Lithium precipitation real-time detection method and system for fast charging cycle of lithium battery

Abstract

The application provides a lithium precipitation real-time detection method and system for fast charging cycles of lithium batteries, comprising: determining initial peak position parameters and initial frequency window based on a relaxation time distribution fingerprint benchmark library; selecting feature frequency points corresponding to the initial frequency window from a detection sub-period and constructing a corresponding multi-sinusoidal superimposed disturbance signal, superimposing the multi-sinusoidal superimposed disturbance signal on the charging current and injecting the lithium battery to be detected; collecting response voltage signals and response current signals of the lithium battery to be detected, and calculating impedance data corresponding to each feature frequency point; determining the current peak position parameters of the feature relaxation time peak based on the impedance data, and calculating the drift amount of the current peak position parameters relative to the initial peak position parameters; comparing the drift amount with a preset drift threshold and judging whether the lithium battery to be detected precipitates lithium. The frequency query range is reduced, the lithium precipitation of the lithium battery during the fast charging cycle is detected in real time, and the generalization ability during the detection of different system batteries is improved.

Description

Technical Field

[0001] This invention relates to the field of lithium battery testing technology, and in particular to a real-time detection method and system for lithium plating in lithium batteries for fast charging cycles. Background Technology

[0002] Electric vehicle power batteries have high safety requirements, especially their safety performance after fast-charging cycles. This has been listed as a mandatory testing item for individual power battery cells, requiring batteries to maintain qualified safety performance after multiple fast-charging cycles. However, under fast-charging conditions, the increased negative electrode polarization and abnormal lithium-ion deposition can easily lead to lithium plating. Lithium plating forms lithium dendrites, which can pierce the separator and cause internal short circuits, a major contributing factor to battery thermal runaway, fire, and explosion. Therefore, achieving early, real-time, and online detection of lithium plating during fast-charging cycles is a core technological requirement for ensuring fast-charging safety and meeting national standard testing requirements.

[0003] Traditional lithium plating detection methods mostly use single-frequency points or a limited number of fixed frequencies for impedance monitoring. However, the lithium plating sensitivity frequencies are not the same for different battery systems, and the fixed-frequency scheme has poor generalization ability across different batches and models of batteries. In addition, traditional EIS testing is time-consuming (several minutes to tens of minutes), making it difficult to achieve real-time detection during fast charging cycles. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a method and system for real-time detection of lithium plating in lithium batteries during fast charging cycles, which solves the problem that it is difficult to achieve real-time detection of lithium plating in lithium batteries during fast charging cycles in the prior art.

[0005] In a first aspect, this application provides a real-time detection method for lithium plating in lithium batteries for fast charging cycles, including:

[0006] The initial peak position parameters and initial frequency window of the characteristic relaxation time peak corresponding to the lithium battery to be tested are determined based on the pre-established relaxation time distribution fingerprint benchmark library.

[0007] In each charging pulse cycle of the fast charging cycle, at least one detection sub-cycle containing the initial frequency window is embedded. Feature frequency points corresponding to the initial frequency window are selected from the detection sub-cycle to form a dynamic frequency set, and a multi-sine superimposed perturbation signal corresponding to the dynamic frequency set is constructed. The multi-sine superimposed perturbation signal is superimposed on the charging current and injected into the lithium battery to be tested.

[0008] The response voltage signal and response current signal of the lithium battery under test in response to the multi-sinusoidal superimposed disturbance signal are collected, and the impedance data corresponding to each characteristic frequency point in the dynamic frequency set is calculated based on the response voltage signal and the response current signal.

[0009] The current peak position parameter of the characteristic relaxation time peak is determined based on the impedance data corresponding to each characteristic frequency point, and the drift of the current peak position parameter relative to the initial peak position parameter is calculated.

[0010] The drift amount is compared with a preset drift threshold to determine whether the lithium battery under test has lithium plating.

[0011] Secondly, this application provides a real-time lithium plating detection system for lithium batteries oriented towards fast charging cycles, including a processor and a memory; wherein the memory stores a computer program, which is used by the processor to load and execute the real-time lithium plating detection method for lithium batteries oriented towards fast charging cycles as described in any one of the first aspects.

[0012] In the real-time lithium plating detection method and system for fast-charging cycles of this embodiment, the frequency range for real-time detection is first compressed from the full frequency band to a local window using a relaxation time distribution fingerprint benchmark library. This reduces the frequency query range and facilitates real-time detection of lithium plating during fast-charging cycles. Simultaneously, each battery obtains its own exclusive initial frequency window through its respective relaxation time distribution fingerprint benchmark library, achieving an adaptive configuration of "one window per battery," thereby improving the generalization ability for detecting different battery systems. Then, a detection sub-cycle is embedded within each charging pulse cycle, thus parallelizing the detection and charging processes in time. This prevents the detection process from occupying an independent test time window, instead utilizing the natural gaps in the charging pulse cycle, avoiding interference from the detection operation during fast charging and fundamentally solving the problem of not being able to insert long-term tests during fast charging. Finally, relative drift is used as a criterion to determine whether lithium plating has occurred, reducing the impact of individual differences and the environment on the judgment results. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart illustrating a real-time lithium plating detection method for lithium batteries oriented towards fast charging cycles, provided in one embodiment of this application.

[0015] Figure 2 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0016] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. Based on the description of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present invention.

[0017] In the description of this invention, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0018] The terms “upper,” “lower,” “left,” “right,” “front,” “back,” “top,” “bottom,” “inner,” and “outer,” etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use. They are only for the convenience of description and simplification, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention.

[0019] The terms “first,” “second,” “third,” etc., are used merely to distinguish elements with similar attributes, not to indicate or imply relative importance or a specific order.

