Battery internal defect detection method, system, medium and equipment

Abstract

The invention discloses a battery internal defect detection method, system, medium and equipment, and the method comprises the steps: preparing battery samples with different defects; performing an electrochemical impedance spectroscopy test on the battery sample under the same temperature environment, and establishing an impedance spectroscopy database of the battery in a normal state and a defect full charge state under a normal temperature working condition; calculating charge transfer impedance, solid electrolyte interfacial film impedance and relaxation time of a corresponding characteristic peak in the impedance spectrum data through the relaxation time distribution curve area, and analyzing the charge transfer impedance, the solid electrolyte interfacial film impedance and the relaxation time of the corresponding characteristic peak; constructing a battery internal defect identification model to accurately identify whether the battery has internal defects or not; performing relaxation time distribution quantification at a constant temperature to obtain charge transfer impedance, solid electrolyte interfacial film impedance and relaxation time of a corresponding characteristic peak; and importing the charge transfer impedance, the solid electrolyte interfacial film impedance and the relaxation time of the corresponding characteristic peak into the battery internal defect identification model to obtain a diagnosis result of whether the defect exists in the single battery or not.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium battery status detection, and in particular to a battery internal defect detection method, system, medium and equipment. Background Art

[0002] In practical applications, the safety and reliability of lithium batteries are crucial. Internal defects, such as airtightness defects and uneven distribution of conductive coatings within the battery, are difficult to detect in the early stages but can lead to poor battery performance. When stacked in series and parallel to form a battery module, these defects can degrade the overall module's balance and reduce battery utilization.

[0003] Among traditional detection methods, commonly used methods include measuring battery voltage and current using Ohm's law, DC internal resistance testing, etc. These methods will cause certain damage to battery performance due to the use of large current or high voltage for testing. At the same time, the detection results are limited by low sensitivity and dependence on a single parameter, making it difficult to accurately identify microscopic defects.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The present invention provides a battery internal defect detection method, system, medium and equipment, which utilize the extraction of the internal characteristic impedance of the battery to track and detect internal defects of the lithium battery after each cycle.

[0006] A method for detecting internal defects of a battery includes:

[0007] S1: Prepare battery samples with different defects, including normal batteries, aged batteries, batteries with airtightness defects, and batteries with uneven conductive coating areas within the battery;

[0008] S2: Conduct electrochemical impedance spectroscopy tests on battery samples under the same temperature environment to establish an impedance spectrum database of normal and defective fully charged batteries under room temperature conditions;

[0009] S3: Calculate the charge transfer impedance, solid electrolyte interface film impedance and relaxation time of the corresponding characteristic peaks in the impedance spectrum data by using the area of ​​the relaxation time distribution curve;

[0010] S4: Using charge transfer impedance, solid electrolyte interface film impedance, and relaxation time of corresponding characteristic peaks as criteria for judging battery defects, a support vector machine classification model is constructed to accurately identify whether the battery has internal defects;

[0011] S5: At a constant temperature, after each charge-discharge cycle, perform impedance spectrum measurement on the single cell to be tested, and quantify the relaxation time distribution to obtain the charge transfer impedance, solid electrolyte interface film impedance, and relaxation time of the corresponding characteristic peak;

[0012] S6: The charge transfer impedance, the solid electrolyte interface film impedance and the relaxation time of the corresponding characteristic peaks are introduced into the battery internal defect identification model to obtain a diagnosis result of whether there is a defect inside the single cell.

[0013] In the battery internal defect detection method, in step S3, the relaxation time distribution (DRT) is used to convert the electrochemical impedance spectrum in the frequency domain into time domain characteristic data to realize the internal charge transfer impedance R of the lithium ion battery. ct and solid electrolyte interface film resistance R SEI Quantitative separation of the battery charge transfer impedance R by analyzing the characteristic peak integral area of ​​the relaxation time distribution function ct and solid electrolyte interface film resistance R SEI The relaxation time distribution function converts the electrochemical impedance spectroscopy data into an expression of multiple differential resistances and capacitances in parallel by using the equivalent circuit method. The formula is as follows:

[0014]

[0015] Where Z(ω) is the total impedance of the battery; n is the number of parallel elements of differential resistance and capacitance; dR i is the ith differential resistance; j is the imaginary unit; ω is the angular frequency; C i is the i-th capacitor; dτ i is the discrete relaxation time differential expression, which is:

[0016] Converting the above differential expression into integral form gives the relaxation time distribution function, which is as follows:

[0017]

[0018] Where Z(ω) is the total impedance of the battery, R0 is the frequency-independent ohmic impedance within the battery, γ(τ) is the relaxation time distribution function, τ is the relaxation time, j is the imaginary unit, and ω is the angular frequency.

