Battery capacity rapid estimation method based on electrochemical impedance spectroscopy

Abstract

The invention discloses a rapid battery capacity estimation method based on an electrochemical impedance spectroscopy, which comprises the following steps: S1, acquiring the electrochemical impedance spectroscopy of a battery of which the capacity is to be estimated and the environment temperature and the battery end voltage during measurement as characteristic data; s2, judging whether the combination of the environment temperature and the battery end voltage during measurement is a non-standard measurement condition or not; when it is judged that the combination of the environment temperature and the battery end voltage during measurement is a non-standard measurement condition, the next step is carried out; s3, inputting the electrochemical impedance spectroscopy into a pre-trained feature matching model to obtain electrochemical impedance spectroscopy data matched to a standard measurement condition; and S4, inputting the electrochemical impedance spectrum under the standard measurement condition into a pre-trained battery capacity estimation model to obtain an estimated value of the battery capacity. According to the invention, the method has the advantages of extremely short measurement time, capability of realizing capacity estimation under any environment temperature and battery end voltage conditions and the like, and is of great significance for improving the capacity estimation speed of a residual value evaluation link in a battery recovery stage.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management, and in particular to a method for quickly estimating battery capacity based on electrochemical impedance spectroscopy. Background Art

[0002] Battery capacity is a key indicator in the residual value assessment link of the battery recycling stage. At present, accurate battery capacity calibration relies on complete charge and discharge tests. According to the product specification, under the preset constant temperature conditions, the battery needs to be charged to the highest cut-off voltage with a preset current size, and the constant voltage charging needs to be maintained until the charging current is less than the preset cut-off current. Then, the battery is discharged to the lowest cut-off voltage with a preset current size, and finally the discharge capacity is used as the capacity calibration value of the tested battery. The above process has a long measurement time and high power consumption, which increases the cost of battery recycling. Therefore, increasing the speed of obtaining battery capacity is of great significance to reducing costs and increasing efficiency in battery recycling.

[0003] Invention patent CN118671598A discloses a method for rapid estimation of battery capacity, which proposes to use the current-time data of the battery constant voltage charging stage and the static stage after constant voltage charging as features, and input the pre-trained battery capacity estimation model to obtain the battery capacity estimation value. However, this technology requires that the battery terminal voltage of the battery under test during the constant voltage charging stage is consistent with the training data. When the battery terminal voltage is uncertain, it still takes a long time to adjust the battery terminal voltage, and the speed of obtaining the battery capacity is limited.

[0004] Invention patent CN110927591B discloses a method for rapid estimation of battery capacity. By measuring the battery charging capacity and voltage, the incremental capacity (IC) curve is calculated, and the IC curve area from the voltage corresponding to the peak point of the IC curve to the preset voltage increment range is used as the health factor. The health factor is corrected based on temperature and charging rate, and the battery capacity estimation value is obtained after inputting the pre-trained battery capacity estimation model. However, the peak point of the IC curve in this technology appears in the special terminal voltage range of the battery, and the battery terminal voltage needs to be adjusted first, and the speed of obtaining the battery capacity is also limited.

[0005] Invention patent CN115097317B discloses a method for rapid estimation of battery capacity. Starting from any battery terminal voltage, the battery can be charged / discharged to each adjacent equal voltage interval sub-interval and aging features can be extracted. The battery capacity estimation value can be obtained after inputting the pre-trained battery capacity estimation model. This technology can obtain the battery capacity in a relatively short time and is not constrained by the initial conditions of the battery terminal voltage, but requires the battery to be measured under preset temperature conditions, which is not suitable for application scenarios with limited temperature control conditions.

[0006] Invention patent CN111337843B discloses a battery differential capacitance calculation method based on electrochemical impedance spectroscopy, and then uses the battery differential capacitance as a feature and inputs a pre-trained battery capacity estimation model to obtain a battery capacity estimation value. However, this method will fail when the electrochemical impedance spectroscopy measurement conditions of the battery under test (including ambient temperature and battery terminal voltage) are inconsistent with the electrochemical impedance spectroscopy measurement conditions used to train the model. Summary of the invention

