Lithium ion battery service life prediction method based on low frequency electrochemical noise
Patent Information
- Application Number
- CN202311178325.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-13
- Publication Date
- 2026-10-27
- Estimated Expiration
- 2043-09-13
AI Technical Summary
[0003]当前针对锂离子电池寿命的检测和预测应用较为广泛的主要有两种手段：一种是直接测量法，在良好的试验环境下直接通过充放电检测锂离子电池的容量，该测量方法对实验设备和环境都有着较高的要求，而且无法实现对电池退化的预见性；另一种是电化学阻抗谱法，利用锂离子电池电化学阻抗谱的变化来预测锂离子电池的寿命，该方法能够间接表现出电池内部材料在老化过程中的变化，但是该方法的预测结果在电池退化的后期误差较大
[0063]本发明基于电池容量和低频电化学噪声的测量结果，提取噪声频域特征参数，采用多元线性模型对电池的容量退化进行建模分析，能够准确地建立电池噪声频域特征参数与电池容量间地定量关系，并且能够预测电池寿命，其预测结果较其他方法真实可信。本发明能够使用电池内部退化机理直接产生的现象进行建模分析，能够用于电池寿命预测的分析和计算，为在指导电池的合理选择与使用，提高电池的使用可靠性及优化提供了技术支撑。
Smart Images
Figure CN116973768B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of lithium-ion battery life prediction and low-frequency electrochemical noise analysis technology, and specifically relates to a method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise. Background Technology
[0002] Lithium-ion batteries are widely used in mobile phones, automobiles, aerospace, industrial manufacturing, energy development, and other fields. Although lithium-ion batteries are characterized by high energy density and long lifespan, equipment failures and damage, and even safety accidents, still occur frequently due to the continuous degradation of battery properties over time. Therefore, timely and reasonable prediction of lithium-ion degradation and lifespan is of great significance for ensuring the stability of equipment functions and the safe use and operation of equipment.
[0003] Currently, there are two main methods widely used for detecting and predicting the lifespan of lithium-ion batteries: one is direct measurement, which directly measures the capacity of lithium-ion batteries through charge and discharge under favorable experimental conditions. This method has high requirements for experimental equipment and environment and cannot predict battery degradation. The other is electrochemical impedance spectroscopy, which uses changes in the electrochemical impedance spectrum of lithium-ion batteries to predict their lifespan. This method can indirectly reflect changes in the internal materials of the battery during the aging process, but the prediction results of this method have large errors in the later stages of battery degradation. In fact, the internal microstructure and chemical composition of lithium-ion batteries are constantly changing during degradation. When these damages and degradations reach a certain level, they lead to a significant decline in battery lifespan. Existing methods detect and predict the lifespan of lithium-ion batteries from the perspective of their electrical performance, while ignoring changes in the internal microstructure of the battery. Some studies have established the relationship between low-frequency electrochemical noise and internal micro-damage in lithium-ion batteries, but there are no studies on quantitatively predicting the health status and lifespan of lithium-ion batteries based on low-frequency electrochemical noise through model fitting. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise. Based on the measurement results of battery capacity and low-frequency electrochemical noise, the method extracts the frequency domain characteristic parameters of the noise and uses a multivariate linear model to model and analyze the capacity degradation of the battery. This method can accurately establish a quantitative relationship between the frequency domain characteristic parameters of battery noise and battery capacity, and can predict battery lifespan with reliable and accurate prediction results.
[0005] To achieve the above objectives, this invention discloses a method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise, comprising:
[0006] S1: Conduct experiments on lithium-ion batteries, construct the battery capacity degradation space, and collect low-frequency noise timing signals;
[0007] Design a charge-discharge cycle test for a lithium-ion battery, design a low-frequency noise measurement test for a lithium-ion battery, conduct charge-discharge and low-frequency noise measurement tests for lithium-ion batteries, and construct a battery capacity degradation space. And collect low-frequency noise timing signals; the battery capacity degradation space It is a set consisting of n sets of experimental data, a The space contains 1 column and k rows of data, specifically:
[0008]
[0009] in, Let i be the capacity degradation vector of the i-th battery. Let be the battery capacity of the j-th battery in the i-th test; i be the number of batteries participating in the test; j be the battery cycle number; n be the number of batteries participating in the test; k be the number of battery cycle groups; and m be the number of charge-discharge test interval cycles.
