Battery cell fault prediction method

Abstract

The invention discloses a battery cell fault prediction method, which comprises the following steps: based on a full-quantity battery pack at a market end, performing electrochemical impedance spectroscopy reconstruction on a battery cell in the full-quantity battery pack to obtain electrochemical impedance spectroscopy data, and recording SOC and temperature corresponding to each battery pack; relaxation time distribution analysis is carried out on the electrochemical impedance spectroscopy data, a first parameter of a battery cell in each battery pack is obtained, and the first parameter comprises at least one relaxation time and impedance corresponding to each relaxation time; for each battery pack, based on the recorded SOC and temperature, performing normalization processing on the first parameter to obtain a second parameter, the second parameter comprising normalized impedance corresponding to each relaxation time; corresponding threshold values are set based on different fault types of the battery cell; and comparing the second parameter with the threshold value, and determining a fault type, a fault battery pack and a fault cell in the fault battery pack.

Description

Technical Field

[0001] The present invention relates to the technical field of batteries, and in particular to a method for predicting battery cell failures. Background Art

[0002] As the electrification of automobiles continues to deepen, the safety issues of electric vehicles are becoming increasingly prominent. Battery fault detection is particularly important. If faults can be effectively predicted and reported in a timely manner, the risk of battery failure and loss of control will be greatly reduced. At present, prediction methods mostly rely on sensors and other equipment, which will increase costs on the one hand, and sensors also have risks such as false alarms on the other hand.

[0003] Electrochemical impedance spectroscopy (EIS), as a non-destructive parameter determination and effective method for determining battery kinetic behavior, is widely used in the study of battery positive and negative electrode materials, lithium ion insertion and extraction kinetic parameters, (solidelectrolyte interphase, SEI) membranes, various electrolytes, and the evaluation and prediction of battery state of charge (SOC), state of health (SOH) and other status information. It is a powerful tool for analyzing lithium battery performance.

[0004] The parameter information of the lithium battery can be obtained by performing relaxation time distribution (DRT) analysis on the electrochemical impedance spectrum. The parameter information includes at least one relaxation time and the resistance corresponding to each relaxation time. When the parameter information of the battery exceeds the range corresponding to the normal operation of the battery, it is determined that the battery has a fault.

[0005] In patent CN114114047B, a battery fault detection method is proposed, which is based on electrochemical impedance spectroscopy and DRT analysis to predict battery faults. However, this patent is aimed at a single battery pack. It considers that the batteries in a battery pack are batteries of a batch, so it is considered that the aging degree of each battery is basically the same, and it is necessary to detect at the same SOC and temperature, so as to eliminate the influence of temperature, SOC, and battery aging on battery parameter information. And the prediction of its fault is first based on when the deviation between the first parameter information of the first battery in multiple batteries and the second parameter information of the second battery meets the preset fault condition, or when the deviation between the first parameter information and the second parameter information meets the preset fault condition, the second parameter information includes at least one relaxation time of the battery at the previous moment and the resistance corresponding to each relaxation time, so as to determine that at least one battery in the battery pack has a fault, and on this basis, the fault type is determined according to the deviation, and finally the severity of the fault is determined according to the degree of deviation from the preset parameter information. Its fault types include overcharge, over discharge, over temperature, internal short circuit, external short circuit, leakage, and extrusion deformation.

[0006] However, when many battery packs are installed on the vehicle or in a battery swap station, the operating conditions of each battery pack vary greatly, and the vehicle-side operating conditions are complex, with SOC and temperature constantly changing. Due to the complex use environment, different temperatures, SOCs, and battery aging must be considered in actual situations. If the above patent is used to predict whether a failure occurs by whether the deviation meets the preset fault conditions, many false alarms or inaccurate predictions will occur. For the complex actual use conditions, the above patent does not give an exact threshold, resulting in the inability to accurately identify batteries with potential failure risks through the above analysis method when facing battery packs (also called full battery packs) on market-side vehicles and / or in battery swap stations under actual use conditions.

[0007] Therefore, it is necessary to develop a method for non-destructive prediction of thermal runaway and abnormal expansion of battery cells, which can analyze the entire battery pack on the market, timely identify abnormal battery packs and abnormal cells, and provide early warning to prevent related failures from occurring. Summary of the invention

[0008] The present application provides a method for predicting battery cell failures to solve or at least improve the problems in the above-mentioned background technology.