[0020] The terms “include,” “comprising,” or any other variation thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0021] like Figure 1 As shown, this embodiment provides a real-time detection method for lithium plating in lithium batteries for fast charging cycles, including:

[0022] Step S100: Determine the initial peak position parameters and initial frequency window of the characteristic relaxation time peak corresponding to the lithium battery to be tested based on the pre-established relaxation time distribution fingerprint benchmark library;

[0023] Step S200: Embed at least one detection sub-cycle containing the initial frequency window in each charging pulse cycle of the fast charging cycle, select the feature frequency points corresponding to the initial frequency window from the detection sub-cycle to form a dynamic frequency set, construct the multi-sine superimposed perturbation signal corresponding to the dynamic frequency set, superimpose the multi-sine superimposed perturbation signal onto the charging current and inject it into the lithium battery to be tested.

[0024] Step S300: Collect the response voltage signal and response current signal of the lithium battery under test in response to the multi-sinusoidal superimposed disturbance signal, and calculate the impedance data corresponding to each characteristic frequency point in the dynamic frequency set based on the response voltage signal and the response current signal;

[0025] Step S400: Determine the current peak position parameter of the characteristic relaxation time peak based on the impedance data corresponding to each characteristic frequency point, and calculate the drift of the current peak position parameter relative to the initial peak position parameter;

[0026] Step S500: Compare the drift amount with a preset drift threshold and determine whether the lithium battery to be tested has lithium plating.

[0027] In the real-time lithium plating detection method for fast-charging cycles in this embodiment, the frequency range for real-time detection is first compressed from the full frequency band to a local window using a relaxation time distribution fingerprint benchmark library. This reduces the frequency query range and facilitates real-time detection of lithium plating during fast-charging cycles. Simultaneously, each battery obtains its own exclusive initial frequency window through its respective relaxation time distribution fingerprint benchmark library, achieving an adaptive configuration of "one window per battery," thereby improving the generalization ability for detecting different battery systems. Then, a detection sub-cycle is embedded within each charging pulse cycle, thus parallelizing the detection and charging processes in time. This prevents the detection process from occupying an independent test time window, instead utilizing the natural gaps in the charging pulse cycle, avoiding interference from the detection operation during fast charging and fundamentally solving the problem of not being able to insert long-term tests during fast charging. Finally, relative drift is used as a criterion to determine whether lithium plating has occurred, reducing the impact of individual differences and the environment on the judgment results.

[0028] Step S100: Determine the initial peak position parameters and initial frequency window of the characteristic relaxation time peak corresponding to the lithium battery to be tested based on the pre-established relaxation time distribution fingerprint benchmark library.

[0029] The relaxation time distribution fingerprint benchmark library refers to the dataset of relaxation time distribution characteristics of a lithium battery in its initial fresh state, obtained through a complete broadband electrochemical impedance spectroscopy test and relaxation time distribution inversion calculation before performing fast charge cycle testing. Each lithium battery under test corresponds to a unique relaxation time distribution fingerprint benchmark library, which is one-to-one bound to the physical entity of the battery and cannot be mixed between different batteries. The relaxation time distribution fingerprint benchmark library records the electrochemical characteristics of the battery in its initial state before undergoing fast charge cycle aging and lithium plating, and can serve as a reference zero point for all subsequent dynamic tests.

[0030] A characteristic relaxation time peak refers to a relaxation time peak in the relaxation time distribution spectrum that has a physical correspondence with the specific electrochemical process of lithium metal deposition. Each peak in the relaxation time distribution spectrum corresponds to a specific electrochemical polarization process within the lithium battery. Characteristic relaxation time peaks exhibit a distinct single-peak morphology in the relaxation time distribution spectrum, with an identifiable peak apex, a measurable half-peak width, and an integrable peak area. They can be determined through offline calibration experiments. Since characteristic relaxation time peaks are located in the mid-frequency range of the relaxation time spectrum, their corresponding characteristic frequency range is typically from 10Hz to 1000Hz. This range overlaps with the characteristic frequency range of the charge transfer process. However, after lithium deposition, the peak position shifts towards lower frequencies, while normal charge transfer peaks do not exhibit this shift. Therefore, characteristic relaxation time peaks can be distinguished by their shift characteristics.

[0031] The initial peak position parameter refers to the original position of the characteristic relaxation time peak on the relaxation time axis in the relaxation time distribution fingerprint benchmark library. The magnitude of the initial peak position parameter reflects the speed of the interfacial reaction kinetics of the lithium battery in its fresh state. Generally, the smaller the value, the faster the charge transfer process and the better the rate performance of the battery; while the larger the value, the slower the charge transfer process, and therefore the higher the risk of lithium plating under high-rate charging.

[0032] The initial frequency window refers to the frequency range corresponding to the characteristic relaxation time peak in the relaxation time distribution fingerprint benchmark library, including the lower and upper frequency limits. Typically, different battery systems have different positions of their characteristic relaxation time peaks due to differences in materials, design, and manufacturing processes, resulting in different initial frequency windows. By establishing an independent benchmark library for each battery and determining its own unique initial frequency window, an adaptive configuration of "one battery, one window" is achieved, solving the problem of poor generalization ability of fixed-frequency schemes. Furthermore, by setting the initial frequency window as the search space for dynamic frequencies during fast charging cycles, the detection system can avoid full-band frequency sweeping and instead dynamically select characteristic frequency points only within the initial frequency window range. This compresses the wideband detection range of 1mHz to 10kHz into a narrowband window, facilitating real-time detection during fast charging cycles.

