Battery charging state estimation method and device and nonvolatile storage medium

Abstract

The invention discloses a battery charging state estimation method and device and a nonvolatile storage medium. The method comprises the following steps: acquiring real-time electrochemical impedance spectroscopy data of a target battery; and inputting the real-time electrochemical impedance spectroscopy data into a target prediction model to obtain a charging state estimation result of the target battery, the target prediction model being obtained by training non-standing electrochemical impedance spectroscopy data of the target battery based on an initial prediction model, and the initial prediction model being obtained by training the non-standing electrochemical impedance spectroscopy data of the target battery. The initial prediction model is a pre-training model obtained through training based on the standing electrochemical impedance spectroscopy data of the target battery. According to the method, the technical problem that the real-time performance required in practical application is difficult to meet due to the fact that relatively long standing time is required for SOC estimation by using the electrochemical impedance spectroscopy traditionally is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management, and in particular to a battery charge state estimation method, device and non-volatile storage medium. Background Art

[0002] In today's era of rapid technological advancement, lithium-ion batteries, as a key technology, have been widely used in everything from personal electronic devices to electric vehicles (EVs) and even grid energy storage systems, demonstrating their enormous potential and importance in the field of energy storage. Especially with the urgent global demand to reduce carbon emissions and dependence on fossil fuels, lithium-ion batteries play an indispensable role in promoting sustainable energy transformation. Accurate and real-time estimation of the battery's state of charge (SOC) is crucial to ensuring the safety and efficiency of batteries in these applications. SOC represents the current available energy of the battery and is a key function in the battery management system (BMS), directly affecting the driving range of electric vehicles and the life management of the battery.

[0003] Traditional SOC estimation methods, such as coulomb counting, are intuitive but limited in practical applications by the precise calibration and long-term maintenance of battery capacity. Model-based methods estimate SOC by constructing an equivalent circuit model (ECM) or electrochemical model (EM) of the battery. Although they can theoretically provide deeper insights into the battery state, their complexity and sensitivity to battery parameters, especially in the face of battery aging, temperature changes, and chemical reaction dynamics, limit their accuracy and practicality. Electrochemical impedance spectroscopy (EIS), as a non-invasive testing method, can provide detailed information about the internal state of the battery, including resistance, capacitance, and battery degradation. However, the typical application of EIS requires testing after the battery is completely at rest to ensure that the measurement results are not affected by battery transient effects, which poses a challenge for real-time battery state monitoring in a dynamic operating environment. In the laboratory, the rest time can be as long as several hours, which is impractical for electric vehicle application scenarios where immediate feedback is expected.

[0004] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0005] Embodiments of the present invention provide a battery state of charge estimation method, device and non-volatile storage medium to at least solve the technical problem that traditional SOC estimation using electrochemical impedance spectroscopy requires a long static time, making it difficult to meet the real-time requirements in practical applications.

[0006] According to one aspect of an embodiment of the present invention, a battery state of charge estimation method is provided, comprising: obtaining real-time electrochemical impedance spectroscopy data of a target battery; and inputting the real-time electrochemical impedance spectroscopy data into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained using non-static electrochemical impedance spectroscopy data of the target battery based on an initial prediction model, and the initial prediction model is a pre-trained model trained using static electrochemical impedance spectroscopy data of the target battery.

[0007] Optionally, the target prediction model is obtained by training the non-static electrochemical impedance spectroscopy data of the target battery based on the initial prediction model, including: obtaining the initial prediction model, the non-static electrochemical impedance spectroscopy data and the training set of the initial prediction model; determining a training set subset in the training set of the initial prediction model based on a preset ratio; merging the non-static electrochemical impedance spectroscopy data and the training set subset as the target training set; and training the initial prediction model again based on the target training set to obtain the target prediction model.

[0008] Optionally, obtaining non-static electrochemical impedance spectroscopy data includes: obtaining the battery capacity of the target battery; dividing the battery capacity based on a first preset division interval to obtain multiple first battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches multiple first battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery is left to stand for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured within a preset time interval after stopping the charging process as the non-static electrochemical impedance spectroscopy data.

[0009] Optionally, based on the target training set, the initial prediction model is trained again to obtain a target prediction model, including: determining multiple hidden layers in the initial prediction model; based on a preset fixed number of layers, determining an adjustable hidden layer among the multiple hidden layers; inputting the first initial data in the target training set into the initial prediction model to obtain a first initial loss function value; reducing the first initial loss function value by adjusting the weight and bias value of the adjustable hidden layer; repeating the above process until the first initial loss function value is less than a first preset threshold, thereby obtaining the target prediction model.