[0009] The embodiment of the present application provides a method for detecting battery cell faults, comprising the following steps: (a) based on a full battery pack on the market, reconstructing the electrochemical impedance spectrum of the battery cells in the full battery pack to obtain electrochemical impedance spectrum data, and recording the SOC and temperature corresponding to each battery pack; (b) performing relaxation time distribution analysis on the electrochemical impedance spectrum data to obtain a first parameter of the battery cells in each battery pack, the first parameter including at least one relaxation time and an impedance corresponding to each relaxation time; (c) for each battery pack, based on the recorded SOC and temperature, normalizing the first parameter to obtain a second parameter, the second parameter including a normalized impedance corresponding to each relaxation time; (d) setting a corresponding threshold value based on different battery cell fault types; (e) comparing the second parameter with the threshold value to determine the fault type, the faulty battery pack, and the faulty battery cell in the faulty battery pack.

[0010] This application targets full battery packs on the market, and takes into account the impact of different SOCs, temperatures, and aging levels of battery packs on fault prediction results. On this basis, for different faults, unified thresholds applicable to full battery packs are set, and the normalized impedance information is directly compared with the threshold to predict whether a fault has occurred. It can be seen that compared with existing fault prediction methods, it expands the scope of application of fault prediction and simplifies the algorithm flow of the prediction method. It can be applied to fault prediction of full battery packs on the actual market based on cloud data collection.

[0011] Furthermore, in the step (c), the normalization processing includes: under the first test condition of controlling the temperature to be constant, testing the battery cell at different SOCs, and obtaining the first parameter of the battery cell at different SOCs based on the electrochemical impedance spectrum of the battery cell and performing relaxation time distribution analysis; determining the first functional relationship between the SOC and the first parameter based on a linear interpolation method; under the second test condition of controlling the SOC to be constant, testing the battery cell at different temperatures, and obtaining the first parameter of the battery cell at different temperatures based on the electrochemical impedance spectrum of the battery cell and performing relaxation time distribution analysis; determining the second functional relationship between temperature and the first parameter based on the Arrhenius equation; and normalizing the first parameter of the battery cell at the SOC and temperature corresponding to each battery pack in the step (c) to the second parameter at the first standard SOC and the first standard temperature based on the first functional relationship and the second functional relationship.

[0012] Furthermore, the second parameter at least includes charge transfer impedance and diffusion impedance, wherein the charge transfer impedance is the sum of the positive electrode charge transfer impedance and the negative electrode charge transfer impedance.

[0013] Furthermore, the fault types include thermal runaway and abnormal expansion of the battery cell.

[0014] Furthermore, in the step (d), the setting of the threshold corresponding to thermal runaway includes: conducting a harsh operating condition experiment on the battery cell to degrade the battery cell to a boundary state of thermal runaway; testing the battery cell at the first standard SOC and the first standard temperature, wherein the test includes obtaining a first relationship curve between different degrees of degradation and the charge transfer impedance and the diffusion impedance based on the electrochemical impedance spectrum of the battery cell and a relaxation time distribution analysis; and setting a first charge transfer impedance threshold and a first diffusion impedance threshold based on the first relationship curve.

[0015] Furthermore, the first charge transfer impedance threshold is 0.05 mΩ; and the first diffusion impedance threshold is 0.8 mΩ.

[0016] Furthermore, in the step (d), the threshold setting also includes: based on the redundancy of the thermal runaway fault detection method, setting a second charge transfer impedance threshold and a second diffusion impedance threshold, the second charge transfer impedance threshold is equal to N1*the first charge transfer impedance threshold, the second diffusion impedance threshold is equal to N2*the first diffusion impedance threshold, wherein N1 and N2 are both greater than 1.

