A method for predicting the capacity of lithium-ion batteries for cascade utilization

Abstract

The present invention provides a method for predicting the capacity of a lithium-ion battery for cascade utilization, which relates to the field of lithium-ion batteries. The present invention uses a frequency stacking method to quickly obtain the FIS curve of the predicted battery, and processes the EIS curve by wavelet denoising to extract the amplitude characteristic |Z| and phase-frequency characteristic of the battery. The ohmic impedance R of the EIS curve is extracted by first-order equivalent circuit fitting. Ω , Charge transfer impedance R ct , diffusion impedance R w As a feature category for quickly predicting battery SOH, the five groups of features are fitted with multivariate quadratic functions to predict the battery capacity, thereby realizing the capacity prediction of lithium-ion batteries for cascade utilization; compared with the full charge and discharge method and the traditional FIS impedance measurement method to extract prediction features, it can effectively reduce the detection time and the workload of battery testing, and in the actual operation process, it will not cause damage to the battery itself, thus reducing the testing cost.

Description

Technical Field

[0001] The present invention relates to the field of lithium-ion batteries, and in particular to a method for predicting the capacity of lithium-ion batteries for cascade utilization. Background Art

[0002] As environmental pollution becomes more and more serious, more and more automobile manufacturers are beginning to develop new energy vehicles, and electric vehicles replacing fuel vehicles has become a future development trend. Since the commercialization of electric vehicles in 2014, according to the average service life of power batteries of 3-8 years, the "scrap wave" of power batteries has arrived. The estimated scrap volume of power batteries in China alone this year may reach 170,000 tons. How to deal with these batteries retired from electric vehicles with rated capacity below 80% has become the main difficulty at this stage. The difficulties in the cascade utilization of lithium-ion batteries are mainly concentrated in three aspects: intelligent disassembly, remaining capacity prediction and system integration technology. The most critical difficulty is the remaining capacity prediction of the battery.

[0003] The existing SOH prediction methods can be divided into capacity testing, ohmic impedance extraction and EIS measurement. Among them, the capacity test tests the capacity of the battery by fully charging the battery and converts its fully charged capacity into the battery's SOH. This method is the most direct way to evaluate the battery's SOH, but its test time is very long and it is difficult to adapt to the remaining capacity estimation of large-scale retired batteries. Ohmic impedance extraction is usually used in combination with EIS measurement, that is, constructing a corresponding equivalent circuit to fit the battery's EIS curve, and extracting the impedance that can characterize the battery capacity through parameter identification to estimate the remaining capacity. This method uses an in-situ detection technology to determine its state by measuring a wide-band impedance spectrum. It is currently recognized as one of the most promising methods for characterizing the aging effects of lithium-ion batteries, because EIS provides information about different aging mechanisms and can reduce energy waste caused by multiple battery charging and discharging.

[0004] Although the traditional EIS detection method can reduce the test time to a certain extent and thus improve efficiency, its test time for 0.01Hz-10kHz is still 8 minutes to 8 minutes and 30 seconds, which is still a long time. In addition, the impedance data obtained by constructing an equivalent circuit for parameter identification are all rough values, and the test error is large. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies in the prior art and provide a method for predicting the capacity of a second-life lithium-ion battery, which improves the prediction efficiency while meeting the requirements for the SOH prediction accuracy of the second-life battery.

[0006] The present invention is implemented by the following technical scheme: a method for predicting the capacity of a lithium-ion battery for cascade utilization, specifically comprising the following steps:

[0007] Step S1: obtaining EIS data of the battery to be predicted by the frequency stacking method, and the EIS data includes the real part, imaginary part and basic characteristics of the battery EIS;

[0008] Step S2: The real part, imaginary part and basic characteristics of the battery EIS obtained in step S1 are subjected to wavelet denoising to obtain denoised EIS data;

[0009] Step S3: The real and imaginary parts of the battery EIS after denoising in step S2 are fitted by a first-order equivalent circuit to obtain the polarization impedance that characterizes each polarization process of the battery. The polarization impedance includes the ohmic impedance R Ω , Charge transfer impedance R ct and diffusion resistance R w , while the basic characteristics of EIS are grouped into the amplitude of EIS, which contains the amplitude-frequency characteristics, and the phase, which contains the phase-frequency characteristics;

[0010] Step S4: using the polarization impedance of each polarization process obtained in step S3 and the amplitude and phase of EIS as feature categories for battery health status prediction;

[0011] Step S5: The ohmic impedance R obtained in step S3 is Ω , Charge transfer impedance R ct , diffusion impedance R w , amplitude-frequency characteristics |Z| and phase-frequency characteristics The weight combination is carried out through grayscale correlation recombination, and then the coefficients of each parameter are obtained through polynomial function fitting to predict the battery capacity.