[0010] S2: Process low-frequency noise time-series signals and construct a frequency domain feature parameter space;
[0011] The low-frequency noise time-series signal in step S1 is filtered and notch-filtered to extract the frequency domain information of the low-frequency noise and construct the frequency domain characteristic parameter space of the low-frequency electrochemical noise. The frequency domain feature parameter space It is a set consisting of n sets of experimental data; a The space contains 3 columns and k rows of data, specifically:
[0012]
[0013] in, Let i be the frequency domain characteristic parameter space of the i-th battery; Let be the slope of the 1 / f noise component in the low-frequency noise of the battery in the j-th i-th test; Let be the corner frequency of the 1 / f noise component in the low-frequency noise of the j-th battery during the i-th test. Let be the bandwidth of the GR noise component in the low-frequency noise of the battery in the j-th and i-th tests;
[0014] S3: Establish a lithium-ion battery life prediction model based on correlation analysis and least squares method;
[0015] Battery capacity degradation space in step S1 and the frequency domain characteristic parameter space in step S2 Correlation analysis was performed; a first-order polynomial from the least squares method was used as the correlation function for regression fitting to obtain the lithium-ion battery life prediction model, specifically:
[0016]
[0017] in, For lithium-ion battery life prediction; a0 is the constant term of the life prediction model function; b1 is the influence coefficient of the first input variable; b2 is the influence coefficient of the second input variable; b3 is the influence coefficient of the third input variable;
[0018] When there are n sets of experimental data, the lithium-ion battery life prediction model is in matrix form, specifically:
[0019] C = X·P;
[0020] Where C is the battery capacity result vector; X is the frequency domain feature parameter matrix; and P is the parameter vector.
[0021] S4: Use a lithium-ion battery life prediction model to predict the lifespan of lithium-ion batteries;
[0022] Obtain the lithium-ion battery life prediction model in step S3 and perform correlation analysis; obtain the frequency domain characteristic parameters of the low-frequency electrochemical noise of the lithium-ion battery under test, obtain the capacity of the lithium-ion battery under test, and then correspond it to the theoretical cycle number of the lithium-ion battery under test. Based on the lithium-ion battery life prediction model, obtain the life prediction result of the lithium-ion battery.
[0023] Preferably, the lithium-ion battery charge-discharge cycle test designed in step S1 is as follows:
[0024] Based on the battery performance parameters, the capacity degradation threshold of the lithium-ion battery under test was set to 80%.
[0025] Based on the expected cycle life Lc of the sample battery, the single charge-discharge cycle time t, and the number of sample batteries n, the number of charge-discharge test interval cycles m is determined. The battery is subjected to km cycles to induce capacity degradation, and the battery capacity and process voltage-current curves after km cycles are recorded to determine the lithium-ion battery life. Specifically, the number of charge-discharge test interval cycles m is:
[0026]
[0027] Where Lc is the expected cycle life of the sample battery; t is the single charge-discharge cycle time of the sample battery; This indicates rounding up to the nearest integer.
[0028] Preferably, the low-frequency noise measurement test for lithium-ion batteries in step S1 is designed as follows:
[0029] First, select a low-pass signal amplifier with low background noise and high amplification gain and a data acquisition card with a set acquisition frequency, and assemble a low-frequency noise measurement test platform with good electromagnetic shielding performance.
[0030] Then, after performing m charge-discharge cycles on the battery sample, the battery was taken in a fully charged state for no-load measurement, and the noise timing data of the battery after k cycles was recorded.
[0031] Preferably, in step S1, lithium-ion battery charging / discharging and low-frequency noise measurement tests are conducted, specifically as follows:
[0032] Charge-discharge cycle experiments were conducted on n selected lithium-ion batteries to obtain their battery capacity after km cycles, and then a capacity degradation sample space of n lithium-ion batteries was constructed. One of them The space contains 1 column and k rows of data, which represents the battery capacity of the nth battery after km cycles;
[0033] Then, using the constructed battery low-frequency noise measurement test platform, the electrochemical noise was measured after km cycles.