[0017] Further, in the step (e), if the charge transfer impedance of the battery cell reaches or exceeds the first charge transfer impedance threshold but does not reach the second charge transfer impedance threshold; or, the diffusion impedance reaches or exceeds the first diffusion impedance threshold but does not reach the second diffusion impedance threshold, then the battery cell has a first thermal runaway risk and a first thermal runaway intervention operation is performed; if the charge transfer impedance of the battery cell reaches or exceeds the second charge transfer impedance threshold; or, the diffusion impedance reaches or exceeds the second diffusion impedance threshold, then the battery cell has a second thermal runaway risk and a second thermal runaway intervention operation is performed.

[0018] Furthermore, in the step (d), the setting of the threshold value corresponding to the abnormal expansion of the battery includes: performing a battery cell aging experiment on the battery cell; testing the battery cell at the first standard SOC and the first standard temperature, wherein the test includes obtaining a second relationship curve between the hard expansion force of the battery cell and the charge transfer impedance and the diffusion impedance based on the electrochemical impedance spectrum of the battery cell and a relaxation time distribution analysis; and setting a charge transfer impedance change threshold value of the battery cell within an aging cycle as well as a third diffusion impedance threshold value and a fourth diffusion impedance threshold value according to the second relationship curve.

[0019] Furthermore, the charge transfer impedance change threshold is 0.02 mΩ; the third diffusion impedance threshold is 0.5 mΩ; the fourth diffusion impedance threshold is 0.75 mΩ; and one aging cycle is defined as a battery pack throughput of 300 kWh.

[0020] Furthermore, within the aging cycle, if the change value of the charge transfer impedance of the battery cell does not exceed the charge transfer impedance change threshold, and the diffusion impedance reaches or exceeds the third diffusion impedance threshold but does not reach the fourth diffusion impedance threshold, the battery cell undergoes a first abnormal expansion, and a first intervention operation is performed; wherein the change value of the charge transfer impedance is defined as the charge transfer impedance at the end of the cycle minus the charge transfer impedance at the beginning of the cycle, taking the absolute value; if the change value of the charge transfer impedance of the battery cell does not exceed the charge transfer impedance change threshold, and the diffusion impedance reaches or exceeds the fourth diffusion impedance threshold, the battery cell undergoes a second abnormal expansion, and a second intervention operation is performed. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of a method for predicting battery cell failure according to an embodiment of the present application.

[0022] Figure 2 It is an example of the relaxation time distribution DRT spectrum of an embodiment of the present application.

[0023] FIG3( a ) is an Rct-SOH diagram of an embodiment of the present application, wherein the horizontal axis is SOH and the vertical axis is Rct.

[0024] FIG3( b ) is an Rcp-SOH diagram of an embodiment of the present application, wherein the horizontal axis is SOH and the vertical axis is Rcp.

[0025] Figure 4 It is a schematic diagram comparing the DRT curves of the same battery cell at the beginning and after degradation in the entire life cycle of the embodiment of the present application.

[0026] Figure 5 It is a schematic diagram of the relationship curve between Rct, Rcp and the growth of the hard expansion force of the battery cell during an aging cycle of an embodiment of the present application. DETAILED DESCRIPTION

[0027] Many specific details are set forth in the following description to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present invention, so the present invention is not limited to the specific implementation disclosed below.

[0028] This application provides a method for predicting battery cell failure. Figure 1 , including: step (a), based on the full battery pack on the market, reconstructing the electrochemical impedance spectroscopy (EIS) of the battery cells in the full battery pack to obtain electrochemical impedance spectroscopy data, and recording the SOC and temperature corresponding to each battery pack. The impedance spectroscopy data of the battery cell includes the impedance of the battery cell at multiple frequencies.

[0029] Among them, the full battery pack on the market side refers to the battery packs on all vehicle ends on the market side, and for swappable vehicles, it also includes the battery packs in the swap station. Since the working conditions of each battery pack are different, the SOC and temperature of each battery pack are also different. In step (a), the present application records the SOC and temperature corresponding to each battery pack, and the consistency of the battery cells in the same battery pack is verified during packaging, so the SOC and temperature corresponding to the battery pack are considered to be the SOC and temperature of the battery cells in the battery pack.

[0030] Step (b), performing a relaxation time distribution (DRT) analysis on the electrochemical impedance spectroscopy data to obtain a first parameter of a battery cell in each battery pack, wherein the first parameter includes at least one relaxation time and an impedance corresponding to each relaxation time.