[0010] Optionally, the method for determining the initial prediction model is as follows: obtaining static electrochemical impedance spectroscopy data; constructing an original prediction model and determining multiple hidden layers in the original prediction model, wherein the original prediction model is a deep neural network model; inputting the second initial data in the static electrochemical impedance spectroscopy data into the original prediction model to obtain a second initial loss function value; reducing the second initial loss function value by adjusting the weights and bias values ​​of the multiple hidden layers in the original prediction model; repeating the above process until the second initial loss function value is less than a second preset threshold, thereby obtaining the initial prediction model.

[0011] Optionally, obtaining static electrochemical impedance spectroscopy data includes: obtaining the battery capacity of the target battery; dividing the battery capacity based on a second preset division interval to obtain multiple second battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches multiple second battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery is left at rest for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured after the target battery is left at rest for a preset time as the static electrochemical impedance spectroscopy data.

[0012] According to another aspect of an embodiment of the present invention, a battery state of charge estimation device is also provided, including: an acquisition module for acquiring real-time electrochemical impedance spectroscopy data of a target battery; a prediction module for inputting the real-time electrochemical impedance spectroscopy data into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on an initial prediction model using non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on static electrochemical impedance spectroscopy data of the target battery.

[0013] According to another aspect of an embodiment of the present invention, a non-volatile storage medium is provided. The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute any of the above-mentioned battery charge state estimation methods.

[0014] According to another aspect of an embodiment of the present invention, a computer device is provided. The computer device includes a processor, and the processor is configured to run a program. When the program is run, any one of the above-mentioned battery charge state estimation methods is executed.

[0015] According to yet another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program. When the computer program is executed by a processor, any one of the above-mentioned battery charge state estimation methods is implemented.

[0016] In an embodiment of the present invention, a battery state of charge estimation method is adopted, by obtaining real-time electrochemical impedance spectroscopy data of a target battery; the real-time electrochemical impedance spectroscopy data is input into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on an initial prediction model using non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on static electrochemical impedance spectroscopy data of the target battery, thereby achieving the purpose of using electrochemical impedance spectroscopy to perform real-time SOC estimation in a non-static state, thereby achieving the technical effect of improving the real-time performance and accuracy of battery SOC estimation, and further solving the technical problem that traditional SOC estimation using electrochemical impedance spectroscopy requires a long static time, making it difficult to meet the real-time performance requirements in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0018] Figure 1 A hardware structure block diagram of a computer terminal for implementing a battery charge state estimation method is shown;

[0019] Figure 2 is a flow chart of a method for estimating a battery charging state according to an embodiment of the present invention;

[0020] Figure 3 2. It is a schematic diagram of regression prediction lines for predicting SOC using multiple models provided in an optional embodiment of the present invention;

[0021] Figure 4 is a schematic diagram of a model with multiple adjustable numbers of layers provided according to an optional embodiment of the present invention;

[0022] Figure 5 2 is a schematic diagram of the results of real-time and steady-state EIS measurement points provided according to an optional embodiment of the present invention;

[0023] Figure 6 4 is a structural block diagram of a battery charging state estimation device provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] According to an embodiment of the present invention, an embodiment of a battery state of charge estimation method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0027] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG1 shows a hardware structure block diagram of a computer terminal for implementing a battery charge state estimation method. Figure 1 As shown, the computer terminal 10 may include one or more (illustrated as 102a, 102b, ..., 102n in the figure) processors (the processor may include but is not limited to a microprocessor MCU or a programmable logic device FPGA and other processing devices), a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0028] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry." The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be incorporated in whole or in part into any of the other components of the computer terminal 10. As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0029] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the battery state of charge estimation method in the embodiment of the present invention. The processor executes the software programs and modules stored in the memory 104 to execute various functional applications and data processing, that is, to implement the battery state of charge estimation method of the above-mentioned application. The memory 104 may include a high-speed random access memory and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely located relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0030] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0031] Figure 2 FIG. 1 is a flow chart of a method for estimating a battery charging state according to an embodiment of the present invention. Figure 2 As shown, the method includes the following steps:

[0032] Step S201 , obtaining real-time electrochemical impedance spectroscopy data of a target battery.