[0034] Preferably, the low-frequency noise timing signal in step S2 is subjected to filtering and notch filtering, specifically as follows:
[0035] S211: Perform a 100Hz low-pass filter on the electrochemical noise of the n samples collected to remove high-frequency electrical signal interference coupled during the acquisition process;
[0036] S212: Perform notch filtering to remove AC interference coupled in during the filtering process in the experiment;
[0037] S213: Retain low-frequency electrochemical noise below 100Hz without AC component to obtain n sets of noise time-series datasets to be analyzed; the electrochemical noise is a time-series signal dataset, if the sampling rate is f s The sampling duration is t s Then the length of the time series signal dataset is f. s *t s In the preprocessing of low-frequency electrochemical noise, it is necessary to retain the data in the middle 300 seconds of the time series data to obtain accurate data.
[0038] Preferably, the extraction of frequency domain information of low-frequency noise in step S2 specifically includes the following steps:
[0039] S221: Perform a Fourier transform on the time-series signal to obtain the frequency domain signal;
[0040] S222: Obtain the power spectral density of the frequency domain signal using the Welch method;
[0041] S223: By comparing the features of low-frequency noise, the exponent n0, corner frequency f0 and GR noise bandwidth Δf are extracted from the 1 / f noise component and GR noise component of the low-frequency noise part, respectively, to obtain the 1 / f noise component and GR noise component of the low-frequency noise part.
[0042] Preferably, the frequency domain feature parameter matrix X and parameter vector P in step S3 are specifically as follows:
[0043] The frequency domain feature parameter matrix X is specifically:
[0044]
[0045] The parameter vector P is specifically:
[0046] P = (a0, b1, b2, b3) T .
[0047] Preferably, in step S3, a first-order polynomial from the least squares method is used as the correlation function for regression fitting to obtain the lithium-ion battery life prediction model, which establishes the battery capacity as the response variable. With a set of input variable frequency domain feature parameters and The functional relationship between them is as follows:
[0048] S31: Extract the battery capacity degradation sample space obtained in step S1. The sample space of the low-frequency electrochemical noise frequency domain characteristic parameters of the battery obtained in step S2. Construct the battery capacity result vector C and the frequency domain feature parameter matrix X, and let the initial parameter vector...
[0049] S32: Calculate the error vector, specifically:
[0050] E = X·PC;
[0051] Where E is the error vector;
[0052] S33: Sample space of characteristic parameters in the frequency domain of low-frequency electrochemical noise Within, iterate through the values of the parameter vector P, such that E 2 The minimum value is used to determine the parameter vector P to be determined.
[0053] Preferably, the correlation analysis in step S3 specifically includes:
[0054] If the correlation coefficient of the correlation analysis results is less than or equal to 0.8, then the fitting effect of the lithium-ion battery life prediction model does not meet the requirements.
[0055] If the correlation coefficient of the correlation analysis results is greater than 0.8, then the fitting effect of the lithium-ion battery life prediction model meets the requirements.
[0056] The specific formula for calculating the correlation coefficient is as follows:
[0057]
[0058] Wherein, ρ(C,Y) i ) represents the correlation coefficient; Y i Let C be the frequency domain feature parameter vector, where i = 1, 2, 3; Cov(C, Y) is the covariance; D(C) is the variance of the battery capacity result vector C; D(Y) is the variance of the frequency domain feature parameter vector C. i Y is the frequency domain feature parameter vector. i The variance;
[0059] The frequency domain feature parameter vector Y i Specifically:
[0060]
[0061] Where Y1 is the first vector of frequency domain feature parameters; Y2 is the second vector of frequency domain feature parameters; and Y3 is the third vector of frequency domain feature parameters.