[0007] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a method for quickly estimating battery capacity based on electrochemical impedance spectroscopy, which has the advantages of extremely short measurement time and the ability to achieve capacity estimation under any ambient temperature and battery terminal voltage conditions, and is of great significance for improving the capacity estimation speed of the residual value assessment link in the battery recycling stage. In order to achieve the above-mentioned purpose and other advantages according to the present invention, a method for quickly estimating battery capacity based on electrochemical impedance spectroscopy is provided, comprising: S1. Obtaining the electrochemical impedance spectrum of the battery whose capacity is to be estimated, as well as the ambient temperature and battery terminal voltage during measurement, as characteristic data; S2, judging whether the combination of the ambient temperature and the battery terminal voltage during measurement is a non-standard measurement condition; if it is judged that the combination of the ambient temperature and the battery terminal voltage during measurement is a non-standard measurement condition, proceeding to the next step; if it is judged that the combination of the ambient temperature and the battery terminal voltage during measurement is not a non-standard measurement condition, proceeding to step S4; S3, inputting the electrochemical impedance spectrum into a pre-trained feature matching model to obtain electrochemical impedance spectrum data matched to standard measurement conditions; S4. Input the electrochemical impedance spectrum under standard measurement conditions into a pre-trained battery capacity estimation model to obtain an estimated value of the battery capacity.

[0008] Preferably, the method for obtaining and training the data set of the pre-trained feature matching model in step S3 includes the following steps: S31, taking a number of batteries of the same model to be recycled, and measuring the electrochemical impedance spectra of the batteries under the conditions of different ambient temperatures and battery terminal voltages at a frequency set f; S32. Using the electrochemical impedance spectrum, ambient temperature and battery terminal voltage data under non-standard measurement conditions as input and the electrochemical impedance spectrum under standard measurement conditions as output, train an electrochemical impedance spectrum feature matching model based on Transformer Encoder.

[0009] Preferably, the battery capacity estimation model pre-trained in step S1 includes but is not limited to XGBoost, LightGBM and Gaussian process regression model.

[0010] Preferably, the data set acquisition and training method of the pre-trained battery capacity estimation model comprises the following steps: S11. Take a number of batteries of the same model to be recycled, measure the electrochemical impedance spectrum of the batteries under the frequency set f under standard measurement conditions, and obtain the battery capacity calibration value under the measurement conditions specified in the product specification; S12. Use the battery electrochemical impedance spectrum under standard measurement conditions as input and the battery capacity calibration value as output to train the battery capacity estimation model Compared with the prior art, the present invention has the following beneficial effects: the method for rapid battery capacity estimation based on electrochemical impedance spectroscopy proposed in the present invention uses the battery broadband impedance as a feature for estimating battery capacity. Depending on the electrochemical impedance spectroscopy measurement method, the capacity estimation time can be controlled to within 1 minute at the shortest, which has the advantage of extremely short time consumption.

[0011] The method for rapid battery capacity estimation based on electrochemical impedance spectroscopy proposed in the present invention matches the electrochemical impedance spectrum under non-standard measurement conditions to that under standard measurement conditions through a feature matching model, and then estimates the battery capacity. This method can achieve battery capacity estimation under any ambient temperature and battery terminal voltage conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flowchart of a method for quickly estimating battery capacity based on electrochemical impedance spectroscopy according to the present invention; Figure 2 A structural diagram of input and output data of a feature matching model of a battery capacity rapid estimation method based on electrochemical impedance spectroscopy according to the present invention; Figure 3 A graph showing the relationship between the electrochemical impedance spectrum and the capacity of a battery during the capacity decay process of the method for rapid battery capacity estimation based on electrochemical impedance spectroscopy according to the present invention; Figure 4 It is a graph showing the electrochemical impedance spectrum characteristic matching result of a 18650 ternary lithium-ion battery according to the battery capacity rapid estimation method based on electrochemical impedance spectroscopy of the present invention; Figure 5 The figure is a battery capacity estimation result diagram of the method for rapid battery capacity estimation based on electrochemical impedance spectroscopy according to the present invention, which takes feature-matched electrochemical impedance spectroscopy data as input. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example 1

[0014] Reference Figure 1 A method for quickly estimating battery capacity based on electrochemical impedance spectroscopy is provided. The method is used to quickly estimate battery capacity in the residual value assessment link of the battery recycling stage, and specifically comprises the following steps: 1) Obtain the electrochemical impedance spectrum of the battery whose capacity is to be estimated under the frequency set f, as well as the ambient temperature and battery terminal voltage during measurement as characteristic data; 2) If the ambient temperature and battery terminal voltage during measurement are combined into non-standard measurement conditions, the electrochemical impedance spectroscopy data, ambient temperature data, and battery terminal voltage data are input into a pre-trained feature matching model to obtain electrochemical impedance spectroscopy data matched to standard measurement conditions; 3) Input the electrochemical impedance spectrum under standard measurement conditions into the pre-trained battery capacity estimation model to obtain an estimated value of the battery capacity.