[0033] In this step, electrochemical impedance spectroscopy (EIS) is a technique that non-destructively evaluates the internal impedance characteristics (including resistance and reactance) of a battery by applying a small sinusoidal AC signal to the battery and measuring its voltage response. In a battery management system (BMS), EIS can provide immediate feedback on the battery's health, state of charge (SOC), and battery aging, which is crucial for optimizing battery performance and extending battery life. Traditional EIS measurements typically require the battery to rest for a long time after the charge and discharge cycle to ensure that the battery terminal voltage reaches the open circuit voltage (OCV), thereby eliminating the transient effects of the battery and obtaining stable and reliable impedance data. However, this resting process is impractical in practical applications, especially in the dynamic operating environment of electric vehicles. Therefore, in this embodiment, real-time EIS data of the target battery in a non-stationary state is selected to provide key input for real-time SOC estimation. These data are then processed and analyzed by a deep neural network (DNN) model combined with transfer learning technology, enabling accurate assessment of the battery SOC even if the battery has not reached a completely static state.

[0034] In step S202, the real-time electrochemical impedance spectroscopy data is input into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on an initial prediction model using the non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on the static electrochemical impedance spectroscopy data of the target battery.

[0035] In this step, the initial prediction model is a model established by collecting EIS data of the battery in a completely static state and training it with a deep neural network (DNN) in an experimental environment. In order to adapt to the real-time SOC prediction of the battery in a non-static state, the concept and technology of transfer learning can be introduced to transfer the knowledge contained in the initial prediction model to a new model, namely the target prediction model, so that it can perform efficient SOC estimation on non-static EIS data. The target prediction model is fine-tuned based on the DNN model (initial prediction model) with existing static EIS data training experience. During model training, by fixing some of the weights and biases in the model, the general features learned from the source task (SOC estimation of static EIS data) are retained, and the remaining parts are retrained to enable it to understand and adapt to the impedance pattern of the target battery in a non-static state. This training strategy greatly improves the adaptability and generalization ability of the model, reduces the amount of data required for retraining, and enables accurate SOC estimation using real-time EIS data even when the battery is in an unstable state, that is, without long-term rest after charging or discharging.

[0036] For example, Table 1 is a table showing the results of SOC prediction using multiple models provided according to an optional embodiment of the present invention. As shown in Table 1, compared with several other models, the prediction results of the DNN model trained using mixed data have huge advantages: MSE is 0.0063, MAE is 0.0664, R-squared is 0.9105, and MAPE is 0.2248. Figure 3 : is a schematic diagram of regression prediction lines for predicting SOC using multiple models provided in an optional embodiment of the present invention, such as Figure 3 As shown in the figure, the points correspond to the estimated SOC values, while the line represents the perfect match between the estimated and true SOC. The closer the point is to the line, the more accurate the estimation is, so it can be concluded that the DNN model trained with mixed data has the most accurate prediction results.

[0037]

[0038] Table 1 shows the results of SOC prediction by multiple models provided in an optional embodiment of the present invention.

[0039] In a laboratory environment, extensive EIS experiments were conducted on 52Ah LFP batteries and 3.6Ah NCM batteries to obtain EIS measurement data with and without rest time. The experimental results show that the DNN-TL model (target prediction model) for LFP batteries achieved high estimation accuracy when using non-rest EIS measurement data for SOC estimation, with an MSE of 0.0063 and a MAE of 0.0664. Compared with other DNN models not trained using transfer learning, the MSE is improved by up to 77.58% and the MAE is improved by 50.92%. In addition, the workload of EIS data collection is significantly reduced, and only 30% of the source dataset is required to retrain the pre-trained DNN model and achieve high SOC estimation accuracy.

[0040] Through the above steps, the goal of using electrochemical impedance spectroscopy to perform real-time SOC estimation in a non-static state is achieved, thereby achieving the technical effect of improving the real-time and accuracy of battery SOC estimation, and further solving the technical problem that traditional SOC estimation using electrochemical impedance spectroscopy requires a long static time, making it difficult to meet the real-time requirements in practical applications.

[0041] As an optional embodiment, the target prediction model is obtained by training the non-static electrochemical impedance spectroscopy data of the target battery based on the initial prediction model, including: obtaining the initial prediction model, the non-static electrochemical impedance spectroscopy data, and the training set of the initial prediction model; determining a training set subset in the training set of the initial prediction model based on a preset ratio; merging the non-static electrochemical impedance spectroscopy data and the training set subset as the target training set; and training the initial prediction model again based on the target training set to obtain the target prediction model.

[0042] Optionally, the initial prediction model is a deep neural network (DNN) model trained based on the electrochemical impedance spectroscopy (EIS) data of the target battery under static conditions. By learning the relationship between EIS data under static conditions and battery SOC, the initial prediction model accumulates rich pattern recognition capabilities, laying the foundation for subsequent model fine-tuning and optimization. From the training set of the initial prediction model, a training set subset can be determined according to a preset ratio. This training set subset is carefully selected from the existing static EIS data and is used to merge with the non-static EIS data to form a standard training set. The main purpose of selecting a preset ratio (for example, 30% of the source data set) is to reduce the demand for new data while ensuring the model training effect, thereby reducing the cost and time consumption of data acquisition. The target training set contains EIS data under static and non-static conditions, providing the model with more comprehensive training materials, enabling it to learn and adapt to the impedance pattern of the battery under different states. Based on the target training set, the initial prediction model is trained again. This process is called model fine-tuning or retraining, thereby obtaining the target prediction model.