[0019] In the battery internal defect detection method, building a battery internal defect recognition model includes:

[0020] Step S401: Collect the battery charge transfer impedance R ct , solid electrolyte interface film impedance R SEIand the relaxation time τ of the corresponding characteristic peak, determine the input feature vector and dimension to form a data set;

[0021] Step S402: Arrange the data set, determine the proportion of the training set to the data set, and shuffle the data set; loop through samples of different categories to achieve data set division and obtain a training set and a test set;

[0022] Step S403: Constructing a support vector machine model, which includes: determining the model kernel function and the model soft margin. ct , solid electrolyte interface film impedance R SEI The model is fed with the relaxation time τ of the corresponding characteristic peak and outputs the corresponding type of internal battery defect. The trained model is validated using a test set. If the test set accuracy is above 90%, the model classification performance is considered to have met expectations. Otherwise, the model hyperparameters are fine-tuned until the classification accuracy meets the requirements.

[0023] In the battery internal defect detection method, in step S403, the model kernel function is a Gaussian kernel function.

[0024]

[0025] Among them, x i 、x j is the battery impedance spectrum data, and γ is the parameter of the Gaussian kernel function.

[0026] In the battery internal defect detection method, in step S403, the objective optimization function of the model soft margin is:

[0027]

[0028] Among them, ω is the normal vector of the hyperplane in the support vector machine model, ε∈R n , C is the penalty parameter, ε i is the relaxation parameter, satisfying ε i ≥0, the objective optimization function needs to maximize the interval and minimize ε i Part of the penalty term is used to achieve a trade-off between the two.

[0029] For the objective optimization function of the model soft margin, the constraints are as follows:

[0030]

[0031] Among them, y i is the index of each training data, ω T To determine the normal vector perpendicular to the hyperplane, x i is the battery impedance spectrum data, b is the distance between the hyperplane and the source point, εi is the relaxation parameter, satisfying ε i ≥0.

[0032] In the battery internal defect detection method, in step S403, the hyperparameter tuning includes the penalty parameter C and the parameter γ of the Gaussian kernel function.

[0033] In the battery internal defect detection method, in step S5, the single lithium battery is in a fully charged state during the detection process.

[0034] A system for implementing the method includes:

[0035] A measuring unit for performing electrochemical impedance spectroscopy testing of a single lithium battery;

[0036] A data unit is used to perform quantitative analysis of the relaxation time distribution of the electrochemical impedance spectroscopy data of batteries with different defect conditions, extract the charge transfer impedance, solid electrolyte interface film impedance and relaxation time data of the corresponding characteristic peaks to establish a database;

[0037] A model unit, which uses the impedance characteristics established by the data unit as characteristic parameters to build a battery internal defect recognition model;

[0038] The detection unit is used to perform electrochemical impedance spectroscopy testing on the single battery to be tested after each cycle, and substitute the quantified extracted data into the established battery internal defect recognition model to obtain the battery internal defect situation.

[0039] A computer storage medium includes computer instructions, which, when executed on a computer, cause the computer to execute the method described above.

[0040] An electronic device, comprising:

[0041] A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein:

[0042] When the processor executes the program, the method described is implemented.

[0043] Compared with the existing technology, the present invention has the following advantages: the present invention solves the difficulty of identifying internal defects of the battery through impedance characteristic modeling and DRT multi-parameter analysis technology. Compared with the traditional method that only relies on total impedance or a single parameter, the present method simultaneously monitors the multiple changes of SEI film internal resistance, charge transfer internal resistance and relaxation time, and maps the internal defects of the battery in multiple dimensions, with high sensitivity, strong environmental adaptability and engineering feasibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Various other advantages and benefits of the present invention will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are intended only to illustrate preferred embodiments and are not to be construed as limiting the present invention. It should be understood that the drawings described below are merely examples of the present invention, and that those skilled in the art will be able to derive other drawings from these drawings without inventive effort. Throughout the drawings, identical reference numerals are used to denote identical components.

[0045] In the attached figure:

[0046] Figure 1 This is a flow chart of a method for detecting internal defects of a battery based on electrochemical impedance spectroscopy, provided in one embodiment of the present disclosure;

[0047] Figure 2 This is a schematic diagram of electrochemical impedance spectroscopy images of lithium-ion batteries under different defect conditions provided by another embodiment of the present disclosure. It shows electrochemical impedance spectroscopy test results for four defect conditions: normal battery, aging, uneven conductive material coating, and battery airtightness. Different internal battery defects have obvious differences in impedance characteristics, and internal battery defects can be identified through impedance parameters.