[0062] Compared with the prior art, the present invention has the following beneficial effects:
[0063] This invention, based on measurements of battery capacity and low-frequency electrochemical noise, extracts noise frequency domain characteristic parameters and employs a multivariate linear model to model and analyze battery capacity degradation. It accurately establishes a quantitative relationship between battery noise frequency domain characteristic parameters and battery capacity, and can predict battery life with more reliable and accurate prediction results than other methods. This invention can use phenomena directly generated by the internal degradation mechanism of the battery for modeling and analysis, enabling analysis and calculation for battery life prediction. It provides technical support for guiding the rational selection and use of batteries, improving battery reliability, and optimizing battery performance. Attached Figure Description
[0064] Figure 1 This is a flowchart of the lithium-ion battery life prediction method based on low-frequency electrochemical noise according to the present invention.
[0065] Figure 2 This is a schematic diagram of timing noise signals shown in an embodiment of the present invention;
[0066] Figure 3 This is a schematic diagram of the filtered timing noise signal shown in an embodiment of the present invention;
[0067] Figure 4The noise power spectral density diagram obtained by the Welch method is shown in an embodiment of the present invention. Detailed Implementation
[0068] Exemplary embodiments, features, and aspects of the present invention will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.
[0069] This invention provides a method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise, such as... Figure 1 As shown, experiments were conducted on lithium-ion batteries to construct a battery capacity degradation space and collect low-frequency noise time-series signals. The low-frequency noise time-series signals were processed to construct a frequency domain characteristic parameter space. Based on correlation analysis and the least squares method, a lithium-ion battery lifetime prediction model was established, and the prediction of lithium-ion battery lifespan was completed. This includes:
[0070] Step S1: Conduct experiments on lithium-ion batteries to construct a battery capacity degradation space and collect low-frequency noise timing signals.
[0071] An analysis was conducted on the selected lithium-ion batteries to be tested. Based on the relevant battery manuals, key performance parameters such as nominal capacity, charging voltage, charging current, charging method, and discharge cut-off voltage were compiled, as shown in Table 1. The nominal capacity was used as the degradation threshold for the battery capacity. Based on the parameter information, the degradation threshold for the selected lithium-ion batteries to be tested was determined. An analysis of the usage conditions of the selected batteries to be tested showed that their predicted cycle life was 110 cycles. The number of battery samples was determined, and the single charge-discharge cycle time of the sample batteries was determined based on the charge-discharge performance parameters of the selected batteries to be tested.
[0072] Table 1. Key performance parameters of the selected lithium-ion batteries to be tested.
[0073]
[0074]
[0075] Design a charge-discharge cycle test for a lithium-ion battery, specifically:
[0076] Based on the expected cycle life Lc of the sample battery, the single charge-discharge cycle time t, and the number of sample batteries n, the number of charge-discharge test interval cycles m is determined. The battery is subjected to km cycles to induce capacity degradation, and the battery capacity and process voltage-current curves after km cycles are recorded to determine the lithium-ion battery life. The specific number of charge-discharge test interval cycles m is as follows:
[0077]
[0078] Where Lc is the expected cycle life of the sample battery; t is the single charge-discharge cycle time of the sample battery; This indicates rounding up to the nearest integer.
[0079] The number of battery samples was determined to be n = 8. Based on the charge / discharge performance parameters of the selected batteries under test, the single charge / discharge cycle time of the sample batteries was determined to be t = 9 min. The number of charge / discharge test intervals for the selected batteries under test was determined to be m = 10 cycles.
[0080] Design a low-frequency noise measurement experiment for lithium-ion batteries, specifically:
[0081] First, select a low-pass signal amplifier with low background noise and high amplification gain, and set the acquisition card with the acquisition frequency to assemble and build a low-frequency noise measurement test platform with good electromagnetic shielding performance.
[0082] Then, after performing m charge-discharge cycles on the battery sample, the battery was taken in a fully charged state for no-load measurement, and the noise timing data of the battery after k cycles was recorded.
[0083] Conducting lithium-ion battery charge / discharge and low-frequency noise measurement tests, specifically:
[0084] Charge-discharge cycle experiments were conducted on n selected lithium-ion batteries to obtain their battery capacity after km cycles, and then a capacity degradation sample space of n lithium-ion batteries was constructed. One of them The space contains 1 column and k rows of data, which represents the battery capacity of the nth battery after km cycles.