[0043] As an optional embodiment, obtaining non-static electrochemical impedance spectroscopy data includes: obtaining the battery capacity of the target battery; dividing the battery capacity based on a first preset division interval to obtain multiple first battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches multiple first battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery is left stationary for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured within a preset time interval after stopping the charging process as the non-static electrochemical impedance spectroscopy data.

[0044] Optionally, measuring and determining the target battery's capacity is fundamental to understanding the battery's health and estimating its state of charge (SOC). Based on predefined SOC intervals (e.g., every 5% SOC), the battery capacity can be segmented to generate a series of state-of-charge thresholds. For example, if the capacity interval is 5%, the thresholds would start at 0% SOC and extend to 100% SOC, with every 5% SOC as a threshold. This segmentation allows for EIS data acquisition at different states of charge (SOCs) to construct a more comprehensive SOC-impedance data mapping. Subsequently, the target battery is subjected to a constant-current pulse charging process. During this process, the battery is charged at a constant current until the SOC reaches the first state-of-charge threshold defined above. Once the SOC reaches the threshold, the charging process is paused to allow for EIS measurements at different charge levels. The EIS data collected immediately after each pause in the constant-current pulse charging process are defined as non-resting electrochemical impedance spectroscopy (EIS) data. These data capture the impedance characteristics of the battery immediately after the charging process and before it reaches full rest.

[0045] For example, in a laboratory setting, after selecting the battery, four constant-temperature chambers were used to maintain a constant temperature of 25°C to minimize variations in experimental conditions. Two batteries were placed in each chamber and connected to power cables, signal cables, and temperature sensors. A battery cycler was used to provide DC power and load to the batteries, completing the charge and discharge process. The design and programming of the charge and discharge curves were loaded into the battery test script to control the entire charge and discharge process. A Biologic EIS test station was used to measure the EIS of the batteries at different states. The test station was equipped with eight channels specifically designed to collect EIS data. Channel selection was based on the physical distance between the EIS tester and the battery. The EIS tester parameters were pre-set to use a 500mA sinusoidal AC excitation current to characterize the dynamic response of the battery within the frequency range of 5kHz to 0.5Hz. Within this frequency range, 60 specific frequency points were selected to characterize changes in the battery's internal impedance. Four measurements were taken at each frequency point to minimize noise, and the final value was the average of these four measurements.

[0046] Eight batteries, including four 52Ah LFP cells and four 3.6Ah NCM cells, were tested in an incubator maintained at a constant temperature of 25°C. Before the experiment began, the state of health (SOH) of each battery was verified to ensure consistent initial capacity across all cells. The batteries first underwent constant current (CC) charging to a cutoff voltage of 3.65V, followed by constant voltage (CV) charging until fully charged. Subsequently, a 0.5C constant current discharge current was applied to fully discharge the batteries. The battery capacity was determined by integrating the discharge current over time. The batteries then underwent a charging cycle that included constant current pulse charging. After each 5% SOC increase, the batteries were rested for 1 hour and 40 minutes. During the rest period following each 5% SOC pulse charge cycle, EIS measurements were performed. The data obtained immediately after the charge current was interrupted constitute the non-resting EIS data.

[0047] As an optional embodiment, the initial prediction model is trained again based on the target training set to obtain the target prediction model, including: determining multiple hidden layers in the initial prediction model; determining an adjustable hidden layer among the multiple hidden layers based on a preset fixed number of layers; inputting the first initial data in the target training set into the initial prediction model to obtain a first initial loss function value; reducing the first initial loss function value by adjusting the weight and bias value of the adjustable hidden layer; repeating the above process until the first initial loss function value is less than a first preset threshold, thereby obtaining the target prediction model.

[0048] Optionally, the hidden layer is the core component of the DNN, responsible for extracting abstract features from the input data. The neurons in each layer are connected to the neurons in the previous layer through weights and bias values, and together they transform the data through nonlinear functions to capture the complex relationship between the input data and the output results (such as battery SOC). During the fine-tuning of the model for transfer learning, it is possible to determine which hidden layers’ weights and bias values ​​can be adjusted and which layers remain unchanged based on a preset fixed number of layers. By fixing the parameters of some hidden layers, the model can retain the general knowledge learned from the source task (training on static EIS data) while fine-tuning specific adjustable hidden layers to adapt to the data distribution and characteristics of the target task (training on non-static EIS data).