[0048] Figure 3 This is a graph showing the calculated relaxation time distribution results of lithium-ion batteries under different defect conditions, provided by another embodiment of the present disclosure. The graph shows changes in battery relaxation time distribution under four defect conditions. The four defective battery samples have differences in height and position on the two characteristic peaks on the right. The corresponding characteristic impedance parameters are obtained by calculating the area enclosed by the characteristic peaks and the abscissa. The abscissa of the corresponding characteristic peak is the corresponding relaxation time.

[0049] Figure 4 This is a classification result diagram of a battery internal defect recognition model constructed based on feature parameters provided by another embodiment of the present disclosure. In the test set classification result diagram of the battery internal defect recognition model, the vertical axis is the actual sample type, and the horizontal axis is the model classification result. The red part is the correctly classified sample, and the blue part is the incorrectly classified sample. The statistically correct classification results account for 93.1% of the total samples.

[0050] The present invention will be further explained below with reference to the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0051] Specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although specific embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0052] It should be noted that certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that technicians may use different nouns to refer to the same component. This specification and claims do not use the difference in nouns as a way to distinguish components, but use the difference in the functions of the components as the criterion for distinction. As mentioned throughout the specification and claims, "including" or "comprising" is an open term, so it should be interpreted as "including but not limited to". The subsequent description of the specification is a preferred embodiment of the present invention, but the description is based on the general principles of the specification and is not intended to limit the scope of the invention. The scope of protection of the present invention shall be as defined in the attached claims.

[0053] To facilitate understanding of the embodiments of the present invention, further explanation will be given below using specific embodiments as examples in conjunction with the accompanying drawings, and the accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0054] like Figures 1 to 4 As shown, the battery internal defect detection method includes the following steps:

[0055] S1: Prepare battery samples with different defects, including normal batteries, aged batteries, batteries with airtightness defects, and batteries with uneven conductive coating areas within the battery;

[0056] S2: Conduct electrochemical impedance spectroscopy tests on battery samples under the same temperature environment to establish an impedance spectrum database of normal and defective fully charged batteries under room temperature conditions;

[0057] S3: Calculate the charge transfer impedance, solid electrolyte interface film impedance and relaxation time of the corresponding characteristic peaks in the impedance spectrum data by using the area of ​​the relaxation time distribution curve;

[0058] S4: Using charge transfer impedance, solid electrolyte interface film impedance, and relaxation time of corresponding characteristic peaks as criteria for judging battery defects, a support vector machine classification model is constructed to accurately identify whether the battery has internal defects;

[0059] S5: At a constant temperature, after each charge-discharge cycle, perform impedance spectrum measurement on the single cell to be tested, and quantify the relaxation time distribution to obtain the charge transfer impedance, solid electrolyte interface film impedance, and relaxation time of the corresponding characteristic peak;

[0060] S6: The charge transfer impedance, the solid electrolyte interface film impedance and the relaxation time of the corresponding characteristic peaks are introduced into the battery internal defect identification model to obtain a diagnosis result of whether there is a defect inside the single cell.

[0061] In a preferred embodiment of the battery internal defect detection method, in step S3, the relaxation time distribution (DRT) is used to convert the electrochemical impedance spectrum in the frequency domain into time domain characteristic data to realize the internal charge transfer impedance R of the lithium ion battery. ct and solid electrolyte interface film resistance R SEI Quantitative separation of the battery charge transfer impedance R by analyzing the characteristic peak integral area of ​​the relaxation time distribution function ct and solid electrolyte interface film resistance R SEI The relaxation time distribution function converts the electrochemical impedance spectroscopy data into an expression of multiple differential resistances and capacitances in parallel by using the equivalent circuit method. The formula is as follows:

[0062]

[0063] Where Z(ω) is the total impedance of the battery; n is the number of parallel elements of differential resistance and capacitance; dR i is the ith differential resistance; j is the imaginary unit; ω is the angular frequency; C i is the i-th capacitor; dτ i is the discrete relaxation time differential expression, which is:

[0064] Converting the above differential expression into integral form gives the relaxation time distribution function, which is as follows:

[0065]

[0066] Where Z(ω) is the total impedance of the battery, R0 is the frequency-independent ohmic impedance within the battery, γ(τ) is the relaxation time distribution function, τ is the relaxation time, j is the imaginary unit, and ω is the angular frequency.

[0067] In a preferred embodiment of the battery internal defect detection method, building a battery internal defect recognition model includes:

[0068] Step S401: Collect the battery charge transfer impedance R ct , solid electrolyte interface film impedance R SEI and the relaxation time τ of the corresponding characteristic peak, determine the input feature vector and dimension to form a data set;

[0069] Step S402: Arrange the data set, determine the proportion of the training set to the data set, and shuffle the data set; loop through samples of different categories to achieve data set division and obtain a training set and a test set;

[0070] Step S403: Constructing a support vector machine model, which includes: determining the model kernel function and the model soft margin. ct , solid electrolyte interface film impedance R SEI The model is fed with the relaxation time τ of the corresponding characteristic peak and outputs the corresponding type of internal battery defect. The trained model is validated using a test set. If the test set accuracy is above 90%, the model classification performance is considered to have met expectations. Otherwise, the model hyperparameters are fine-tuned until the classification accuracy meets the requirements.