[0012] According to the above technical solution, preferably, step S1 specifically includes the following sub-steps:

[0013] Step S1-1: charging a plurality of batteries to be predicted by constant current and constant voltage charging until the charge state of each battery to be predicted is 50%; leaving the charged batteries to be predicted to stand for at least 30 minutes;

[0014] Step S1-2: Take 12 frequency points at logarithmic frequency intervals for every 10 octaves in the test frequency range;

[0015] Step S1-3: superimposing the sine waves of the selected frequencies, and using the superimposed frequencies as the excitation of the battery;

[0016] Step S1-4: inputting the excitation signal into the battery to be predicted to obtain the response of the test battery;

[0017] Step S1-5: split the response into responses at each frequency point using fast Fourier transform;

[0018] Step S1-6: Obtain battery EIS related data through stimulation and response.

[0019] According to the above technical solution, preferably, step S2 specifically includes the following sub-steps:

[0020] Step S2-1: Select a wavelet basis function and a decomposition layer number N, and then perform N-layer wavelet decomposition on the noisy signal to obtain a wavelet sequence and wavelet decomposition coefficients;

[0021] Step S2-2: using a threshold function to perform threshold processing on the decomposed wavelet sequence, and denoising the wavelet sequence to obtain a wavelet quantity;

[0022] Step S2-3: reconstruct the wavelet quantity after threshold processing to obtain a denoised signal.

[0023] According to the above technical solution, preferably, the denoised signal in step S2-3 is verified by KK transformation.

[0024] According to the above technical solution, preferably, step S5 specifically includes the following sub-steps:

[0025] Step S5-1: Determine the grey relational sequence, including a capacity sequence x 0 =(x 0 (1),x 0 (2),…,x 0 (n)) and the characteristic sequence x of the three types of impedance characteristics and amplitude and phase characteristics obtained according to the first-order equivalent circuit i =(x i (1),x i (2),…,x i (n)), i represents five different features, and n is the sequence length;

[0026] Step S5-2: Parameter normalization:

[0027]

[0028]

[0029] in and They represent the normalized capacity sequence and feature sequence respectively;

[0030] Step S5-3: Determine the grey correlation coefficient and calculate the correlation coefficient of each indicator in the reference sequence:

[0031]

[0032] k=1,2,…,n

[0033] Among them, ρ is the resolution coefficient, k is the serial number of the correlation coefficient of the characteristic sequence;

[0034] Step S5-4: Calculate the correlation r between the comparison sequence and the reference sequence i :

[0035]

[0036] Step S5-5: Calculate the weight β of each indicator according to the grey correlation degree i ;

[0037]

[0038] Step S5-5: Calculate the gray weight of each parameter using the gray correlation degree of each type of parameter, and combine them to obtain the EIS impedance R EIS ; The predicted capacity f(R EIS );

[0039]

[0040]

[0041] Where A, B, and C are the coefficients of the polynomial function.

[0042] The beneficial effects of the present invention are:

[0043] (1) Compared with the method of extracting prediction features by using full charge and discharge and the traditional EIS impedance measurement method, the present invention can effectively reduce the detection time and the workload of battery testing, and will not cause damage to the battery itself during the actual operation, thereby reducing the testing cost;

[0044] (2) Predicting the battery health status by using the method of predicting battery capacity according to multivariate quadratic function fitting can improve the prediction efficiency while meeting the requirements for the accuracy of SOH prediction of cascade utilization batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A schematic diagram of a process according to an embodiment of the present invention is shown;

[0046] Figure 2 A capacity decay trend diagram of a lithium-ion battery according to an embodiment of the present invention is shown;

[0047] Figure 3 Flowchart diagram of step 1.