[0033] In one embodiment, before determining the initial peak position parameters and initial frequency window of the characteristic relaxation time peak corresponding to the lithium battery to be detected based on the pre-established relaxation time distribution fingerprint benchmark library, the method further includes:

[0034] Step S101: Apply an electrochemical impedance spectroscopy excitation at a preset frequency to the lithium battery under test, and collect the full-band impedance spectroscopy data corresponding to the lithium battery under test.

[0035] Electrochemical impedance spectroscopy (EIS) excitation involves applying a small-amplitude sinusoidal AC voltage or current signal with a frequency that varies over time to the lithium-ion battery under test, thereby stimulating the corresponding electrochemical response within the battery. To avoid interfering with normal battery operation, the excitation current is typically 5% to 15% of the lithium-ion battery's rated current. This ensures that the frequency of the output signal under excitation is the same as the input signal frequency, with only the amplitude and phase changing. Simultaneously, it avoids the excitation signal having a substantial impact on the battery's state of charge, temperature, and other state parameters. Furthermore, the frequency of EIS excitation covers the characteristic frequency range corresponding to all electrochemical processes within the lithium-ion battery within a preset frequency range, extending from 10kHz (corresponding to ohmic conduction and SEI film processes) in the high-frequency band to 1mHz (corresponding to lithium-ion solid-phase diffusion processes) in the low-frequency band, and varies continuously or in steps according to a certain pattern.

[0036] Full-band impedance spectroscopy data refers to the set of complex impedance values ​​of a lithium battery under test collected at different frequency points under electrochemical impedance spectroscopy excitation, reflecting the static electrochemical characteristics of the battery during testing. The impedance data for each frequency point consists of two parts: the real part and the imaginary part of the impedance.

[0037] When applying electrochemical impedance spectroscopy excitation, the lithium battery under test is first placed in a constant temperature environment, with the temperature controlled within the range of 20℃ to 45℃, to eliminate the influence of temperature changes on the impedance spectrum. Then, the battery is charged to a preset state of charge and left to stand for a sufficient period of time to wait for the battery to reach electrochemical equilibrium. Next, the battery is connected to the impedance spectroscopy testing equipment, and the test parameters are set (for example, the frequency range is set to 1mHz to 10kHz, the number of frequency points is set to 30 to 50, and the excitation amplitude is set to 5% to 15% of the battery's rated current). Finally, the test is started, and the impedance spectroscopy testing equipment can automatically apply the excitation signal according to the set frequency sequence and collect the amplitude and phase of the response signal at each frequency point, and calculate the real part and imaginary part of the impedance at that frequency point.

[0038] Step S102: Perform deconvolution calculation on the full-band impedance spectrum data based on the preset relaxation time distribution algorithm to obtain a relaxation time distribution spectrum containing multiple relaxation time peaks.

[0039] The relaxation time distribution spectrum refers to the functional relationship curve between relaxation time and distribution function obtained after converting the electrochemical impedance spectroscopy from the frequency domain to the time domain. In the relaxation time distribution spectrum, the peak area of ​​each relaxation time peak is proportional to the contribution of the electrochemical process to the total polarization resistance of the battery, the peak height reflects the intensity of the process, and the peak width reflects the distribution range of the time constant of the process. Furthermore, the relaxation time distribution spectrum can be represented in Nyquist or Bode plots as overlapping electrochemical processes separated in the time domain. The relaxation time peaks corresponding to different relaxation times represent different electrochemical polarization processes, each corresponding to an electrochemical process with a specific time constant, thus ensuring that each relaxation time peak corresponds to an electrochemical process with a specific time constant.

[0040] In one embodiment, the deconvolution calculation of the full-band impedance spectrum data based on a preset relaxation time distribution algorithm to obtain a relaxation time distribution spectrum containing multiple relaxation time peaks specifically includes: acquiring full-band electrochemical impedance spectroscopy data of the lithium-ion battery under test within a preset frequency range; constructing a full-band relaxation time distribution function based on the full-band electrochemical impedance spectroscopy data within the preset frequency range; discretizing the full-band relaxation time distribution function on a preset relaxation time axis to obtain multiple discrete points; solving for the discrete distribution value corresponding to each discrete point on the full-band relaxation time distribution function; and generating the relaxation time distribution spectrum based on the relative position of the discrete distribution value on the relaxation time axis.

[0041] Specifically, based on the relaxation time distribution theory, the continuous relaxation time distribution function γ(lnτ) can be deconvolved by inversely calculating from the impedance data Z(ω) at discrete frequency points. The expression for the full-band relaxation time distribution function is:

[0042]

[0043] in, For ohmic resistance at the high-frequency limit, Angular frequency, The relaxation characteristic time, It is the imaginary unit.

[0044] Since directly solving the full-band relaxation time distribution function leads to severe oscillations, it is necessary to first discretize the continuous relaxation time axis into N discrete points within a preset range, where N ranges from 100 to 200. After discretization, the continuous distribution function γ(lnτ) is approximated as a discrete vector γ, which transforms the full-band relaxation time distribution function into a system of linear equations: ,in It is a vector composed of complex impedances at M frequency points (M is the number of frequency points, usually 30 to 50). ={ For an M×N kernel matrix, its elements , Let be the angular frequency at the i-th frequency point. This represents the k-th relaxation time point.

[0045] To overcome the ill-posedness of deconvolution calculations when solving for the discrete distribution values ​​corresponding to each discrete point, a Tikhonov regularization constraint is introduced, transforming the problem into an optimization problem with the objective function:

[0046]

[0047] in ={ For an M×N kernel matrix, its elements ,in Let i be the frequency of the i-th characteristic frequency point; For regularization parameters; This represents the sum of squares of the deviations between the reconstructed impedance and the measured impedance. This is a regularization term.