[0049] For example, Figure 4 Schematic diagram of a model with multiple adjustable layers according to an optional embodiment of the present invention. Figure 4As shown, lighter-colored neurons represent knowledge learned from the source task and store pre-trained weights and bias values, while darker-colored neurons represent adjustable layers. Without fixed layers, the weights of all neurons in the hidden layer can be updated based on previously stored values ​​during model retraining. In contrast, when the number of fixed layers is three, the weights of all neurons in the hidden layer remain unchanged during retraining. Given the similarity between the target and source tasks, only a small number of layers in the neural network need to be fine-tuned to transfer previously acquired knowledge to the target task. To achieve optimal performance, retaining the knowledge learned from the source task is beneficial, which can improve model training efficiency by fixing the values ​​of the weights in the hidden layers. Adjustable weights in the hidden layers that respond to newly collected data help improve accuracy on the target task. Therefore, in SOC estimation, a careful balance must be struck between model accuracy and efficiency, and the number of fixed layers in the neural network must be appropriately selected.

[0050] As an optional embodiment, the method for determining the initial prediction model is as follows: obtaining static electrochemical impedance spectroscopy data; constructing an original prediction model and determining multiple hidden layers in the original prediction model, wherein the original prediction model is a deep neural network model; inputting the second initial data in the static electrochemical impedance spectroscopy data into the original prediction model to obtain a second initial loss function value; reducing the second initial loss function value by adjusting the weights and bias values ​​of the multiple hidden layers in the original prediction model; repeating the above process until the second initial loss function value is less than a second preset threshold, thereby obtaining the initial prediction model.

[0051] Optionally, electrochemical impedance spectroscopy data of the target battery in a completely static state is obtained through an EIS experiment in the laboratory. These data are obtained after the battery is charged or discharged and has been static for a sufficiently long time, so as to obtain real impedance information reflecting the internal state and characteristics of the battery. During the pre-training stage, the DNN model absorbs knowledge from the verified EIS static data and stores this knowledge in the hidden layer neurons. Each connection between neurons in the hidden layer has a unique weight, which is continuously updated during the training stage through the Adam optimization algorithm until the optimal state is reached. Typically, battery impedance changes with different SOC levels. The DNN model learns patterns from the EIS measurement data during the pre-training stage to estimate the SOC and establish the relationship between EIS and battery capacity. The neuron weights are then used to store this information in the hidden layer and output layer.

[0052] The DNN model starts with randomly assigned weights and is fine-tuned using training data. When the training data is fed into the first DNN model, a loss function measures the difference between the estimated value and the actual battery capacity. An optimization algorithm minimizes this loss through a backpropagation process. After training on the source task, the weights stored in the neurons capture the optimal parameters needed to describe battery degradation. Through this training process, an EIS pre-trained model for SOC estimation at steady-state battery conditions is developed. The pre-training phase is also used to identify differences between EIS measurements obtained when the battery is steady-state and data obtained when the battery is not at rest. By fine-tuning the weights and biases on the resting EIS measurement data, the DNN model is able to effectively learn the differences introduced by different battery states. This meticulous fine-tuning process demonstrates the transfer of knowledge from the source task (resting EIS data) to the target task (non-resting EIS data) and provides a comprehensive understanding of the electrochemical changes that affect battery capacity.

[0053] As an optional embodiment, obtaining static electrochemical impedance spectroscopy data includes: obtaining the battery capacity of the target battery; dividing the battery capacity based on a second preset division interval to obtain multiple second battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches multiple second battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery is left at rest for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured after the target battery is left at rest for a preset time as the static electrochemical impedance spectroscopy data.

[0054] Optionally, similar to the process of obtaining non-static electrochemical impedance spectroscopy data, obtaining static electrochemical impedance spectroscopy data also requires charging the target battery by constant current pulse charging after the battery capacity state threshold is determined. When the battery SOC reaches a preset division point (the second battery capacity state threshold), charging is stopped immediately. After the battery has been stationary for a preset time, EIS data measurement is performed to obtain static electrochemical impedance spectroscopy data.