[0048] Figure 4 It is the EIS curve obtained by quickly detecting the battery to be predicted by the frequency superposition method of the present invention;

[0049] Figure 5This is a schematic diagram of the calculation results of KK transformation;

[0050] Figure 6 It is the relationship diagram between phase-frequency characteristics and capacity at each frequency point;

[0051] Figure 7 It is the relationship diagram between amplitude-frequency characteristics and capacity at each frequency point;

[0052] Figure 8 It is the battery capacity prediction result diagram; DETAILED DESCRIPTION

[0053] The technical solution of the invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the invention, not all of them. Based on the embodiments of the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0054] like Figure 1 As shown, the present invention provides a method for predicting the capacity of a lithium-ion battery for cascade utilization, which specifically includes the following steps:

[0055] Step S1: quickly detecting the EIS data of the battery to be predicted by the overlapping frequency method, and the EIS data includes the real part, imaginary part and basic characteristics of the battery EIS;

[0056] Step S2: The real part, imaginary part and basic characteristics of the battery EIS obtained in step S1 are processed by wavelet denoising to obtain more accurate EIS data, and the reliability is verified by KK transformation;

[0057] Step S3: The real and imaginary parts of the battery EIS are fitted by a first-order equivalent circuit to obtain the polarization impedance that characterizes each polarization process of the battery. The polarization impedance includes the ohmic impedance R Ω , Charge transfer impedance R ct and diffusion resistance R w , while the basic characteristics of EIS are grouped into the amplitude of EIS, which contains the amplitude-frequency characteristics, and the phase, which contains the phase-frequency characteristics;

[0058] Step S4: taking the polarization impedance of each polarization process and the amplitude and phase of EIS as feature categories for battery health state prediction;

[0059] Step S5: The above ohmic impedance R Ω , Charge transfer impedance R ct , diffusion impedance R w The amplitude-frequency characteristics and phase-frequency characteristics are weighted and combined through grayscale correlation recombination, and then the coefficients of each parameter are obtained through polynomial function fitting to predict the capacity of the battery, and the error analysis of the predicted value is performed.

[0060] like Figure 2 As shown, when the capacity of a lithium-ion battery decays to 80% of its initial capacity, it is considered obsolete, i.e., in a 0% health state, with a total of 150 cycles;

[0061] Further, the process of step S1 to quickly detect the EIS data of the battery to be predicted by the overlapping frequency method is as follows: Figure 3 As shown, it specifically includes the following steps:

[0062] Step S1-1: charging a plurality of batteries to be predicted in a constant current and constant voltage charging mode until the charge state of each battery to be predicted is 50%; leaving the charged batteries to be predicted to stand for at least 30 minutes;

[0063] Step S1-2: Take 12 frequency points at logarithmic frequency intervals for every 10 octaves in the test frequency range;

[0064] Step S1-3: superimposing the sine waves of the selected frequencies, and using the superimposed frequencies as the excitation of the battery;

[0065] Step S1-4: inputting the excitation signal into the battery to obtain the response of the test battery;

[0066] Step S1-5: split the response into responses at each frequency point using fast Fourier transform;

[0067] Step S1-6: Obtain battery EIS related data through stimulation and response.

[0068] Further, the EIS curve measured by the frequency superposition method in step S1 is as follows Figure 4 As shown by the dotted line, the accuracy of the low-frequency part is exchanged for the reduction of measurement time. The average measurement time of the 10-times overlapping frequency method when the measurement frequency range is 0.1Hz to 10kHz is only 16 seconds, which greatly shortens the measurement time and improves efficiency compared with traditional measurement methods.

[0069] Further, in step S2, the low-frequency parts of the real part and the imaginary part are processed by using a wavelet denoising method, which specifically includes the following steps:

[0070] Step S2-1: Select a suitable wavelet basis function and the number of decomposition layers N, and then perform N-layer wavelet decomposition on the noisy signal to obtain a wavelet sequence and wavelet decomposition coefficients;

[0071] Step S2-2: Select an appropriate threshold function to perform threshold processing on the decomposed wavelet sequence, and denoise the wavelet sequence to obtain a wavelet quantity;

[0072] Step S2-3: reconstruct the wavelet quantity after threshold processing to obtain a denoised signal.

[0073] This embodiment uses sym2 wavelet as the wavelet basis function, selects the soft threshold function as the threshold function, and the number of decomposition layers is selected according to the different characteristics of different components. Wavelet denoising can retain the authenticity of the original sequence to the greatest extent and remove the noise signal in the sequence. The processed EIS curve is as follows Figure 4 Shown as solid line.

[0074] The processed EIS data is transformed by KK to calculate the residuals of the real and imaginary parts, and the results are as follows: Figure 5 As shown, none of them is greater than ±0.2%. Therefore, the EIS curve measured by the above method greatly improves the test efficiency while ensuring the accuracy.

[0075] Step S3 specifically includes the following contents:

[0076] The battery EIS curve after the first-order equivalent circuit fitting process is used to obtain the battery's ohmic impedance R Ω , Charge transfer impedance R ct and diffusion resistance R w , and record;

[0077] The amplitude and phase of the EIS of the battery to be predicted are obtained by the frequency superposition method. The fluctuation of the low-frequency part of the amplitude and phase is also large. Therefore, the wavelet denoising method is also used to process this feature, and then multiple characteristic frequency points are selected to test the amplitude-frequency characteristics and phase-frequency characteristics.