[0085] As shown in Table 2, the established test loading scheme was used to conduct charge-discharge cycle tests on the sample batteries. The battery capacity of each sample battery after km of charge-discharge cycles during each test was used as a set of capacity data, thus constructing a capacity degradation data space for eight sample batteries. Figure 2 As shown, the low-frequency electrochemical noise time series data of each sample battery after km charge-discharge cycles during each experiment are used as a set of noise data to construct the original low-frequency electrochemical noise time series dataset of 8 sample batteries.
[0086] Table 2 shows the capacity degradation data of the sample batteries after km of charge-discharge cycles.
[0087]
[0088] Then, using the constructed battery low-frequency noise measurement test platform, the electrochemical noise was measured after km cycles.
[0089] Building a space for battery capacity degradation And collect low-frequency noise timing signals; battery capacity degradation space Ω1 n It is a set consisting of n sets of experimental data, a The space contains 1 column and k rows of data, specifically:
[0090]
[0091] in, Let i be the capacity degradation vector of the i-th battery. Let be the battery capacity of the j-th battery in the i-th test; i be the battery number participating in the test; j be the battery cycle number; n be the number of batteries participating in the test; k be the battery cycle number; and m be the number of charge-discharge test interval cycles. Step S2: Process the low-frequency noise time-series signal and construct the frequency domain feature parameter space. The low-frequency noise time-series signal in step S1 is filtered and notched. Specifically, step S211: The electrochemical noise of the n samples is subjected to a 100Hz low-pass filter to remove high-frequency electrical signal interference coupled during the acquisition process.
[0092] Step S212: Perform notch filtering to remove AC interference coupled in during the filtering process of the experiment.
[0093] Step S213: Retain the low-frequency electrochemical noise below 100Hz without AC component to obtain n sets of noise time series datasets to be analyzed; the electrochemical noise is a time series signal dataset, and if the sampling rate is f s The sampling duration is t s Then the length of the time series signal dataset is f. s *t s In the preprocessing of low-frequency electrochemical noise, it is necessary to retain the data in the middle 300 seconds of the time series data to obtain accurate data.
[0094] like Figure 3 As shown, the data in the middle 300s of the time series was selected and subjected to 100Hz low-pass digital filtering to remove high-frequency electrical signal interference coupled during the acquisition process. Then, notch filtering was performed on the data to remove AC interference coupled into the signal during the filtering process, resulting in 8 sets of low-frequency electrochemical noise time series datasets to be analyzed.
[0095] The specific steps for extracting the frequency domain information of low-frequency noise are as follows:
[0096] Step S221: Perform Fourier transform on the time-series signal to obtain the frequency domain signal.
[0097] Step S222: Use the Welch method to obtain the power spectral density of the frequency domain signal.
[0098] like Figure 4The noise power spectral density map obtained by the Welch method is shown in the embodiment of the present invention. The filtered time-series signal is subjected to Fourier transform, and then spectrum analysis is performed by Welch power spectral density to calculate the energy distribution of the signal at different frequencies and extract frequency domain feature parameters.
[0099] Step S223: By comparing the features of low-frequency noise, extract the exponent n0, corner frequency f0 and GR noise bandwidth Δf of the 1 / f noise component and GR noise component of the low-frequency noise part, respectively, to obtain the 1 / f noise component and GR noise component of the low-frequency noise part, and construct the data space of the frequency domain feature parameters of the 8 sample batteries, as shown in Table 3.
[0100] Table 3. Frequency domain characteristic parameters of a sample battery after km of charge-discharge cycles.
[0101]
[0102] Constructing the frequency domain characteristic parameter space of low-frequency electrochemical noise Frequency domain characteristic parameter space It is a set consisting of n sets of experimental data; a The space contains 3 columns and k rows of data, specifically:
[0103]
[0104] in, Let i be the frequency domain characteristic parameter space of the i-th battery; Let be the slope of the 1 / f noise component in the low-frequency noise of the battery in the j-th i-th test; Let be the corner frequency of the 1 / f noise component in the low-frequency noise of the j-th battery during the i-th test. Let be the bandwidth of the GR noise component in the low-frequency noise of the battery during the j-th and i-th tests.