[0048] We can find a γ that minimizes the value of the objective function. Solving this minimization problem yields the optimal γ, which is the desired relaxation time distribution function.

[0049] After determining the relaxation time distribution function γ, the discrete distribution values ​​of the relaxation time distribution function γ at N relaxation time points are calculated respectively. The coordinates of the kth relaxation time point are obtained as follows: .

[0050] Finally With γ as the horizontal axis and γ as the vertical axis, the relaxation time distribution spectrum is formed by plotting k relaxation time points on the coordinate system and connecting them.

[0051] Step S103: Determine at least one characteristic relaxation time peak in the relaxation time distribution spectrum, and record the initial peak position parameters and initial frequency window corresponding to the characteristic relaxation time peak.

[0052] In one embodiment, determining at least one characteristic relaxation time peak in the relaxation time distribution spectrum specifically includes: marking relaxation time peaks in the relaxation time distribution spectrum that are located within a preset characteristic time constant range and whose peak intensity exceeds a preset intensity threshold as the characteristic relaxation time peaks.

[0053] The characteristic time constant of the lithium metal deposition process is usually 0.001s-0.1s. Therefore, peaks in the relaxation time spectrum that fall within this range can be used as candidate peaks. Then, candidate peaks with peak intensities exceeding 0.2 (the intensity threshold is usually 0.05-0.2, but can also be customized as needed) are recorded as characteristic relaxation time peaks.

[0054] In steps S101-S103, before the detection begins, a unique electrochemical fingerprint benchmark for the lithium battery under test is established through a one-time full-band DRT analysis. The original impedance data is gradually transformed into benchmark parameters that can be used for real-time detection, which facilitates millisecond-level real-time detection during fast charging cycles. This achieves adaptive detection with "one battery, one window" and solves the problem of poor generalization ability of fixed frequency schemes.

[0055] Step S200: Embed at least one detection sub-cycle containing the initial frequency window in each charging pulse cycle of the fast charging cycle, select the feature frequency points corresponding to the initial frequency window from the detection sub-cycle to form a dynamic frequency set, construct the multi-sine superimposed perturbation signal corresponding to the dynamic frequency set, superimpose the multi-sine superimposed perturbation signal onto the charging current and inject it into the lithium battery to be tested.

[0056] A charging pulse cycle refers to a complete time unit in which a pulsed charging current is applied to a lithium battery during a fast charging cycle. Each charging pulse cycle consists of a current injection phase and an interval phase. The duration of the current injection phase is typically 1 to 10 seconds, and the duration of the interval phase is typically 0.1 to 1 second. The entire charging process is composed of several repeated charging pulse cycles.

[0057] The detection sub-cycle is a time segment specifically allocated within the charging pulse cycle for performing lithium plating detection. The detection sub-cycle is embedded within the interval phase of the charging pulse cycle, performing impedance measurements at characteristic frequency points during this time period. In other words, the detection sub-cycle is not a time segment independent of the charging process but rather a component of the charging pulse cycle. Typically, the detection sub-cycle is embedded in the interval phase of the charging pulse cycle. Since the interval phase is a period with no or very low current, the superimposed multiple sinusoidal perturbation signals during this time will not interfere with the ongoing charging process. Simultaneously, the static state of the interval phase provides a relatively stable electrochemical environment for impedance measurement. Furthermore, since impedance measurements only need to be performed at 3 to 5 characteristic frequency points within the detection sub-cycle, and the excitation and acquisition at each frequency point only takes milliseconds, the duration of the entire detection sub-cycle is typically 10 to 50 milliseconds. This duration is much shorter than the duration of the interval phase, thus allowing it to be fully embedded without affecting the original timing structure of the charging pulse cycle.

[0058] When embedding a detection sub-cycle within each charging pulse cycle of a fast charging cycle, the embedding strategy for the detection sub-cycle can be preset before the start of the fast charging cycle to trigger its embedding. The embedding strategy can be triggered based on the pulse cycle interval, the number of pulse cycles, or the battery state. For example, detection can be triggered once every 5 charging pulse cycles; or once every 10 pulse cycles in the first 100 cycles, once every 5 pulse cycles in the 100 to 200 cycles, and once every pulse cycle after 200 cycles. The triggering strategy can be customized based on the lithium battery system.

[0059] A multi-sine superimposed perturbation signal refers to a composite excitation signal formed by linearly superimposing multiple sinusoidal signals of different frequencies in the time domain. This signal is superimposed on the charging current to excite the battery's impedance response at multiple characteristic frequency points. Typically, the multi-sine superimposed perturbation signal contains sinusoidal components corresponding to M characteristic frequency points, where M is an integer from 3 to 5. These characteristic frequency points are selected within an initial frequency window at logarithmic intervals, covering the frequency range where the lithium plating sensitive characteristic peak is located.

[0060] Furthermore, the step of selecting feature frequency points corresponding to the initial frequency window from the detection sub-period to form a dynamic frequency set, and constructing a multi-sine superimposed perturbation signal corresponding to the dynamic frequency set, specifically includes:

[0061] Step S201: Select a preset number of feature frequency points in real time from the dynamic neighborhood of the initial frequency window corresponding to the detection sub-cycle to construct a dynamic frequency set.