[0055] For example, Figure 5 1 is a schematic diagram of the results of a real-time and steady-state EIS measurement point provided according to an optional embodiment of the present invention. Figure 5 As shown in the figure, the first EIS measurement is performed immediately after the charging current is interrupted to capture data when the battery is in an unstable state. The second EIS measurement is performed during the 1-hour rest period after the pulse charge. During the charging and discharging process, the terminal voltage and current of the battery are recorded simultaneously. When the constant current (CC) charging reaches the cutoff voltage, the battery is considered fully charged. Therefore, the EIS measurement procedure is repeated 21 times, systematically covering every 5% SOC interval from 0% to 100% SOC.

[0056] At the end of a charge or discharge cycle, battery electrodes are in a non-equilibrium state. When the current is interrupted, the battery voltage immediately drops due to the influence of ohmic resistance and charge transfer resistance. Even after the current stops, the ions within the battery system remain in a non-equilibrium state, and diffusion processes continue, requiring the system to redistribute ion concentrations to reach equilibrium. When the ions are completely redistributed and the internal reactions stabilize, the battery enters equilibrium. During this equilibrium process, the timing of the EIS measurement has a significant impact on the results. In the high and medium frequency ranges, the effect of battery quiescence is negligible, as the ohmic resistance values ​​remain nearly constant. However, in the low frequency range, particularly in the region where charge transfer resistance dominates, the effect of battery quiescence becomes significant, with the charge transfer resistance showing an increasing trend for LFP and NCM batteries. Therefore, it can be concluded that longer battery quiescence leads to higher charge transfer resistance and diffusion resistance. EIS measurement data show that, at the same frequency, the impedance values ​​measured after one hour of quiescence (measured with an excitation current below 1 Hz) are higher than those measured without quiescence. This increase in impedance limits the application of EIS in scenarios requiring real-time state estimation.

[0057] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0058] Through the description of the above embodiments, those skilled in the art can clearly understand that the battery charge state estimation method according to the above embodiment can be implemented by software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the existing technology, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0059] According to an embodiment of the present invention, a device for implementing the above-mentioned battery charging state estimation method is also provided. Figure 6 FIG. 1 is a structural block diagram of a battery charging state estimation device according to an embodiment of the present invention. Figure 6 As shown, the device includes: an acquisition module 61 and a prediction module 62. The device is described below.

[0060] The acquisition module 61 is used to acquire real-time electrochemical impedance spectroscopy data of the target battery.

[0061] The prediction module 62 is connected to the acquisition module 61 and is used to input the real-time electrochemical impedance spectroscopy data into the target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on the initial prediction model using the non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on the static electrochemical impedance spectroscopy data of the target battery.

[0062] It should be noted that the acquisition module 61 and prediction module 62 described above correspond to steps S201 to S202 in the embodiment. The examples and application scenarios implemented by the various modules and corresponding steps are the same, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above modules, as part of the device, can be run in the computer terminal 10 provided in the embodiment.

[0063] An embodiment of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0064] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the battery state of charge estimation method and device in the embodiments of the present invention. The processor executes the software programs and modules stored in the memory to perform various functional applications and data processing, thereby implementing the above-mentioned battery state of charge estimation method. The memory may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory may further include a memory remotely located relative to the processor, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0065] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtaining real-time electrochemical impedance spectroscopy data of the target battery; inputting the real-time electrochemical impedance spectroscopy data into the target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on the initial prediction model using the non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on the static electrochemical impedance spectroscopy data of the target battery.

[0066] Optionally, the processor may also execute the program code of the following steps: a target prediction model is obtained based on the initial prediction model through training of the non-static electrochemical impedance spectroscopy data of the target battery, including: obtaining the initial prediction model, the non-static electrochemical impedance spectroscopy data, and the training set of the initial prediction model; determining a training set subset in the training set of the initial prediction model based on a preset ratio; merging the non-static electrochemical impedance spectroscopy data and the training set subset as a target training set; and training the initial prediction model again based on the target training set to obtain the target prediction model.

[0067] Optionally, the processor may also execute the program code of the following steps: obtaining non-static electrochemical impedance spectroscopy data, including: obtaining the battery capacity of the target battery; dividing the battery capacity based on a first preset division interval to obtain a plurality of first battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches a plurality of first battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery is left stationary for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured within a preset time interval after stopping the charging process as the non-static electrochemical impedance spectroscopy data.

[0068] Optionally, the processor may also execute the program code of the following steps: based on the target training set, the initial prediction model is trained again to obtain the target prediction model, including: determining multiple hidden layers in the initial prediction model; based on a preset fixed number of layers, determining an adjustable hidden layer among the multiple hidden layers; inputting the first initial data in the target training set into the initial prediction model to obtain a first initial loss function value; reducing the first initial loss function value by adjusting the weight and bias value of the adjustable hidden layer; repeating the above process until the first initial loss function value is less than a first preset threshold value to obtain the target prediction model.