[0105] Step S3: Establish a lithium-ion battery life prediction model based on correlation analysis and least squares method;
[0106] Battery capacity degradation space in step S1 and the frequency domain characteristic parameter space in step S2 Correlation analysis was performed; a first-order polynomial in the least squares method was used as the correlation function for regression fitting to obtain the lithium-ion battery life prediction model, which establishes the battery capacity as the response variable. With a set of input variable frequency domain feature parameters and The functional relationship between them is as follows:
[0107] Step S31: The battery capacity degradation sample space obtained in step S1 is... The sample space of the low-frequency electrochemical noise frequency domain characteristic parameters of the battery obtained in step S2. Construct the battery capacity result vector C and the frequency domain feature parameter matrix X, and let the initial parameter vector...
[0108] Step S32: Calculate the error vector, specifically:
[0109] E = X·PC;
[0110] Where E is the error vector.
[0111] Step S33: Sample space of low-frequency electrochemical noise frequency domain characteristic parameters Within, iterate through the values of the parameter vector P, such that E 2 The minimum value is used to determine the parameter vector P to be determined.
[0112] The lithium-ion battery life prediction model is as follows:
[0113]
[0114] in, For lithium-ion battery life prediction; a0 is the constant term of the life prediction model function; b1 is the influence coefficient of the first input variable; b2 is the influence coefficient of the second input variable; b3 is the influence coefficient of the third input variable.
[0115] When there are n sets of experimental data, the lithium-ion battery life prediction model is in matrix form, specifically:
[0116] C = X·P;
[0117] Where C is the battery capacity result vector; X is the frequency domain feature parameter matrix; and P is the parameter vector.
[0118] The frequency domain feature parameter matrix X is as follows:
[0119]
[0120] The parameter vector P is specifically:
[0121] P = (a0, b1, b2, b3) T .
[0122] Step S4: Use the lithium-ion battery life prediction model to predict the lifespan of lithium-ion batteries;
[0123] Obtain the lithium-ion battery life prediction model in step S3 and perform correlation analysis. If the correlation coefficient of the correlation analysis result is less than or equal to 0.8, the fitting effect of the lithium-ion battery life prediction model does not meet the requirements.
[0124] If the correlation coefficient of the correlation analysis results is greater than 0.8, then the fitting effect of the lithium-ion battery life prediction model meets the requirements.
[0125] The specific formula for calculating the correlation coefficient is as follows:
[0126]
[0127] Wherein, ρ(C,Y) i ) represents the correlation coefficient; Y i Let C be the frequency domain feature parameter vector, where i = 1, 2, 3; Cov(C, Y) is the covariance; D(C) is the variance of the battery capacity result vector C; D(Y) is the variance of the frequency domain feature parameter vector C. i Y is the frequency domain feature parameter vector. i The variance.
[0128] Frequency domain feature parameter vector Y i Specifically:
[0129]
[0130] Where Y1 is the first vector of frequency domain feature parameters; Y2 is the second vector of frequency domain feature parameters; and Y3 is the third vector of frequency domain feature parameters.
[0131] The frequency domain characteristic parameters of the low-frequency electrochemical noise of the lithium-ion battery under test are obtained, the capacity of the lithium-ion battery under test is obtained, and then the theoretical cycle number of the lithium-ion battery under test is correspondingly obtained. Based on the lithium-ion battery life prediction model, the life prediction result of the lithium-ion battery is obtained.
[0132] Based on the correlation coefficient formula, the correlation coefficients between the capacity degradation vector and the three frequency domain characteristic parameter vectors of low-frequency electrochemical noise for the selected lithium-ion battery under test are 0.92, 0.93, and 0.89, respectively. Combined with the capacity degradation data space... and frequency domain characteristic parameter data space Linear regression fitting using the least squares method yielded the following results for the lifetime prediction model of the selected lithium-ion battery under test based on low-frequency electrochemical noise:
[0133] C=1120.367+41.834*n0+9.355*f0-7.415*Δf.