[0062] The dynamic neighborhood of the initial frequency window refers to the frequency range that expands dynamically based on the battery state at the current detection time, centered on the center frequency of the initial frequency window. Typically, the initial frequency window is the initial value of the dynamic neighborhood. As detection progresses, the dynamic neighborhood gradually deviates from the initial frequency window, moving in the direction of peak drift. Furthermore, the center frequency of the dynamic neighborhood is updated in real-time with the actual drift position of the relaxation time peak, rather than being fixed to the center frequency of the initial frequency window, ensuring that the dynamic neighborhood is always dynamically adjusted with the actual position of the current feature peak as the center. Simultaneously, the bandwidth of the dynamic neighborhood is dynamically adjusted based on the current detection confidence and battery state. When the detection confidence is high, the neighborhood bandwidth can be appropriately narrowed to reduce the number of frequency points and improve detection efficiency; when the detection confidence is low or the battery is in a rapidly changing state, the neighborhood bandwidth can be appropriately widened to prevent exceeding the search range due to excessively rapid peak changes.

[0063] Furthermore, the step of selecting a preset number of feature frequency points in real time from the dynamic neighborhood of the initial frequency window corresponding to the detection sub-cycle to construct a dynamic frequency set specifically includes: taking the real-time peak position of the feature relaxation time peak at the previous detection time as the center, extending a preset neighborhood bandwidth in both the low-frequency and high-frequency directions to form the dynamic neighborhood; selecting a preset number of feature frequency points in the dynamic neighborhood according to the logarithmic spacing method; sorting the feature frequency points according to preset rules and constructing the dynamic frequency set.

[0064] Because the detection time interval is short, the real-time peak position at the previous detection moment is usually used as the estimated position of the current feature peak. Then, based on the bandwidth expansion factor β (ranging from 2 to 5), the upper boundary of the dynamic neighborhood is determined. and lower boundary The dynamic neighborhood can then be obtained. , ].

[0065] Furthermore, since the relaxation time peak exhibits a logarithmically symmetric Gaussian peak shape in the frequency domain, taking the logarithm of the frequency can transform the peak shape into a symmetrical distribution. Therefore, using the logarithmic equal-interval sampling method ensures uniform coverage on the logarithmic frequency axis, making the number of characteristic frequency points distributed on both sides of the peak shape symmetrical, which is beneficial for accurately capturing the real-time peak position of characteristic frequency points.

[0066] The characteristic frequency points can be arranged in ascending (or descending) order of frequency to form a dynamic frequency set F = {f1, f2, ..., f...}. m}

[0067] Step S202: Generate a corresponding single-frequency sine wave signal for each characteristic frequency point in the dynamic frequency set, and linearly superimpose the single-frequency sine wave signals corresponding to each characteristic frequency point in the time domain to generate a composite time domain signal.

[0068] A single-frequency sinusoidal signal refers to a pure sinusoidal excitation signal whose frequency is a specific characteristic frequency point within a dynamic frequency concentration. The pure sinusoidal excitation signal at a specific characteristic frequency point can be determined using the following formula. :

[0069]

[0070] in, This is the amplitude coefficient, with a value ranging from 0.1 to 0.5. Let be the frequency of the i-th characteristic frequency point, and t be the time variable. is the initial phase angle, and is the initial phase value of the sine wave at time t=0.

[0071] A composite time-domain signal is a signal obtained by linearly superimposing single-frequency sine wave signals corresponding to all characteristic frequency points in the dynamic frequency set in the time domain.

[0072] Step S203: Adjust the peak amplitude of the composite time-domain signal to a preset injection amplitude range, and obtain the multi-sine superimposed perturbation signal.

[0073] Composite time domain signal peak amplitude The maximum value of the composite time-domain signal within a complete detection sub-cycle is, i.e. Then, based on the preset injection peak value... and peak amplitude Calculate the scaling factor k= (When k < 1, the composite time domain signal is compressed; when k > 1, the composite time domain signal is amplified); then the composite time domain signal is multiplied by the scaling factor k and normalized to obtain the normalized composite time domain signal. Finally, the normalized composite time domain signal is output as the final multi-sine superimposed perturbation signal and superimposed on the charging current to obtain the multi-sine superimposed perturbation signal.

[0074] In steps S201 to S203, by setting a dynamic domain, the selection range of the characteristic frequency point is bound to the actual position of the current peak. This can prevent the characteristic frequency point from falling outside the effective bandwidth of the peak when the drift reaches a certain level, thus avoiding detection failure.

[0075] Step S300: Collect the response voltage signal and response current signal of the lithium battery under test in response to the multi-sine superimposed disturbance signal, and calculate the impedance data corresponding to each characteristic frequency point in the dynamic frequency set based on the response voltage signal and the response current signal.

[0076] The testing equipment is equipped with a voltage acquisition channel and a current acquisition channel. The voltage acquisition channel acquires the response voltage signal of the lithium battery by setting a differential amplifier circuit, and the current acquisition channel is equipped with a Hall effect current sensor to indirectly obtain the response current signal by detecting the magnetic field generated by the current.

[0077] For each characteristic frequency point in the dynamic frequency set, the impedance data corresponding to the characteristic frequency point can be calculated based on the fast Fourier transform. First, the acquired voltage signal V(t) and current signal I(t) are subjected to fast Fourier transforms respectively to obtain their respective voltage signal spectrum V(k) and current signal spectrum I(k). Then, according to the formula... The corresponding impedance data can then be obtained. .

[0078] Step S400: Determine the current peak position parameter of the characteristic relaxation time peak based on the impedance data corresponding to each characteristic frequency point, and calculate the drift of the current peak position parameter relative to the initial peak position parameter.

[0079] The drift of the current peak position parameter relative to the initial peak position parameter refers to the change in the current peak position parameter relative to the initial peak position parameter. Generally, a positive drift indicates that the characteristic relaxation time peak has drifted towards a longer time direction, with a corresponding increase in relaxation time and a decrease in frequency; while a negative drift indicates that the characteristic relaxation time peak has drifted towards a shorter time direction, with a corresponding decrease in relaxation time and an increase in frequency.