[0071] In a preferred embodiment of the battery internal defect detection method, in step S403, the model kernel function is a Gaussian kernel function.

[0072]

[0073] Among them, x i 、x j is the battery impedance spectrum data, and γ is the parameter of the Gaussian kernel function.

[0074] In a preferred embodiment of the battery internal defect detection method, in step S403, the objective optimization function of the model soft margin is:

[0075]

[0076] Among them, ω is the normal vector of the hyperplane in the support vector machine model, ε∈R n , C is the penalty parameter, ε i is the relaxation parameter, satisfying ε i ≥0, the objective optimization function needs to maximize the interval and minimize ε i Part of the penalty term is used to achieve a trade-off between the two.

[0077] For the objective optimization function of the model soft margin, the constraints are as follows:

[0078]

[0079] Among them, y i is the index of each training data, ω T To determine the normal vector perpendicular to the hyperplane, x i is the battery impedance spectrum data, b is the distance between the hyperplane and the source point, ε i is the relaxation parameter, satisfying ε i ≥0.

[0080] In a preferred embodiment of the battery internal defect detection method, in step S403, the hyperparameter tuning includes the penalty parameter C and the parameter γ of the Gaussian kernel function.

[0081] In a preferred embodiment of the battery internal defect detection method, in step S5, the single lithium battery is in a fully charged state during the detection process.

[0082] A system for implementing the method includes:

[0083] A measuring unit for performing electrochemical impedance spectroscopy testing of a single lithium battery;

[0084] A data unit is used to perform quantitative analysis of the relaxation time distribution of the electrochemical impedance spectroscopy data of batteries with different defect conditions, extract the charge transfer impedance, solid electrolyte interface film impedance and relaxation time data of the corresponding characteristic peaks to establish a database;

[0085] A model unit, which uses the impedance characteristics established by the data unit as characteristic parameters to build a battery internal defect recognition model;

[0086] The detection unit is used to perform electrochemical impedance spectroscopy testing on the single battery to be tested after each cycle, and substitute the quantified extracted data into the established battery internal defect recognition model to obtain the battery internal defect situation.

[0087] A computer storage medium includes computer instructions, which, when executed on a computer, cause the computer to execute the method described above.

[0088] An electronic device, comprising:

[0089] A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein:

[0090] When the processor executes the program, the method described is implemented.

[0091] In one embodiment, preferably, in step S2, the normal ambient temperature is 25°C.

[0092] In one embodiment, the method steps include:

[0093] Prepare battery samples with different defect conditions, including normal batteries, aged batteries, batteries with airtight defects, and battery samples with uneven conductive coating areas inside the battery; conduct electrochemical impedance spectroscopy tests on single battery samples under the same temperature environment, and establish an impedance spectrum database for normal and defective fully charged batteries under room temperature conditions; calculate the charge transfer internal resistance, SEI internal resistance value, and relaxation time in the impedance spectrum data by using the area of ​​the relaxation time distribution curve for analysis; use the charge transfer resistance, SEI internal resistance value, and relaxation time as criteria for judging the battery defect condition, and construct a battery internal defect recognition model to accurately identify whether the battery has internal defects; at a constant temperature, after each charge and discharge cycle, perform impedance spectrum measurements on the single battery to be tested, and quantify the relaxation time distribution to obtain the charge transfer internal resistance, SEI internal resistance value, and relaxation time; import the charge transfer internal resistance, SEI internal resistance value, and relaxation time into the battery internal defect recognition model to obtain a diagnosis result on whether the single battery has defects. Based on the differences in impedance characteristics between battery cells in normal state and those with internal defects, the present invention proposes a battery internal defect detection method based on electrochemical impedance spectroscopy. By tracking and detecting during the battery cycle, it is possible to quickly and economically detect whether there are defects inside the single cell.

[0094] In one embodiment, in step S3, the relaxation time distribution method (DRT) is an impedance spectrum analysis method based on mathematical inversion technology. Its core principle is to convert the frequency domain impedance response into the time domain relaxation time distribution function. This method converts the frequency domain electrochemical impedance spectrum into time domain characteristic data to achieve the internal charge transfer impedance (R ct ) and solid electrolyte interface (SEI) impedance (R SEI Specifically, under the DRT analysis framework, the electrochemical polarization process of the battery system can be decomposed into several elementary responses with characteristic relaxation times, where the contribution of each relaxation process is characterized by the integral area of ​​the distribution function. By analyzing the evolution of the integral area of ​​the characteristic peak of the DRT distribution function with the number of cycles, the R ct and R SEI The impedance increase.