[0078] Step S4 specifically includes the following contents:

[0079] The impedance amplitude at different frequency points gradually increases with battery aging, and exhibits strong monotonicity in the low-frequency part of the battery EIS curve, such as Figure 6 As shown, the characteristic amplitude of the low-frequency part is suitable for use as a prediction feature for cascade utilization, and the processed amplitude-frequency feature |Z| is recorded.

[0080] The characteristic phase below the selected 10kHz frequency point will fluctuate up and down as the battery capacity decays, and does not show an obvious monotonic law. The characteristic phase at the 10kHz frequency point will gradually decrease as the battery capacity decays, such as Figure 7 As shown, the high-frequency part of EIS measured by the overlapping frequency method has higher reliability, so the phase-frequency characteristics of the high-frequency part are suitable to be used as a screening factor for cascade utilization.

[0081] Step S5 specifically includes the following contents:

[0082] The ohmic resistance R Ω , Charge transfer impedance R ct , diffusion impedance R w, amplitude-frequency features and phase-frequency features are weighted and combined by grayscale correlation recombination, which specifically includes the following steps:

[0083] Determine the grey relational sequence, including a capacity sequence x 0 =(x 0 (1),x 0 (2),…,x 0 (n)) and the characteristic sequence x of the three types of impedance characteristics and amplitude and phase characteristics obtained according to the first-order equivalent circuit i =(x i (1),x i (2),…,x i (n)), i represents five different features and n is the sequence length.

[0084] Normalize the parameters to avoid the impact of different dimensions on the results:

[0085]

[0086]

[0087] in and They represent the normalized capacity sequence and feature sequence respectively;

[0088] Determine the grey correlation coefficient and calculate the correlation coefficient of each index in the reference sequence:

[0089]

[0090] k=1,2,…,n

[0091] Among them, ρ is the resolution coefficient, usually taken as 0.5, and k is the serial number of the correlation coefficient of the characteristic sequence;

[0092] Calculate the correlation r between the comparison sequence and the reference sequence i :

[0093]

[0094] Calculate the weight β of each indicator according to the grey correlation degree i ;

[0095]

[0096] The obtained correlation degree is extremely strong within 0.8-1.0, strongly correlated within 0.6-0.8, moderately correlated within 0.4-0.6, weakly correlated within 0.2-0.4, extremely weakly correlated within 0.0-0.2 or no correlation.

[0097] Ohmic resistance R Ω , Charge transfer impedance Rct , diffusion impedance R w , the amplitude-frequency characteristics of EIS |Z| and the phase-frequency characteristics of EIS The gray correlations with capacity are 0.612, 0.845, 0.863, 0.91 and 0.874 respectively. It can be seen that the ohmic impedance R Ω , Charge transfer impedance R ct , diffusion impedance R w , EIS amplitude |Z| and EIS phase-frequency characteristics The correlation with the battery capacity is high, so the features are strongly correlated with the battery capacity. Therefore, the battery capacity can be predicted more accurately by using the features.

[0098] The grey weight of each parameter is calculated by the grey correlation degree of each parameter, and the EIS impedance R is obtained by combining them. EIS In order to efficiently predict retired batteries, this embodiment uses the above characteristics as parameters to establish a polynomial function:

[0099]

[0100]

[0101] Where A, B, and C are coefficients of the polynomial function, and the battery capacity is obtained by fitting the above function.

[0102] The capacity measured within the first 20% SOH of the eight batteries and the above characteristics were used as fitting data to obtain the capacity prediction function of each battery. Then the capacity outside the first 20% SOH and the above characteristics were used to verify the accuracy of the function prediction. The prediction results are shown in Figure 8 As shown;

[0103] Methods for verifying the accuracy of prediction values ​​include: mean square error (MSE), root mean square error (RMSE) and mean absolute error (MAE). Among them, MSE can evaluate the degree of change of data. The smaller the MSE value, the better the accuracy of the prediction model. RMSE is the arithmetic square root of MSE. MAE can better reflect the actual situation of the prediction value error. Among them:

[0104] The specific calculation formula of mean square error is:

[0105]

[0106] The specific calculation formula of the root mean square error is:

[0107]

[0108] The specific calculation formula for the mean absolute error is:

[0109]

[0110] C 0 and C p They represent the actual capacity value and the predicted capacity value of length N respectively;

[0111] The beneficial effects of this embodiment are as follows: the present invention uses ohmic impedance, charge transfer impedance, diffusion impedance, EIS amplitude and phase as feature categories for battery SOH prediction, and utilizes multivariate quadratic functions to fit the features and capacity to obtain the coefficients of each feature, so as to more accurately predict the capacity of the battery. This method can effectively improve the speed of battery EIS detection and prediction, reduce the workload of battery testing, and meet the requirements for battery life capacity prediction accuracy.