[0080] Furthermore, determining the current peak position parameter of the characteristic relaxation time peak based on the impedance data corresponding to each characteristic frequency point specifically includes:

[0081] Step S401: Using the impedance data corresponding to the characteristic frequency point as input, the impedance data is inverted and fitted using a pre-built analytical basis function model library, and the corresponding local relaxation time distribution curve of the characteristic relaxation time peak is generated; wherein, the analytical basis functions in the analytical basis function model library include at least one of Cole-Cole type relaxation basis functions and Havriliak-Negami type relaxation basis functions.

[0082] Because only 3 to 5 characteristic frequency points are collected for real-time detection, it is difficult to reconstruct a continuous relaxation time distribution curve. Therefore, this embodiment introduces analytical basis functions to parameterize the relaxation time distribution. Instead of directly solving for the values ​​of unknown discrete points, it represents them as a weighted superposition of a few analytical basis functions, transforming the problem into solving for the weight coefficients and shape parameters of the analytical basis functions, thereby reducing the computational difficulty.

[0083] Each analytic basis function incorporates the shape information of the relaxation time distribution. For example, the frequency domain expression of a Cole-Cole type relaxation basis function is: ,in, The value of α is the broadening factor, which takes the range 0 < α ≤ 1. When α = 1, it is the standard Debye model. When α < 1, the relaxation time distribution exhibits a symmetrical broadening shape. The ohmic resistance of a lithium battery. This refers to the polarization resistor of a lithium battery. α represents the current peak position parameter of the characteristic relaxation time peak. The relaxation time distribution function of the Cole-Cole type relaxation basis functions exhibits a bell-shaped distribution that is symmetrical about the left and right sides on the logτ axis, and the peak width increases as α decreases.

[0084] Furthermore, the step of inverting and fitting the impedance data using a pre-constructed analytical basis function model library to generate the corresponding local relaxation time distribution curve of the characteristic relaxation time peak specifically includes: constructing a local relaxation time distribution function based on the impedance data; solving the local relaxation time distribution function based on a preset coarse-resolution relaxation time grid; identifying the coarse peak position of the characteristic relaxation time peak from the local relaxation time distribution function; determining a local time window centered on the coarse peak position and using a preset window expansion factor; constructing a refined relaxation time grid within the local time window; representing the relaxation time distribution on the refined relaxation time grid as an objective function containing the analytical basis function; solving the objective function of the analytical basis function based on Tikhonov regularization to determine the weight coefficients of the objective function; and reconstructing the relaxation time distribution curve within the local time window based on the weight coefficients to obtain the local relaxation time distribution curve.

[0085] The local relaxation time distribution function refers to the distribution function reconstructed within a preset relaxation time range based on impedance data at characteristic frequency points at the current detection time. It is denoted as... The expression for its constraint condition is:

[0086]

[0087] in, For impedance data, Let M be the frequency of the i-th characteristic frequency point, and M be the number of characteristic frequency points.

[0088] On a coarse-resolution relaxation time grid, L1 norm sparse regularization is used for solving the problem, and the objective function is expressed as follows:

[0089]

[0090] in, ={ For an M×N kernel matrix, its elements ,in For the k-th relaxation time point, For regularization parameters, Let be the relaxation time distribution vector to be solved. The L1 norm.

[0091] The objective function can be solved using an iterative soft thresholding algorithm or the alternating direction multiplier method. Through continuous iteration, the objective function is solved until it converges. This is the relaxation time distribution vector obtained by solving.

[0092] Then, by scanning the relaxation time distribution vector starting from k=2, the local maximum points are determined. ,at this time That is, the local maximum point at this time. The coarse peak position is determined by the maximum value of the local maximum point. If multiple local maxima exist, the local maximum point with the largest numerical value is selected as the coarse peak position.

[0093] Determining the position of the rough peak Then, based on its corresponding relaxation time point The boundary of the local time window is determined by the preset window expansion factor β, and its lower boundary is... upper boundary ,in The value range is 2-5.

[0094] Then, in the local time window [ , Within the [ ], a encrypted relaxation time grid is constructed in a logarithmically evenly spaced manner, with the number of grid points in the encrypted grid ranging from 20 to 40.

[0095] The relaxation time distribution to be solved on the encrypted relaxation time grid. It can be represented as a weighted superposition of K analytic basis functions, i.e.:

[0096]

[0097] in, Let be the weight coefficients of the q-th analytic basis function. Let be the distribution function of the q-th analytic basis function. The shape parameters are for the analytical basis functions.

[0098] Correspondingly, the expression for the objective function at this time is:

[0099]

[0100] Where M is the number of characteristic frequency points in the dynamic frequency set; For the i-th characteristic frequency point Impedance data obtained from actual measurements; To calculate the i-th feature frequency point under the current model parameters. The theoretical impedance value at that point is calculated using the following formula: , Let be the contribution weight of the q-th analytic basis function in the relaxation time distribution. This indicates that the q-th analytic basis function is at the characteristic frequency point. The contribution impedance value at that point, Let q be the shape parameter of the analytic basis function; This is the regularization parameter, and its value ranges from 10⁻ 6 Up to 10²; This is the regularization matrix; Let w be the weight coefficient vector of the analytic basis functions. If the number of analytic basis functions is K, then w = [w1, w2, ... w2]. q …, w k ]ᵀ. Each w q This represents the contribution weight of the q-th analytic basis function in the relaxation time distribution.