[0095] like Figure 1 As shown, a method for detecting internal defects of a battery based on electrochemical impedance spectroscopy includes the following steps:

[0096] S1: Prepare battery samples with different defect conditions, including normal batteries, aged batteries, batteries with airtightness defects, and batteries with uneven conductive coating areas within the battery;

[0097] S2: Conduct electrochemical impedance spectroscopy (EIS) tests on single-cell battery samples under the same temperature environment to establish an impedance spectrum database for normal and defective fully charged batteries under room temperature conditions.

[0098] S3: Calculate the charge transfer internal resistance, SEI internal resistance and relaxation time in the impedance spectrum data by using the area of ​​the relaxation time distribution curve;

[0099] S4: Using charge transfer resistance, SEI internal resistance and relaxation time as criteria for judging battery defects, a battery internal defect recognition model is constructed to accurately identify whether the battery has internal defects;

[0100] S5: At a constant temperature, after each charge and discharge cycle, perform impedance spectrum measurement on the single cell to be tested, and quantify the relaxation time distribution to obtain the charge transfer internal resistance, SEI internal resistance and relaxation time;

[0101] S6: The charge transfer internal resistance, SEI internal resistance and relaxation time are introduced into the battery internal defect identification model to obtain a diagnosis result of whether there is a defect inside the single cell.

[0102] The above embodiments constitute a complete technical solution of the present disclosure. The method described in this embodiment solves the difficulty of identifying internal battery defects through impedance characteristic modeling and DRT multi-parameter analysis technology. Compared with traditional methods that only rely on total impedance or a single parameter, this embodiment simultaneously monitors multiple changes in SEI film internal resistance, charge transfer internal resistance, and relaxation time, mapping the internal battery defects in multiple dimensions, with high sensitivity, strong environmental adaptability, and engineering feasibility.

[0103] In another embodiment, the comparative analysis of the impedance spectra of lithium batteries with different defect conditions is as follows: Figure 2 As shown in the figure, there are mainly four types of battery samples, namely: normal battery sample, aged battery sample, battery airtightness defect sample and battery material area uneven coating sample. All four battery samples are controlled in the full charge state for impedance spectrum testing. Figure 2 The results show that there are significant differences in the electrochemical impedance spectroscopy test results for different battery internal defects. For battery samples with the same condition, the overall curve changes are nearly consistent between the two, but there are significant differences in the amplitude of the change in the semicircular arc on the right side of the curve between different samples. Therefore, by considering the impedance information corresponding to the semicircular portion of the impedance spectrum, a mapping relationship between impedance characteristic parameters and internal battery defects can be established, enabling the identification of internal battery defects.

[0104] In another embodiment, the relaxation time distribution of the single cell under different defect conditions is transformed. The relaxation time distribution method (DRT) is an impedance spectrum analysis method based on mathematical inversion technology. Its core principle is to convert the frequency domain impedance response into a time domain relaxation time distribution function. This method converts the frequency domain electrochemical impedance spectrum into time domain characteristic data to achieve the internal charge transfer impedance (R ct ) and solid electrolyte interface film impedance (R SEI ) for quantitative separation.

[0105] The calculation method of the relaxation time distribution function is: using the equivalent circuit method to convert the electrochemical impedance spectroscopy data into an expression of multiple differential resistances and capacitances in parallel, the formula is as follows:

[0106]

[0107] Where Z(ω) is the total impedance of the battery; n is the number of parallel elements of differential resistance and capacitance; dR i is the ith differential resistance; j is the imaginary unit; ω is the angular frequency; C i is the i-th capacitor; dτ i is the discrete relaxation time differential expression, which is:

[0108] Converting the above differential expression into integral form gives the relaxation time distribution function, which is as follows:

[0109]

[0110] Where Z(ω) is the total impedance of the battery, R0 is the frequency-independent ohmic impedance within the battery, γ(τ) is the relaxation time distribution function, τ is the relaxation time, j is the imaginary unit, and ω is the angular frequency.