[0015] In step 1), the measurement of the electrochemical impedance spectrum is specifically carried out by applying a disturbance current signal to the battery and synchronously sampling the disturbance current signal and the response voltage signal at both ends of the battery; then, the voltage and current signals in the time domain are converted to the frequency domain using a complex frequency domain calculation method such as Fourier transform, wavelet transform or Laplace transform, and the battery impedance is further calculated; in this example, the measurement disturbance signal of the electrochemical impedance spectrum can be a sinusoidal signal of a single frequency or a synthetic signal containing multiple frequency components.

[0016] In step 2), the standard measurement condition is a combination of an ambient temperature of 25°C and a battery terminal voltage corresponding to 50% SOC, and the non-standard measurement condition is a combination of an ambient temperature or a battery terminal voltage that is different from the standard measurement condition.

[0017] In step 2), the pre-trained feature matching model is an electrochemical impedance spectroscopy feature matching model based on the Transformer Encoder model. In this example, electrochemical impedance spectroscopy measurements were performed on multiple 18650 batteries under 20 combinations of ambient temperatures of 0°C, 10°C, 25°C and 40°C and battery terminal voltages corresponding to 10%, 30%, 50%, 70% and 90% SOC, with a frequency range of 50mHz to 5kHz. The electrochemical impedance spectroscopy feature matching model is trained using the electrochemical impedance spectrum, ambient temperature and battery terminal voltage data under non-standard measurement conditions as input and the electrochemical impedance spectrum under standard measurement conditions as output. The input and output data structures of the feature matching model are shown below. Figure 2 As shown, each column of input data is the real part and imaginary part of the battery impedance at frequency fi, as well as the ambient temperature and battery terminal voltage during measurement, and each column of output data is the characteristic matching result of the real part and imaginary part of the battery impedance at frequency fi.

[0018] In step 3), the pre-trained battery capacity estimation model is a multivariate nonlinear regression model. In this example, under standard measurement conditions, electrochemical impedance spectroscopy measurements are performed on multiple 18650 batteries with a frequency range of 50mHz to 5kHz; in addition, the capacity of the tested battery is calibrated according to the charging and discharging current parameters specified in the product specification to obtain the battery capacity calibration value. With the electrochemical impedance spectrum under standard measurement conditions as input and the battery capacity calibration value as output, the battery capacity estimation model is trained based on the XGBoost model.

[0019] The implementation principle of the present invention is as follows: The battery capacity decays during the battery calendar aging or charge-discharge cycle aging process, and the battery's electrochemical impedance also increases. The present invention uses electrochemical impedance measurement equipment to perform AC impedance measurement on a batch of 18650 ternary lithium-ion batteries with a rated capacity of 3.0Ah under standard measurement conditions, and observes that there is a clear correlation between battery impedance and battery capacity. Specifically, the larger the real part of the battery impedance and the larger the radius of the impedance arc, the lower the battery capacity, such as Figure 3 shown.

[0020] In addition, by measuring the electrochemical impedance spectroscopy of 18650 ternary lithium-ion batteries under different ambient temperatures and battery terminal voltages, it can be observed that the ambient temperature and the battery terminal voltage have a nonlinear effect on the electrochemical impedance spectrum of the battery. In general, the battery impedance decreases with the increase of the ambient temperature, and first decreases and then increases with the increase of the battery terminal voltage. Based on the above phenomenon, by collecting the electrochemical impedance spectra of the same model of batteries measured under different ambient temperatures and battery terminal voltages, based on a machine learning model, feature matching of the electrochemical impedance spectra under non-standard measurement conditions to standard measurement conditions can be achieved. In view of the fact that the electrochemical impedance spectrum can exist as a kind of sequence structure data from the frequency domain perspective, the self-attention mechanism of the Transformer Encoder model can better identify the relationship between different elements in the same sequence and between the elements in the input and output sequences. Therefore, based on the Transformer Encoder model, the present invention constructs a feature matching model for the electrochemical impedance spectrum on the electrochemical impedance spectrum data set of 18650 ternary lithium-ion batteries, and achieves a better feature matching effect. The feature matching test results of a battery running under cycle aging conditions at the end of aging are shown in the figure. Figure 4 As shown, Figure 4 (6) is the electrochemical impedance spectrum under standard measurement conditions, and the remaining sub-figures are the electrochemical impedance spectra under non-standard measurement conditions and the corresponding feature matching results. The average absolute percentage error of the impedance modulus is 0.93%, and the average absolute error of the impedance angle is 0.265°. Note Figure 4 The two data points corresponding to the low-frequency impedance in the input data of (10) have serious real part offsets, but the feature matching model can still give a relatively reasonable matching result, which reflects the strong anti-interference ability of the feature matching model in the present invention.