[0101] Then, Tikhonov regularization is used to solve the above objective function, and the solution process is iteratively optimized using the Levenberg-Marquardt algorithm. In each iteration, the contribution weights of the objective function are calculated. and shape parameters By calculating the gradient until the iteration results converge, the optimal contribution weight can be obtained. and shape parameters Weighting coefficients used as the objective function.

[0102] Finally, based on contribution weight and shape parameters The local relaxation time distribution curve is obtained by reconstructing the relaxation time distribution curve within the local time window.

[0103] Step S402: Extract the current peak position parameter of the characteristic relaxation time peak from the local relaxation time distribution curve.

[0104] On the local relaxation time distribution curve, scan all discrete points to find the discrete point with the largest distribution function value to initially locate the characteristic relaxation time peak. Then, take the largest discrete point and its two adjacent discrete points, and fit a quadratic parabola using these three points. Solve for the coefficients of the parabola by substituting the coordinates of the three points, and take the vertex position of the parabola as the current position of the characteristic relaxation time peak, and determine the current peak position parameters at this point. .

[0105] The current peak position parameter can be calculated using the logarithmic difference formula. Relative to the initial peak position parameter drift amount The expression for the logarithmic difference formula is:

[0106]

[0107] Step S500: Compare the drift amount with a preset drift threshold and determine whether the lithium battery to be tested has lithium plating.

[0108] The drift threshold ranges from 0.03 to 0.2, and can be customized based on different battery safety requirements. If the drift amount is greater than or equal to the drift threshold, lithium plating is determined to have occurred in the lithium battery; if the drift amount is less than the drift threshold, lithium plating is determined not to have occurred.

[0109] Furthermore, as the number of fast charging cycles increases, the value of the drift threshold decreases.

[0110] In the early stages of fast charging cycles, the battery is in a stable state with minimal normal aging drift. Using a higher threshold can effectively avoid misjudgments caused by measurement noise or minor fluctuations, ensuring the specificity of the detection. However, in the later stages of the cycle, battery aging intensifies, and the risk of lithium plating increases. Using a lower threshold can improve detection sensitivity, allowing for early warnings at the stage of lithium plating, thus providing time for subsequent handling.

[0111] In summary, the real-time lithium plating detection method for fast-charging cycles in this embodiment achieves several improvements. First, by obtaining a unique initial frequency window from a separate relaxation time distribution fingerprint benchmark library, an adaptive configuration of "one window per battery" is realized, improving the generalization ability for detecting different battery systems. Second, the detection process is parallelized with the charging process in time, ensuring that the detection process does not occupy an independent test time window, avoiding interference from the detection operation in the fast-charging process, and fundamentally solving the problem of not being able to insert long-term tests during fast charging. Finally, the relative drift is used as a criterion to determine whether lithium plating has occurred, reducing the impact of individual differences and the environment on the judgment results.

[0112] Based on the same inventive concept as the above embodiments, this embodiment also provides a real-time lithium plating detection system for fast charging cycles, including a processor and a memory; wherein, the memory stores a computer program, which is loaded by the processor and executed as described above for the real-time lithium plating detection method for fast charging cycles.

[0113] like Figure 2 As shown, based on the same inventive concept as the above embodiments, this embodiment also provides an electronic device. The electronic device is provided with a computer-readable storage medium storing instructions for loading and executing the method as described above by a processor.

[0114] The embodiments of the mobile terminal and computer-readable storage medium provided in this application include all the technical features of the embodiments of the above control method. The extended and explanatory content of the specification is basically the same as that of the embodiments of the above method, and will not be repeated here.

[0115] This application also provides a computer program product, which includes computer program code. When the computer program code is run on a computer, it causes the computer to perform the methods described in the various possible implementations above.

[0116] This application also provides a chip, including a memory and a processor. The memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that a device with the chip installed performs the methods described in the various possible implementations above.

[0117] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0118] In this application, the same or similar terms, concepts, technical solutions and / or application scenario descriptions are generally described in detail only when they appear for the first time. When they appear again, they are generally not repeated for the sake of brevity. When understanding the technical solutions and other contents of this application, the same or similar terms, concepts, technical solutions and / or application scenario descriptions that are not described in detail later can be referred to their previous relevant detailed descriptions.

[0119] In this application, the descriptions of the various embodiments have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0120] The technical features of the present application can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present application.

[0121] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in the above-mentioned storage medium and includes several instructions to cause a terminal device to execute the methods of each embodiment of this application. The above are only preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0122] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0123] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A real-time detection method for lithium plating in lithium batteries for fast charging cycles, characterized in that, include: The initial peak position parameters and initial frequency window of the characteristic relaxation time peak corresponding to the lithium battery to be tested are determined based on the pre-established relaxation time distribution fingerprint benchmark library. In each charging pulse cycle of the fast charging cycle, at least one detection sub-cycle containing the initial frequency window is embedded. Feature frequency points corresponding to the initial frequency window are selected from the detection sub-cycle to form a dynamic frequency set, and a multi-sine superimposed perturbation signal corresponding to the dynamic frequency set is constructed. The multi-sine superimposed perturbation signal is superimposed on the charging current and injected into the lithium battery to be tested. The response voltage signal and response current signal of the lithium battery under test in response to the multi-sinusoidal superimposed disturbance signal are collected, and the impedance data corresponding to each characteristic frequency point in the dynamic frequency set is calculated based on the response voltage signal and the response current signal. The current peak position parameter of the characteristic relaxation time peak is determined based on the impedance data corresponding to each characteristic frequency point, and the drift of the current peak position parameter relative to the initial peak position parameter is calculated. The drift amount is compared with a preset drift threshold to determine whether the lithium battery under test has lithium plating.