[0112] In the description of the invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0113] In the description of the invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the invention can be understood according to specific circumstances.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the invention rather than to limit it. Although the invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solution to deviate from the scope of the technical solution of the embodiments of the invention.

Claims

1. A method for predicting the capacity of a lithium-ion battery for cascade utilization, characterized in that: The specific steps include: Step S1: obtaining EIS data of the battery to be predicted by the frequency stacking method, and the EIS data includes the real part, imaginary part and basic characteristics of the battery EIS; Step S1-1: charging a plurality of batteries to be predicted by constant current and constant voltage charging until the charge state of each battery to be predicted is 50%; leaving the charged batteries to be predicted to stand for at least 30 minutes; Step S1-2: Take 12 frequency points at logarithmic frequency intervals for every 10 octaves in the test frequency range; Step S1-3: superimposing the sine waves of the selected frequencies, and using the superimposed frequencies as the excitation of the battery; Step S1-4: inputting the excitation signal into the battery to be predicted to obtain the response of the test battery; Step S1-5: split the response into responses at each frequency point using fast Fourier transform; Step S1-6: obtaining battery EIS related data through stimulation and response; Step S2: The real part, imaginary part and basic characteristics of the battery EIS obtained in step S1 are subjected to wavelet denoising to obtain denoised EIS data; Step S3: The real and imaginary parts of the battery EIS after denoising in step S2 are fitted by a first-order equivalent circuit to obtain the polarization impedance that characterizes each polarization process of the battery. The polarization impedance includes the ohmic impedance R Ω , Charge transfer impedance R ct and diffusion resistance R w , while the basic characteristics of EIS are grouped into the amplitude of EIS, which contains the amplitude-frequency characteristics, and the phase, which contains the phase-frequency characteristics; Step S4: using the polarization impedance of each polarization process obtained in step S3 and the amplitude and phase of EIS as feature categories for battery health status prediction; Step S5: The ohmic impedance R obtained in step S3 is Ω , Charge transfer impedance R ct , diffusion impedance R w , amplitude-frequency characteristics |Z| and phase-frequency characteristics The weight combination is carried out through grayscale correlation recombination, and then the coefficients of each parameter are obtained through polynomial function fitting to predict the battery capacity.

2. The method for predicting the capacity of a lithium-ion battery for cascade utilization according to claim 1, characterized in that: Step S2 specifically includes the following sub-steps: Step S2-1: Select a wavelet basis function and a decomposition layer number N, and then perform N-layer wavelet decomposition on the noisy signal to obtain a wavelet sequence and wavelet decomposition coefficients; Step S2-2: using a threshold function to perform threshold processing on the decomposed wavelet sequence, and denoising the wavelet sequence to obtain a wavelet quantity; Step S2-3: reconstruct the wavelet quantity after threshold processing to obtain a denoised signal.

3. The method for predicting the capacity of a lithium-ion battery for cascade utilization according to claim 2, characterized in that: The denoised signal of step S2-3 is verified by KK transformation.

4. The method for predicting the capacity of a lithium-ion battery for cascade utilization according to claim 1, characterized in that: Step S5 specifically includes the following sub-steps: Step S5-1: Determine the grey relational sequence, including a capacity sequence And the characteristic sequences of three types of impedance characteristics and amplitude and phase characteristics obtained according to the first-order equivalent circuit i represents five different features, and n is the sequence length; Step S5-2: Parameter normalization: in, and They represent the normalized capacity sequence and feature sequence respectively; Step S5-3: Determine the grey correlation coefficient and calculate the correlation coefficient of each indicator in the reference sequence: in, is the resolution coefficient, is the serial number of the correlation coefficient of the characteristic sequence; Step S5-4: Calculate the correlation between the comparison sequence and the reference sequence : Step S5-5: Calculate the weight of each indicator according to the grey correlation degree ; Step S5-5: Calculate the gray weight of each parameter using the gray correlation degree of each type of parameter, and combine them to obtain the EIS impedance R EIS ; The predicted capacity f(R EIS ); Where A, B, and C are the coefficients of the polynomial function.

Images