[0021] Finally, more than 4,000 feature-matched electrochemical impedance spectroscopy data were input into the pre-trained battery capacity estimation model to obtain the final capacity estimation result. Figure 5 As shown, the model fitting goodness reaches 0.966, and the average absolute percentage error of the estimation result is 1.49%. The horizontal axis in the figure is the calibrated value of the battery capacity, and the vertical axis is the estimated value of the battery capacity. The lighter the color of the estimated point, the greater the distribution density of the estimated point.

[0022] The number of devices and processing scales described here are used to simplify the description of the present invention, and the application, modification and variation of the present invention will be obvious to those skilled in the art.

[0023] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and implementation modes. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A method for rapid estimation of battery capacity based on electrochemical impedance spectroscopy, characterized in that: The following steps are involved: S1. Obtaining the electrochemical impedance spectrum of the battery whose capacity is to be estimated, as well as the ambient temperature and battery terminal voltage during measurement, as characteristic data; S2, judging whether the combination of the ambient temperature and the battery terminal voltage during measurement is a non-standard measurement condition; if it is judged that the combination of the ambient temperature and the battery terminal voltage during measurement is a non-standard measurement condition, proceeding to the next step; if it is judged that the combination of the ambient temperature and the battery terminal voltage during measurement is not a non-standard measurement condition, proceeding to step S4; S3, inputting the electrochemical impedance spectrum into a pre-trained feature matching model to obtain electrochemical impedance spectrum data matched to standard measurement conditions; S4. Input the electrochemical impedance spectrum under standard measurement conditions into a pre-trained battery capacity estimation model to obtain an estimated value of the battery capacity.

2. A method for rapid estimation of battery capacity based on electrochemical impedance spectroscopy according to claim 1, characterized in that: In the step S1, the electrochemical impedance spectrum of the battery whose capacity is to be estimated is obtained under a frequency set f, and the frequency set f is 51 frequency points evenly distributed in a logarithmic ratio within a range of 0.1 Hz to 10 kHz.

3. A method for rapid estimation of battery capacity based on electrochemical impedance spectroscopy as claimed in claim 2, characterized in that: The specific contents of the measurement of the electrochemical impedance spectroscopy include: Applying a disturbance current signal to the battery, and synchronously sampling the disturbance current signal and a response voltage signal at both ends of the battery; Calculation of electrochemical impedance spectroscopy, wherein the methods for calculating the electrochemical impedance spectroscopy include Fourier transform method, wavelet transform method and Laplace transform method; A measurement disturbance signal of an electrochemical impedance spectroscopy, wherein the measurement disturbance signal comprises a sinusoidal signal of a single frequency and a synthetic signal comprising multiple frequency components.

4. A method for rapid estimation of battery capacity based on electrochemical impedance spectroscopy according to claim 1, characterized in that: The standard measurement conditions in step S2 are that the ambient temperature is 25°C and the battery terminal voltage corresponds to 50% battery state of charge; The non-standard measurement condition satisfies any combination of the following conditions: the ambient temperature is not 25°C or the battery terminal voltage corresponds to a battery SOC state other than 50% SOC.

5. A method for rapid estimation of battery capacity based on electrochemical impedance spectroscopy according to claim 1, characterized in that: The data set acquisition and training method of the pre-trained feature matching model in step S3 includes the following steps: S31, taking a number of batteries of the same model to be recycled, and measuring the electrochemical impedance spectra of the batteries under the conditions of different ambient temperatures and battery terminal voltages at a frequency set f; S32. Using the electrochemical impedance spectrum, ambient temperature and battery terminal voltage data under non-standard measurement conditions as input and the electrochemical impedance spectrum under standard measurement conditions as output, train an electrochemical impedance spectrum feature matching model based on Transformer Encoder.

6. A method for rapid estimation of battery capacity based on electrochemical impedance spectroscopy according to claim 1, characterized in that: The battery capacity estimation model pre-trained in step S1 includes but is not limited to XGBoost, LightGBM and Gaussian process regression model.

7. A method for rapid estimation of battery capacity based on electrochemical impedance spectroscopy as claimed in claim 6, characterized in that: The data set acquisition and training method of the pre-trained battery capacity estimation model includes the following steps: S11. Take a number of batteries of the same model to be recycled, measure the electrochemical impedance spectrum of the batteries under the frequency set f under standard measurement conditions, and obtain the battery capacity calibration value under the measurement conditions specified in the product specification; S12. Using the battery electrochemical impedance spectrum under standard measurement conditions as input and the battery capacity calibration value as output, a battery capacity estimation model is trained.

Images