[0031] Among them, the relaxation time distribution technology (DRT) is a fast visualization analysis method for EIS data. This method does not require the definition of a specific design of equivalent circuit, thereby minimizing the need for "a priori" assumptions in electrochemical system research, and can accurately provide basic information about each impedance inside the battery cell. DRT uses an infinite number of extremely small differential RC elements to represent impedance. By applying DRT technology, the g(τ) function can be obtained by deconvolution method through the following formula and the battery cell EIS impedance information. Each electrochemical process can be analyzed with a corresponding g(τ) function. This method can avoid the construction of an impedance model of any electrochemical system, thereby avoiding a series of a priori assumption problems caused by building an electrochemical model.

[0032]

[0033] Where R0 is the ohmic resistance of the electrochemical system, R pol is the parameter of polarization resistance, τ is the relaxation characteristic time, and g(τ) is the relaxation characteristic time distribution function. For the battery cell, there are multiple characteristic values ​​such as ohmic impedance, charge transfer impedance, lithium ion diffusion impedance, etc. It is difficult to obtain detailed information of all impedances by traditional methods. Using DRT technology, the g(τ) function corresponding to different impedances can be given based on different τ, such as Figure 2 As described above, different peak areas represent impedances at different frequencies.

[0034] In this embodiment, the relaxation time distribution curve of the battery cell after decoupling by DRT technology can be clearly distinguished into five peaks S1-S5, corresponding to the five impedance types inside the battery cell. Among them, the S1 peak area represents the ohmic internal resistance, the S2 peak area represents the SEI membrane impedance, the S3 peak area represents the positive electrode charge transfer impedance, the S4 peak area represents the negative electrode charge transfer impedance, and the S5 peak area represents the diffusion impedance Rcp, which represents the influence of the diffusion of lithium ions inside the electrode material particles and in the liquid on the electrical performance of the battery; among them, the sum of the S3 and S4 peak areas is called the charge transfer impedance Rct.

[0035] Step (c), for each battery pack, based on the recorded SOC and temperature, normalize the first parameter to obtain a second parameter, wherein the second parameter at least includes the normalized impedance corresponding to each relaxation time.

[0036] Step (d), setting corresponding thresholds based on different fault types of the battery cells.

[0037] Step (e), comparing the second parameter with the threshold value to determine the fault type, the faulty battery pack, and the faulty battery cell in the faulty battery pack.

[0038] In this application, since the full battery packs on the market are targeted, the SOC and temperature of different battery packs vary greatly. Therefore, when fault prediction is performed based on the impedance information obtained by DRT analysis, both SOC and temperature will affect the impedance, which will in turn affect the prediction results. Therefore, this application targets the full battery packs, and for the SOC and temperature of different battery packs, the first parameter is normalized in step (c) to obtain the second parameter.

[0039] Specifically, the normalization processing includes: under the first test condition of controlling the temperature to be constant, testing the battery cell at different SOCs, obtaining the first parameter of the battery cell at different SOCs based on the electrochemical impedance spectrum of the battery cell and performing relaxation time distribution analysis; and determining a first functional relationship between the SOC and the first parameter based on a linear interpolation method.

[0040] Under the second test condition of controlling the SOC unchanged, the battery cell is tested at different temperatures, and the first parameter of the battery cell at different temperatures is obtained based on the electrochemical impedance spectrum of the battery cell and relaxation time distribution analysis; based on the Arrhenius equation, the second functional relationship between temperature and the first parameter is determined.

[0041] Based on the first functional relationship and the second functional relationship, the first parameter of the battery cell at the SOC and temperature corresponding to each of the battery packs in the step (c) is normalized to the second parameter at the first standard SOC and the first standard temperature.

[0042] In a specific embodiment, an EIS test is performed on the battery cell at every 5% SOC (0% to 100% SOC) to obtain EIS impedance information of the battery cell at different SOCs.

[0043] The EIS test is performed on the battery cell at every 10°C (-10°C to 40°C) to obtain the EIS impedance information of the battery cell at different temperatures.