[0111] Under the DRT analysis framework, the impedance spectrum data of battery samples with four defect conditions are converted into relaxation time distribution data as shown below: Figure 3 As shown. The impedance values ​​of the charge transfer resistance and the SEI film internal resistance correspond to the two characteristic peaks in the relaxation time distribution diagram within 0.1-10s (i.e., the battery impedance spectrum is 0.1-10Hz). The area of ​​the left characteristic peak is the SEI internal resistance value, and the right is the charge transfer resistance value. The horizontal axis coordinates corresponding to the peak values ​​of the two characteristic peaks are the relaxation times τ1 and τ2. Figure 3 The corresponding characteristic parameters are shown in the following table:

[0112] Table 1 Battery characteristic parameters under different defect conditions

[0113]

[0114] In another embodiment, a battery internal defect recognition model was constructed. Multiple electrochemical impedance spectroscopy tests were performed on the above battery samples, ensuring that each test was conducted at a constant temperature of 25°C and the battery samples were fully charged. The electrochemical impedance spectroscopy was inverted using the relaxation time distribution method, and the characteristic parameters of the different defect samples in the above embodiment were extracted. A characteristic parameter database was constructed, totaling 190 sets of battery data samples. The data was divided into training and test sets in a ratio of 7:3, and a support vector machine classification training model was constructed.

[0115] The Support Vector Machine (SVM) is a binary generalized linear classifier based on a supervised learning framework. Its core mechanism is to construct a decision hyperplane that maximizes the interval between training samples. Compared to algorithms such as logistic regression and neural networks, this model exhibits more outstanding algorithmic properties when solving nonlinear classification problems: for linearly inseparable sample sets, by introducing slack variables and using kernel mapping techniques, samples in the low-dimensional original input space can be nonlinearly transformed into a high-dimensional feature space, achieving linear separability, and then solving the optimal classification hyperplane in this reproducing kernel Hilbert space. Its mathematical modeling process can be formally expressed as the following optimization problem:

[0116] 1. Confirmation of kernel function. Nonlinear data is usually not separable, which requires the use of kernel functions to map the original data to a high-dimensional space, so that the data can be separated in the high-dimensional space. At the same time, the kernel function can simplify the dot product operation in the high-dimensional space into a low-dimensional space kernel function calculation, greatly reducing the complexity of the model. In this embodiment, the model kernel function adopts the Gaussian kernel function, which is as follows:

[0117]

[0118] Among them, x i 、x j is the battery impedance spectrum data, and γ is the parameter of the Gaussian kernel function.

[0119] 2. Soft margin design. To handle noise or outliers, slack variables and penalty parameters must be introduced. The optimization objective function is:

[0120]

[0121] Among them, ω is the normal vector of the hyperplane in the support vector machine model, ε∈R n , C is the penalty parameter, ε i is the relaxation parameter, satisfying ε i ≥ 0. The objective function needs to maximize the interval and minimize ε i Part of the penalty term is used to achieve a trade-off between the two.

[0122] For the soft margin optimization objective function, the constraints are as follows:

[0123]

[0124] Among them, y i is the index of each training data, ω T To determine the normal vector perpendicular to the hyperplane, x i is the battery impedance spectrum data, b is the distance between the hyperplane and the source point, ε i is the relaxation parameter, satisfying ε i ≥0.

[0125] By adjusting the penalty parameter C and the kernel function parameter γ, the model is tuned. The penalty parameter C is finally determined to be 1 and the kernel function parameter γ is determined to be 0.1. The classification results of the test set after model training are as follows Figure 4 As shown in the figure, 54 of the 58 test sets were correctly classified, for an accuracy rate of 93.1%. Based on this, by testing the electrochemical impedance spectroscopy of battery samples and performing relaxation time distribution conversion, the required characteristic parameters are extracted and introduced into the battery internal defect recognition model, allowing for rapid identification of internal defects in battery samples.

[0126] In another embodiment, the present disclosure further provides a method for detecting internal defects of a battery based on electrochemical impedance spectroscopy, comprising:

[0127] Measuring unit, used for electrochemical impedance spectroscopy testing of single lithium batteries;

[0128] The data unit is used to quantitatively analyze the relaxation time distribution of the electrochemical impedance spectroscopy data of batteries with different defect conditions, extract the charge transfer internal resistance, SEI internal resistance and relaxation time data to establish a database;

[0129] The model unit uses the impedance characteristics established by the data unit as characteristic parameters to build a battery internal defect recognition model;

[0130] The detection unit is used to perform electrochemical impedance spectroscopy testing on the single battery to be tested after each cycle, and substitute the quantified extracted data into the established battery internal defect recognition model to obtain the internal defect situation of the battery.

[0131] The present invention prepares battery samples with different defect types, including normal batteries, aged batteries, batteries with airtight defects, and battery samples with uneven conductive coating areas. Constructing a diverse database: By introducing multiple typical defect types, model training and testing are ensured to be representative, improving the recognition model's ability to distinguish between different types of defects. Enhancing diagnostic universality: Covering the main defect types that may occur during actual production and use, improving the adaptability and reliability of the detection method in industrial applications. Electrochemical impedance spectroscopy (EIS) testing is performed on various types of batteries in a fully charged state at a constant temperature (e.g., 25°C) to establish an impedance spectrum database. Non-destructive testing: EIS is a non-destructive, rapid, and sensitive electrochemical analysis method that can reflect the internal microstructure and interfacial reaction characteristics of batteries. Feature extraction basis: Providing raw frequency domain data for subsequent DRT analysis, it is the basis for quantifying key parameters such as SEI film internal resistance and charge transfer resistance. Unifying the testing environment (temperature, SOC) eliminates external interference factors and ensures data consistency and comparability.

