; acquiring electrochemical impedance spectroscopy (EIS) data and battery voltage data of the target lithium battery at multiple frequencies of current when reaching multiple charging states; acquiring resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current when reaching multiple charging states, respectively, using a random forest algorithm; determining the importance of the battery voltage data, resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current when reaching multiple charging states, respectively, to the charging state; and determining the target frequency from the multiple frequencies based on the importance, such that the EIS data or battery voltage data at the target frequency has the highest contribution to the prediction of the charging state. The target model is used to characterize the relationship between the electrochemical impedance spectroscopy data, battery voltage data and state of charge of the target lithium battery at the target frequency current. The target model is trained based on a convolutional neural network using elastic network regularization.

2. The method according to claim 1, characterized in that, The process of charging the target lithium battery in a fully discharged state, and acquiring electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at multiple frequencies of current under various charging states, includes: The plurality of charging states are selected at predetermined intervals between a fully discharged state and a fully charged state; The target lithium battery in a fully discharged state is charged, and when multiple charging states are reached, the target lithium battery is left to stand for a predetermined time. The target lithium battery after resting is excited by currents of the multiple frequencies respectively; The electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at currents at multiple frequencies are obtained.

3. The method according to claim 1, characterized in that, Before charging the target lithium battery in a fully discharged state, and before acquiring electrochemical impedance spectroscopy data and battery voltage data of the target lithium battery at multiple frequencies of current, respectively, when multiple states of charge are reached, the process includes: The target lithium battery is subjected to capacity detection to determine the initial capacity of the target lithium battery, wherein the initial capacity is used to determine the plurality of charging states.

4. A lithium battery charging state determination device, characterized in that, include: The excitation module is used to excite the target lithium battery with a current at the target frequency; The first acquisition module is used to acquire the target electrochemical impedance spectroscopy data and target battery voltage data of the target lithium battery at the target frequency current. The first determining module is used to determine the target charging state corresponding to the target lithium battery based on the target electrochemical impedance spectroscopy data and the target battery voltage data, as well as the relationship between the electrochemical impedance spectroscopy data, battery voltage data and charging state of the target lithium battery at the target frequency current. The target lithium battery is a lithium iron phosphate battery. The device is further configured to charge the target lithium battery in a fully discharged state before determining the target frequency, and to acquire electrochemical impedance spectroscopy (EIS) data and battery voltage data of the target lithium battery at multiple frequencies of current when it reaches multiple charging states. Based on the EIS data of the target lithium battery at multiple frequencies of current when it reaches multiple charging states, the device acquires resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current when it reaches multiple charging states. Using a random forest algorithm, the device determines the importance of the battery voltage data, resistance data, reactance data, amplitude data, and phase angle data of the target lithium battery at multiple frequencies of current when it reaches multiple charging states to the charging states. Based on the importance, the device determines the target frequency from the multiple frequencies such that the EIS data or battery voltage data at the target frequency contributes the most to the prediction of the charging state. The target model is used to characterize the relationship between the electrochemical impedance spectroscopy data, battery voltage data and state of charge of the target lithium battery at the target frequency current. The target model is trained based on a convolutional neural network using elastic network regularization.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device containing the computer-readable storage medium to perform the lithium battery charging state determination method according to any one of claims 1 to 3.

6. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program executes the lithium battery charging state determination method according to any one of claims 1 to 3.

7. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 3.

Images