[0134] The beneficial effects of this invention are as follows: This invention provides a method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise. Based on an example analysis of lithium-ion batteries, this invention measures battery capacity and low-frequency electrochemical noise, extracts noise frequency domain characteristic parameters, and uses a multivariate linear model to model and analyze battery capacity degradation. This accurately establishes a quantitative relationship between battery noise frequency domain characteristic parameters and battery capacity, and can predict battery lifespan. The embodiments of this invention can use phenomena directly generated by the internal degradation mechanism of the battery for modeling and analysis, and can be used for battery life prediction analysis and calculation. This provides technical support for guiding the rational selection and use of batteries, improving battery reliability, and optimization. Verification and analysis of examples demonstrate that this method has good application effects.
[0135] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.
Claims
1. A method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise, characterized in that, It includes: S1: Conduct experiments on lithium-ion batteries, construct the battery capacity degradation space, and collect low-frequency noise timing signals; Design a charge-discharge cycle test for a lithium-ion battery, design a low-frequency noise measurement test for a lithium-ion battery, conduct charge-discharge and low-frequency noise measurement tests for lithium-ion batteries, and construct a battery capacity degradation space. And collect low-frequency noise timing signals; The battery capacity degradation space It is by A set of experimental data, a Space includes List Row data, specifically: ； in, For the first A battery capacity degradation vector; For the first The first The battery capacity of the battery in this test; Number the batteries that participated in the experiment; Number the number of battery cycles; The number of batteries participating in the experiment; This refers to the number of battery cycle groups. Take a positive integer; Number of charge / discharge test cycles; S2: Process low-frequency noise time-series signals and construct a frequency domain feature parameter space; The low-frequency noise time-series signal in step S1 is filtered and notch-filtered to extract the frequency domain information of the low-frequency noise and construct the frequency domain characteristic parameter space of the low-frequency electrochemical noise. The frequency domain feature parameter space It is by A collection of experimental data; a The space contains 3 columns Row data, specifically: ； in, For the first The frequency domain characteristic parameter space of a battery; For the first The first The slope of the 1 / f noise component in the low-frequency noise of the battery in this experiment; For the first The first The corner frequency of the 1 / f noise component in the low-frequency noise of the battery in this experiment; For the first The first The bandwidth of the GR noise component in the low-frequency noise of the battery in this experiment; The specific steps for extracting the frequency domain information of low-frequency noise in step S2 are as follows: S221: Perform a Fourier transform on the time-series signal to obtain the frequency domain signal; S222: Obtain the power spectral density of the frequency domain signal using the Welch method; S223: By comparing the features of low-frequency noise, the exponents of the 1 / f noise component and the GR noise component of the low-frequency noise part are extracted respectively. Transition frequency and GR noise bandwidth The 1 / f noise component and GR noise component of the low-frequency noise part are obtained. S3: Establish a lithium-ion battery life prediction model based on correlation analysis and least squares method; Battery capacity degradation space in step S1 and the frequency domain characteristic parameter space in step S2 Correlation analysis was performed; a first-order polynomial from the least squares method was used as the correlation function for regression fitting to obtain the lithium-ion battery life prediction model, specifically: ； in, For predicting the lifespan of lithium-ion batteries; This is a constant term in the lifetime prediction model function; The influence coefficient of the first input variable; The influence coefficient of the second input variable; The influence coefficient of the third input variable; When there are n sets of experimental data, the lithium-ion battery life prediction model is in matrix form, specifically: ； in, This is a vector representing the battery capacity results; This is the frequency domain characteristic parameter matrix; For parameter vectors; S4: Use a lithium-ion battery life prediction model to predict the lifespan of lithium-ion batteries; Obtain the lithium-ion battery life prediction model in step S3, perform correlation analysis to determine the fitting effect of the lithium-ion battery life prediction model; obtain the frequency domain characteristic parameters of the low-frequency electrochemical noise of the lithium-ion battery under test, obtain the capacity of the lithium-ion battery under test, and the corresponding theoretical cycle number of the lithium-ion battery under test; obtain the life prediction result of the lithium-ion battery based on the lithium-ion battery life prediction model.