[0132] Distribution of relaxation time (DRT) analysis inverts EIS frequency-domain data into a time-domain relaxation time distribution function, separating Rct (charge transfer resistance), RSEI (SEI film internal resistance), and the corresponding relaxation time τ. Multi-parameter quantitative analysis: This breaks through the limitations of traditional EIS, which relies solely on total impedance or semicircular arc diameter, and enables quantitative characterization of multiple independent electrochemical processes (such as SEI formation and charge transfer). High-sensitivity defect identification: Interface changes caused by different defects (such as SEI thickening and poor electrode contact) will produce significant differences in Rct, RSEI, and τ, thereby enabling accurate identification of defect types. Dynamic monitoring basis: By tracking the changing trends of Rct and RSEI with the number of cycles, it can be used to assess the degree of battery aging and the evolution of defects.

[0133] A support vector machine (SVM) classification model is constructed, using the following input features: Rct, RSEI, and τ. A Gaussian kernel function (RBF) is used, along with soft margin optimization and fine-tuning of the penalty parameter C and kernel parameter γ. High-dimensional mapping and nonlinear classification: The RBF kernel is used to map low-dimensional inputs to a high-dimensional space, resolving the linear inseparability of the original data and improving classification accuracy. Robustness: The soft margin mechanism allows for partial misclassification, preventing overfitting and improving the model's stability in the presence of noisy data. Automatic defect classification: Based on extracted electrochemical features, automated identification of defect types such as "normal / abnormal," "aging / airtightness defects," and "uneven coating" is achieved. Parameter tuning improves performance: The C and γ parameters are appropriately selected to achieve an optimal balance between model generalization and classification accuracy. Post-cycle impedance measurement and online testing are performed, with EIS measurements performed after each cycle. Rct, RSEI, and τ are extracted and fed into the SVM model for diagnosis. Real-time health assessment: Through periodic testing, continuous monitoring of internal defect evolution throughout the battery's lifecycle is achieved. Early warning function: Even at the early stages of defects (such as the slow growth phase of the SEI), parameter changes can be used to promptly identify potential problems. Closed-loop quality control: Suitable for production line quality inspection or in-service health management, improving battery system safety and reliability. Database and model training: A dataset containing 190 samples is established; the training and test sets are divided into a 7:3 ratio; and the SVM model is trained and validated. Support for intelligent diagnosis systems: This provides high-quality training data and mature algorithm models for the subsequent development of an AI-based battery defect detection platform. Strong model transferability: The proposed method is independent of specific battery models and is applicable across systems and processes. High classification accuracy: Experimental results demonstrate a 93.1% accuracy rate on the test set, validating the method's effectiveness in practical applications. Modular design facilitates integration: Each unit has clear functions, making it suitable for integration into battery management systems (BMS) or smart manufacturing testing equipment. Fully automated testing: From data acquisition to final diagnostic output, this process achieves unmanned automated testing. High feasibility: Deployment is simple using standard EIS instruments and a general-purpose computing platform, eliminating the need for specialized hardware and keeping costs manageable.

[0134] Although the embodiments of the present invention have been described above with reference to the accompanying drawings, the present invention is not limited to the above-mentioned specific embodiments and application fields. The above-mentioned specific embodiments are merely illustrative and instructive, and are not restrictive. A person skilled in the art, guided by this specification and without departing from the scope of protection of the claims of the present invention, may also devise various forms, all of which fall within the scope of protection of the present invention.

Claims

1. A method for detecting internal defects of a battery, characterized in that: The steps include: S1: Prepare battery samples with different defects, including normal batteries, aged batteries, batteries with airtightness defects, and batteries with uneven conductive coating areas within the battery; S2: Conduct electrochemical impedance spectroscopy tests on battery samples under the same temperature environment to establish an impedance spectrum database of normal and defective fully charged batteries under room temperature conditions; S3: Calculate the charge transfer impedance, solid electrolyte interface film impedance and relaxation time of the corresponding characteristic peaks in the impedance spectrum data by using the area of ​​the relaxation time distribution curve; S4: Using charge transfer impedance, solid electrolyte interface film impedance, and relaxation time of corresponding characteristic peaks as criteria for judging battery defects, a support vector machine classification model is constructed to accurately identify whether the battery has internal defects; S5: At a constant temperature, after each charge-discharge cycle, perform impedance spectrum measurement on the single cell to be tested, and quantify the relaxation time distribution to obtain the charge transfer impedance, solid electrolyte interface film impedance, and relaxation time of the corresponding characteristic peak; S6: The charge transfer impedance, the solid electrolyte interface film impedance and the relaxation time of the corresponding characteristic peaks are introduced into the battery internal defect identification model to obtain a diagnosis result of whether there is a defect inside the single cell.