[0044] DRT analysis is used to decompose the above EIS data. According to the various impedance parameters obtained by DRT analysis at different SOCs and different temperatures, the first functional relationship G(SOC) between SOC and various impedance parameters (i.e., the first parameter) and the functional relationship F(1 / T) between temperature and various impedance parameters (i.e., the first parameter) are established based on the linear interpolation method and the Arrhenius equation.

[0045] Using the functions F(1 / T) and G(SOC), each impedance parameter obtained by DRT analysis at different temperatures and different SOCs, specifically S1 to S5 in this embodiment, is normalized to the impedance parameters at 80% SOC and 25°C.

[0046] So far, this application has taken into account the influence of different SOCs and different temperatures for the full range of battery packs on the market, and normalized the impedance information at different SOCs and different temperatures to the impedance information at the first standard SOC and the first standard temperature through a normalization method, thereby facilitating the use of these impedance information for fault prediction.

[0047] In one embodiment of the present application, the above method is used to predict thermal runaway.

[0048] Based on the mechanism analysis, the charge transfer impedance Rct and the diffusion impedance Rcp are significantly correlated with the thermal stability of the battery cell, as shown in Figure 3(a) and Figure 3(b), where the horizontal axis is SOH, the vertical axis is charge transfer impedance Rct, and the diffusion impedance Rcp, respectively. SOH reflects the degree of degradation of the battery cell. As the battery cell deteriorates, the charge transfer impedance Rct and the diffusion impedance Rcp have a significant increasing trend, and the charge transfer impedance Rct is linearly related to the degree of degradation of the battery cell. The diffusion impedance Rcp has an obvious jump point, which proves that the internal degradation of the battery cell is aggravated, the thermal stability of the battery cell is significantly reduced, and there is a risk of thermal runaway.

[0049] Figure 4 A comparison of the DRT curves of the same battery cell at the beginning of its life cycle (battery cell-BOL) and after degradation (battery cell-EOL) is given. It can be found that the charge transfer impedance Rct corresponding to the sum of the peak areas of S3 and S4 increases relatively linearly without a jump point, while the diffusion impedance Rcp has jumped and increased significantly, proving that the thermal stability of the battery cell has been significantly reduced and there is a greater risk of thermal runaway.

[0050] Therefore, based on the above mechanism analysis, the threshold can be set according to the relationship curve between different degradation degrees and the charge transfer impedance Rct and diffusion impedance Rcp, and the thermal runaway of the battery cell can be predicted based on the relationship between the charge transfer impedance Rct and diffusion impedance Rcp of the battery cell and the threshold.

[0051] Specifically, in the step (d), the setting of the threshold corresponding to thermal runaway includes: conducting a harsh operating condition experiment on the battery cell to degrade the battery cell to a boundary state of thermal runaway; exemplarily, the harsh operating condition experiment includes high-temperature storage, large current charging and discharging, local overtemperature, local high pressure, etc.; testing the battery cell at the first standard SOC and the first standard temperature, the test including obtaining a first relationship curve between different degrees of degradation and the charge transfer impedance Rct and the diffusion impedance Rcp based on the electrochemical impedance spectrum of the battery cell and performing a relaxation time distribution analysis; and setting a first charge transfer impedance threshold and a first diffusion impedance threshold based on the first relationship curve.

[0052] In a specific embodiment of the present application, setting the first charge transfer impedance threshold and the first diffusion impedance threshold based on the first relationship curve includes: using an adiabatic accelerating calorimeter ARC to obtain the T1 temperature of the battery after degradation, wherein the T1 temperature is the starting temperature of the battery cell self-heating. Under normal circumstances, the T1 temperature is directly related to the thermal stability of the battery cell. The lower the T1 temperature, the worse the thermal stability of the battery cell. After the T1 temperature drops to a certain level, the thermal stability of the battery cell drops significantly and is on the edge of thermal runaway; associating the corresponding relationship between the T1 temperature drop and the diffusion impedance Rcp jump point, setting the value after the diffusion impedance Rcp jump point as the first diffusion impedance threshold, and performing rollback verification based on the severe working condition experimental data and the actual thermal runaway data of the full battery pack, setting the first charge transfer impedance threshold and the first diffusion impedance threshold. Of course, it can be understood that when the severe working condition experimental data and the actual thermal runaway data of the full battery pack are rolled back for verification, the actual thermal runaway data of the full battery pack is also normalized to the data at the first standard SOC and the first standard temperature.