2. The method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise according to claim 1, characterized in that: The lithium-ion battery charge-discharge cycle test designed in step S1 is as follows: Based on the battery performance parameters, the capacity degradation threshold of the lithium-ion battery under test was set to 80%. Based on the expected cycle life of the sample battery Single charge / discharge cycle time and sample battery count Determine the number of charge / discharge test cycles. ,pass The cycle caused battery capacity degradation, and the cycle was recorded. The battery capacity and process voltage-current curves after each charge-discharge test are used to determine the lifespan of the lithium-ion battery; the number of charge-discharge test cycles is also considered. Specifically: ； in, Expected cycle life for the sample batteries; The single charge-discharge cycle time of the sample battery; This indicates rounding up to the nearest integer.
3. The method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise according to claim 1, characterized in that: The design of the low-frequency noise measurement experiment for lithium-ion batteries in step S1 is as follows: First, select a low-pass signal amplifier with low background noise and high amplification gain and a data acquisition card with a set acquisition frequency, and assemble a low-frequency noise measurement test platform with good electromagnetic shielding performance. Then, after performing m inter-group charge-discharge cycles on the battery samples, the batteries were taken in a fully charged state for no-load measurement, and the noise timing data of the batteries after k cycles were recorded.
4. The method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise according to claim 1, characterized in that: The lithium-ion battery charging and discharging and low-frequency noise measurement tests conducted in step S1 are as follows: For the selected A charge-discharge cycle experiment was conducted on a lithium-ion battery to obtain... A lithium-ion battery Battery capacity after each cycle, and thus construction Sample space of capacity degradation of lithium-ion batteries One of them Space includes List Row data, for the first One battery in Battery capacity after one cycle; Then, using the constructed battery low-frequency noise measurement test platform, in Electrochemical noise was measured after one cycle.
5. The method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise according to claim 1, characterized in that: The low-frequency noise timing signal in step S2 is filtered and notch-filtered. The specific steps are as follows: S211: The collected data Electrochemical noise of 100 samples Low-pass filtering removes high-frequency electrical signal interference coupled during the acquisition process; S212: Perform notch filtering to remove AC interference coupled in during the filtering process in the experiment; S213: Leave 100 The following low-frequency electrochemical noise without alternating current component is obtained. A set of noisy time-series datasets to be analyzed; The electrochemical noise is a time-series signal dataset, if the sampling rate is... Sampling time is Then the length of the time series signal dataset is ; In the preprocessing of low-frequency electrochemical noise, it is necessary to retain the data in the middle 300 seconds of the time series data to obtain accurate data.
6. The method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise according to claim 1, characterized in that: The frequency domain feature parameter matrix in step S3 and parameter vector Specifically: The frequency domain feature parameter matrix Specifically: ； The parameter vector Specifically: 。 7. The method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise according to claim 1, characterized in that: In step S3, a first-order polynomial from the least squares method is used as the correlation function for regression fitting to obtain the lithium-ion battery life prediction model, which establishes the response variable as battery capacity. With a set of input variable frequency domain feature parameters , and The functional relationship between them is as follows: S31: Extract the battery capacity degradation sample space obtained in step S1. The sample space of the low-frequency electrochemical noise frequency domain characteristic parameters of the battery obtained in step S2. Construct a battery capacity result vector and frequency domain characteristic parameter matrix Let the initial parameter vector ; S32: Calculate the error vector, specifically: ； in, This is the error vector; S33: Sample space of characteristic parameters in the frequency domain of low-frequency electrochemical noise Inside, traverse the parameter vector The value of makes such that Minimize to determine the parameter vector to be solved .
8. The method for predicting the lifespan of lithium-ion batteries based on low-frequency electrochemical noise according to claim 1, characterized in that: In step S4, correlation analysis is performed, specifically as follows: If the correlation coefficient of the correlation analysis results is less than or equal to 0.8, then the fitting effect of the lithium-ion battery life prediction model does not meet the requirements. If the correlation coefficient of the correlation analysis results is greater than 0.8, then the fitting effect of the lithium-ion battery life prediction model meets the requirements. The specific formula for calculating the correlation coefficient is as follows: ； in, The correlation coefficient; Let be the frequency domain feature parameter vector, where ; For covariance; Battery capacity result vector The variance; Frequency domain feature parameter vector The variance; The frequency domain feature parameter vector Specifically: ； in, This is the first vector of frequency domain feature parameters; This is the second vector of frequency domain feature parameters; This is the third vector of frequency domain characteristic parameters.