[0069] Optionally, the processor may also execute the program code of the following steps: the method for determining the initial prediction model is as follows: obtaining static electrochemical impedance spectroscopy data; constructing an original prediction model and determining multiple hidden layers in the original prediction model, wherein the original prediction model is a deep neural network model; inputting the second initial data in the static electrochemical impedance spectroscopy data into the original prediction model to obtain a second initial loss function value; reducing the second initial loss function value by adjusting the weights and bias values ​​of the multiple hidden layers in the original prediction model; repeating the above process until the second initial loss function value is less than a second preset threshold value to obtain the initial prediction model.

[0070] Optionally, the processor may also execute the program code of the following steps: obtaining static electrochemical impedance spectroscopy data, including: obtaining the battery capacity of the target battery; dividing the battery capacity based on a second preset division interval to obtain a plurality of second battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches a plurality of second battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery has been left at rest for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured after the target battery has been left at rest for a preset time as the static electrochemical impedance spectroscopy data.

[0071] An embodiment of the present invention provides a method for estimating the state of charge of a battery. This method obtains real-time electrochemical impedance spectroscopy data of a target battery and inputs the real-time electrochemical impedance spectroscopy data into a target prediction model to obtain a state of charge estimation result for the target battery. The target prediction model is trained using the non-stationary electrochemical impedance spectroscopy data of the target battery based on an initial prediction model, which is a pre-trained model trained using the static electrochemical impedance spectroscopy data of the target battery. This method achieves the purpose of using electrochemical impedance spectroscopy for real-time SOC estimation in a non-stationary state, thereby achieving the technical effect of improving the real-time performance and accuracy of battery SOC estimation. This method further addresses the technical problem that conventional SOC estimation using electrochemical impedance spectroscopy requires a long static period, making it difficult to meet the real-time requirements of practical applications.

[0072] A person skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a non-volatile storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0073] The embodiment of the present invention further provides a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store the program code executed by the battery charge state estimation method provided in the above embodiment.

[0074] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0075] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining real-time electrochemical impedance spectroscopy data of the target battery; inputting the real-time electrochemical impedance spectroscopy data into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on an initial prediction model using non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on static electrochemical impedance spectroscopy data of the target battery.

[0076] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: a target prediction model is obtained by training the non-static electrochemical impedance spectroscopy data of the target battery based on the initial prediction model, including: obtaining the initial prediction model, the non-static electrochemical impedance spectroscopy data, and the training set of the initial prediction model; based on a preset ratio, determining a training set subset in the training set of the initial prediction model; merging the non-static electrochemical impedance spectroscopy data and the training set subset as a target training set; and based on the target training set, training the initial prediction model again to obtain the target prediction model.

[0077] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining non-static electrochemical impedance spectroscopy data, including: obtaining the battery capacity of the target battery; dividing the battery capacity based on a first preset division interval to obtain multiple first battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches multiple first battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery is left to stand for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured within a preset time interval after stopping the charging process as the non-static electrochemical impedance spectroscopy data.

[0078] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for executing the following steps: based on the target training set, the initial prediction model is trained again to obtain a target prediction model, including: determining multiple hidden layers in the initial prediction model; based on a preset fixed number of layers, determining an adjustable hidden layer among the multiple hidden layers; inputting the first initial data in the target training set into the initial prediction model to obtain a first initial loss function value; reducing the first initial loss function value by adjusting the weight and bias value of the adjustable hidden layer; repeating the above process until the first initial loss function value is less than a first preset threshold value to obtain the target prediction model.

[0079] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: the method for determining the initial prediction model is as follows: obtaining static electrochemical impedance spectroscopy data; constructing an original prediction model and determining multiple hidden layers in the original prediction model, wherein the original prediction model is a deep neural network model; inputting the second initial data in the static electrochemical impedance spectroscopy data into the original prediction model to obtain a second initial loss function value; reducing the second initial loss function value by adjusting the weights and bias values ​​of the multiple hidden layers in the original prediction model; repeating the above process until the second initial loss function value is less than a second preset threshold, and obtaining the initial prediction model.

[0080] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining static electrochemical impedance spectroscopy data, including: obtaining the battery capacity of the target battery; dividing the battery capacity based on a second preset division interval to obtain multiple second battery capacity state thresholds; performing constant current pulse charging on the target battery, and whenever the charging state of the target battery reaches multiple second battery capacity state thresholds, stopping the charging process, and restarting the charging process after the target battery is left at rest for a preset time; and using the electrochemical impedance spectroscopy data measurement value measured after the target battery is left at rest for a preset time as the static electrochemical impedance spectroscopy data.