2. The real-time detection method for lithium plating in lithium batteries oriented towards fast charging cycles according to claim 1, characterized in that, Before determining the initial peak position parameters and initial frequency window of the characteristic relaxation time peak corresponding to the lithium battery to be tested based on the pre-established relaxation time distribution fingerprint benchmark library, the method further includes: An electrochemical impedance spectroscopy excitation at a preset frequency is applied to the lithium battery under test, and the impedance spectral data of the lithium battery under test across the entire frequency band is collected. Based on a preset relaxation time distribution algorithm, the full-band impedance spectrum data is deconvolved to obtain a relaxation time distribution spectrum containing multiple relaxation time peaks. Identify at least one characteristic relaxation time peak in the relaxation time distribution spectrum, and record the initial peak position parameters and initial frequency window corresponding to the characteristic relaxation time peak.

3. The real-time detection method for lithium plating in lithium batteries for fast charging cycles according to claim 2, characterized in that, The preset relaxation time distribution algorithm is used to perform deconvolution calculation on the full-band impedance spectrum data to obtain a relaxation time distribution spectrum containing multiple relaxation time peaks, specifically including: Acquire full-band electrochemical impedance spectroscopy data of the lithium-ion battery under test within a preset frequency range; A full-band relaxation time distribution function is constructed based on the full-band electrochemical impedance spectroscopy data within the preset frequency range; The full-band relaxation time distribution function is discretized on a preset relaxation time axis to obtain multiple discrete points; Solve for the discrete distribution values ​​of each of the discrete points corresponding to the relaxation time distribution function across the entire frequency band; The relaxation time distribution spectrum is generated based on the relative positions of the discrete distribution values ​​on the relaxation time axis.

4. The real-time detection method for lithium plating in lithium batteries for fast charging cycles according to claim 2, characterized in that, Determining at least one characteristic relaxation time peak in the relaxation time distribution spectrum specifically includes: The relaxation time peaks in the relaxation time distribution spectrum that are within the preset characteristic time constant range and whose peak intensity exceeds the preset intensity threshold are marked as the characteristic relaxation time peaks.

5. The real-time detection method for lithium plating in lithium batteries for fast charging cycles according to claim 1, characterized in that, The step of selecting feature frequency points corresponding to the initial frequency window from the detection sub-cycle to form a dynamic frequency set, and constructing a multi-sine superposition perturbation signal corresponding to the dynamic frequency set, specifically includes: A preset number of feature frequency points are selected in real time from the dynamic neighborhood of the initial frequency window corresponding to the detection sub-cycle to construct a dynamic frequency set; For each characteristic frequency point in the dynamic frequency set, a corresponding single-frequency sine wave signal is generated, and the single-frequency sine wave signals corresponding to each characteristic frequency point are linearly superimposed in the time domain to generate a composite time domain signal; The peak amplitude of the composite time-domain signal is adjusted to a preset injection amplitude range to obtain the multi-sine superimposed perturbation signal.

6. The real-time detection method for lithium plating in lithium batteries for fast charging cycles according to claim 5, characterized in that, The step of selecting a preset number of feature frequency points in real time from the dynamic neighborhood of the initial frequency window corresponding to the detection sub-cycle to construct a dynamic frequency set specifically includes: Taking the real-time peak position of the characteristic relaxation time peak at the previous detection time as the center, a preset neighborhood bandwidth is extended in both the low-frequency and high-frequency directions to form the dynamic neighborhood. A preset number of characteristic frequency points are selected in the dynamic neighborhood according to the logarithmic spacing method; The characteristic frequency points are sorted according to preset rules and the dynamic frequency set is constructed.

7. The real-time detection method for lithium plating in lithium batteries for fast charging cycles according to claim 1, characterized in that, The determination of the current peak position parameter of the characteristic relaxation time peak based on the impedance data corresponding to each characteristic frequency point specifically includes: Using the impedance data corresponding to the characteristic frequency point as input, the impedance data is inverted and fitted through a pre-built analytical basis function model library, and the local relaxation time distribution curve corresponding to the characteristic relaxation time peak is generated; wherein, the analytical basis functions in the analytical basis function model library include at least one of Cole-Cole type relaxation basis functions and Havriliak-Negami type relaxation basis functions. Extract the current peak position parameter of the characteristic relaxation time peak from the local relaxation time distribution curve.

8. The real-time detection method for lithium plating in lithium batteries for fast charging cycles according to claim 7, characterized in that, The process of inverting and fitting the impedance data using a pre-built analytical basis function model library to generate the local relaxation time distribution curve corresponding to the characteristic relaxation time peak specifically includes: A local relaxation time distribution function is constructed based on the impedance data. The local relaxation time distribution function is solved based on a preset coarse-resolution relaxation time grid. The coarse peak position of the characteristic relaxation time peak is identified from the local relaxation time distribution function. The local time window is determined with the coarse peak position as the center and a preset window expansion factor. An encrypted relaxation time grid is constructed within the local time window. The relaxation time distribution on the encrypted relaxation time grid is characterized as an objective function containing the analytical basis function. The objective function of the analytical basis function is solved based on Tikhonov regularization to determine the weight coefficients of the objective function. The relaxation time distribution curve within the local time window is reconstructed based on the weight coefficients to obtain the local relaxation time distribution curve.

9. The real-time detection method for lithium plating in lithium batteries oriented towards fast charging cycles according to any one of claims 1-8, characterized in that, As the number of fast charging cycles increases, the value of the drift threshold decreases.

10. A real-time lithium plating detection system for lithium batteries oriented towards fast charging cycles, characterized in that, It includes a processor and a memory; wherein the memory stores a computer program for being loaded by the processor and executed as described in any one of claims 1-9, a real-time lithium plating detection method for fast charging cycles.

Images