2. A battery internal defect detection method according to claim 1, characterized in that: Preferably, in step S3, the relaxation time distribution DRT converts the electrochemical impedance spectrum in the frequency domain into time domain characteristic data to realize the internal charge transfer impedance R ct and solid electrolyte interface film resistance R SEI Quantitative separation of the battery charge transfer impedance R by analyzing the characteristic peak integral area of ​​the relaxation time distribution function ct and solid electrolyte interface film resistance R SEI The relaxation time distribution function converts the electrochemical impedance spectroscopy data into an expression of multiple differential resistances and capacitances in parallel by using the equivalent circuit method. The formula is as follows: ， Where Z(ω) is the total impedance of the battery; n is the number of parallel elements of differential resistance and capacitance; dR i is the ith differential resistance; j is the imaginary unit; ω is the angular frequency; C i is the i-th capacitor; dτ i is the discrete relaxation time differential expression, which is: , Converting the above differential expression into integral form gives the relaxation time distribution function, which is as follows: ， Where Z(ω) is the total impedance of the battery, R0 is the frequency-independent ohmic impedance in the battery, γ(τ) is the relaxation time distribution function, τ is the relaxation time, and j is the imaginary unit.

3. A battery internal defect detection method according to claim 2, characterized in that: Building a battery internal defect recognition model includes: Step S401: Collect the battery charge transfer impedance R ct , solid electrolyte interface film impedance R SEI and the relaxation time τ of the corresponding characteristic peak, determine the input feature vector and dimension to form a data set; Step S402: Arrange the data set, determine the proportion of the training set to the data set, and shuffle the data set; Loop out samples of different categories to divide the data set into training set and test set; Step S403: constructing a support vector machine model, which includes: determining the model kernel function, the model soft interval, and converting the charge transfer impedance R ct , solid electrolyte interface film impedance R SEI The relaxation time τ of the corresponding characteristic peak is input into the model, and the corresponding type of internal battery defects is output. The trained model is verified using a test set. If the accuracy of the test set results is above 90%, it is considered that the model classification effect meets the expectations. Otherwise, the model hyperparameters are tuned until the classification accuracy meets the requirements.

4. A battery internal defect detection method according to claim 3, characterized in that: In step S403, the model kernel function is a Gaussian kernel function. ， Among them, x i 、x j is the battery impedance spectrum data, and γ is the parameter of the Gaussian kernel function.

5. A battery internal defect detection method according to claim 4, characterized in that: In step S403, the objective optimization function of the model soft margin is: ， Among them, ω is the normal vector of the hyperplane in the support vector machine model, ε∈R n , C is the penalty parameter, ε i is the relaxation parameter, satisfying ε i ≥0, the objective optimization function needs to maximize the interval and minimize ε i Part of the penalty term is used to achieve a trade-off between the two. For the objective optimization function of the model soft margin, the constraints are as follows: ， Among them, y i is the index of each training data, ω T To determine the normal vector perpendicular to the hyperplane, x i is the battery impedance spectrum data, b is the distance between the hyperplane and the source point, ε i is the relaxation parameter, satisfying ε i ≥0.

6. A battery internal defect detection method according to claim 5, characterized in that: In step S403 , the hyperparameter tuning includes the penalty parameter C and the parameter γ of the Gaussian kernel function.

7. The method for detecting internal defects of a battery according to claim 1, wherein: In step S5, the single lithium battery is in a fully charged state during the detection process.

8. A system for implementing the method according to any one of claims 1 to 7, characterized in that: It includes: A measuring unit for performing electrochemical impedance spectroscopy testing of a single lithium battery; A data unit is used to perform quantitative analysis of the relaxation time distribution of the electrochemical impedance spectroscopy data of batteries with different defect conditions, extract the charge transfer impedance, solid electrolyte interface film impedance and relaxation time data of the corresponding characteristic peaks to establish a database; A model unit, which uses the impedance characteristics established by the data unit as characteristic parameters to build a battery internal defect recognition model; The detection unit is used to perform electrochemical impedance spectroscopy testing on the single battery to be tested after each cycle, and substitute the quantified extracted data into the established battery internal defect recognition model to obtain the battery internal defect situation.

9. A computer storage medium, characterized in that The storage medium includes computer instructions, which, when executed on a computer, enable the computer to perform the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The electronic device comprises: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

Images