[0081] An embodiment of the present invention also provides a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can achieve: obtaining real-time electrochemical impedance spectroscopy data of a target battery; inputting the real-time electrochemical impedance spectroscopy data into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on an initial prediction model using non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on static electrochemical impedance spectroscopy data of the target battery.

[0082] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0083] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0084] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0085] The units described as separate components may or may not be physically separate, and 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 may be selected according to actual needs to achieve the purpose of the present embodiment.

[0086] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0087] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, 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. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, and other media that can store program code.

[0088] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for estimating a battery state of charge, characterized in that: include: Obtain real-time electrochemical impedance spectroscopy data of the target battery; The real-time electrochemical impedance spectroscopy data is input into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained based on an initial prediction model using the non-static electrochemical impedance spectroscopy data of the target battery, and the initial prediction model is a pre-trained model trained based on the static electrochemical impedance spectroscopy data of the target battery.

2. The method according to claim 1, characterized in that The target prediction model is obtained by training the non-stationary electrochemical impedance spectroscopy data of the target battery based on the initial prediction model, including: Obtaining the initial prediction model, the non-stationary electrochemical impedance spectroscopy data, and a training set of the initial prediction model; Determining a training set subset in the training set of the initial prediction model based on a preset ratio; merging the non-stationary electrochemical impedance spectroscopy data and the training set subset as a target training set; Based on the target training set, the initial prediction model is trained again to obtain the target prediction model.

3. The method according to claim 2, characterized in that The obtaining of the non-stationary electrochemical impedance spectroscopy data comprises: Obtaining the battery capacity of the target battery; Dividing the battery capacity based on a first preset division interval to obtain a plurality of first battery capacity state thresholds; Performing constant current pulse charging on the target battery, stopping the charging process whenever the charging state of the target battery reaches the plurality of first battery capacity state thresholds, and restarting the charging process after the target battery is left to stand for a preset time; The electrochemical impedance spectroscopy data measured within a preset time interval after the charging process is stopped is used as the non-stationary electrochemical impedance spectroscopy data.

4. The method according to claim 2, characterized in that The step of training the initial prediction model again based on the target training set to obtain the target prediction model includes: determining a plurality of hidden layers in the initial prediction model; Determining an adjustable hidden layer among the plurality of hidden layers based on a preset fixed number of layers; Inputting the first initial data in the target training set into the initial prediction model to obtain a first initial loss function value; By adjusting the weight and bias value of the adjustable hidden layer, the first initial loss function value is reduced; Repeat the above process until the first initial loss function value is less than a first preset threshold, and obtain the target prediction model.

5. The method according to claim 1, characterized in that The method for determining the initial prediction model is as follows: Acquiring the static electrochemical impedance spectroscopy data; Constructing an original prediction model and determining a plurality of hidden layers in the original prediction model, wherein the original prediction model is a deep neural network model; Inputting second initial data in the static electrochemical impedance spectroscopy data into the original prediction model to obtain a second initial loss function value; By adjusting the weights and bias values ​​of the plurality of hidden layers in the original prediction model, the value of the second initial loss function is reduced; Repeat the above process until the second initial loss function value is less than a second preset threshold, and obtain the initial prediction model.

6. The method according to claim 5, characterized in that The obtaining of the static electrochemical impedance spectroscopy data comprises: Obtaining the battery capacity of the target battery; Dividing the battery capacity based on a second preset division interval to obtain a plurality of second battery capacity state thresholds; Performing constant current pulse charging on the target battery, stopping the charging process whenever the charging state of the target battery reaches the plurality of second battery capacity state thresholds, and restarting the charging process after the target battery is left to stand for a preset time; The electrochemical impedance spectroscopy data measured after the target battery is left at rest for the preset time is used as the static electrochemical impedance spectroscopy data.

7. A battery charge state estimation device, characterized in that: include: An acquisition module is used to obtain real-time electrochemical impedance spectroscopy data of the target battery; A prediction module is configured to input the real-time electrochemical impedance spectroscopy data into a target prediction model to obtain a charge state estimation result of the target battery, wherein the target prediction model is trained using the non-static electrochemical impedance spectroscopy data of the target battery based on an initial prediction model, and the initial prediction model is a pre-trained model trained using the static electrochemical impedance spectroscopy data of the target battery.

8. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the battery charge state estimation method according to any one of claims 1 to 6.

9. A computer device, characterized in that: include: memory and processor, The memory stores a computer program; The processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, the processor executes the battery state of charge estimation method according to any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the battery state of charge estimation method according to any one of claims 1 to 6 is implemented.

Images