[0053] So far, through the severe operating condition experiment, the first charge transfer impedance threshold and the first diffusion impedance threshold at the first standard SOC and the first standard temperature are obtained.

[0054] In a specific implementation of the present application, the first charge transfer impedance threshold is 0.05 mΩ; the first diffusion impedance threshold is 0.8 mΩ.

[0055] Furthermore, considering the individual subtle differences of each battery pack and setting the algorithm redundancy, the detection of the thermal runaway fault also includes setting a second charge transfer impedance threshold and a second diffusion impedance threshold, wherein the second charge transfer impedance threshold is equal to N1*the first charge transfer impedance threshold, and the second diffusion impedance threshold is equal to N2*the first diffusion impedance threshold, wherein N1 and N2 are algorithm redundancies, and their values ​​are both greater than 1. Exemplarily, N1 is 1.2, that is, for the setting of the charge transfer impedance threshold, the algorithm redundancy is set to 20%; N1 is 1.1, that is, for the setting of the diffusion impedance threshold, the algorithm redundancy is set to 10%.

[0056] Further, in step (e), if the charge transfer impedance Rct of a certain battery cell reaches or exceeds the first charge transfer impedance threshold, but does not reach the second charge transfer impedance threshold; or, the diffusion impedance Rcp reaches or exceeds the first diffusion impedance threshold, but does not reach the second diffusion impedance threshold, that is, the first charge transfer impedance threshold ≤ charge transfer impedance Rct < second charge transfer impedance threshold, or the first diffusion impedance threshold ≤ diffusion impedance Rcp < second diffusion impedance threshold, then the battery cell has a first thermal runaway risk, and a first thermal runaway intervention operation is performed. Exemplarily, the first thermal runaway intervention operation includes marking the battery cell and the corresponding battery pack as a risky battery pack, and performing manual intervention in the cloud in a timely manner: including adjusting the available cycle depth of the battery pack (i.e., locking the battery in the usual sense), reducing the charging current, reducing the discharge power, and other measures.

[0057] If the charge transfer impedance Rct reaches or exceeds the second charge transfer impedance threshold; or the diffusion impedance Rcp reaches or exceeds the second diffusion impedance threshold, that is, the charge transfer impedance Rct ≥ the second charge transfer impedance threshold, or the diffusion impedance Rcp ≥ the second diffusion impedance threshold, then the battery cell has a second thermal runaway risk and performs a second thermal runaway intervention operation. Exemplarily, the second thermal runaway intervention operation includes: the vehicle side issues a warning to the user in a timely manner, initiates a battery pack recall, prohibits the battery pack from charging and discharging, and other measures.

[0058] In another embodiment of the present application, the above method is used to predict abnormal expansion of a battery cell.

[0059] The mechanism analysis is conducted on the correlation between the hard expansion of the battery cell and the DRT analytical impedance. If the battery cell in the battery pack expands hard, the battery cells squeeze each other, and the expansion force between the large surfaces of the battery cells increases significantly. The internal stress of the battery cell accumulates, and the accumulated stress will be applied to the internal pole piece, resulting in the diaphragm closed hole, insufficient electrolyte between the pole pieces, and causing changes in the battery cell impedance information.

[0060] Mechanism analysis proves that the change of expansion force between cells will significantly affect the peak intensity of S5 peak representing the diffusion impedance Rcp of the cell, while it has little effect on S3 and S4 peaks representing the charge transfer impedance Rct of the cell. It proves that the expansion force on the cell has a significant correlation with the diffusion impedance Rcp of the cell. Based on the analysis of the change of S5 peak, it proves that the expansion force on the cell is positively correlated with the diffusion impedance Rcp of the cell. This positive correlation is mainly attributed to the squeezing of the cell by the expansion force, which causes the internal core to deform, the diaphragm is compressed and closed, and the electrolyte between the pole pieces is gradually squeezed out, resulting in insufficient electrolyte between the pole pieces and increased diffusion impedance Rcp. Therefore, there is a certain correlation between the abnormal expansion force between the cells and the diffusion impedance Rcp of the cells. Therefore, the present invention reasonably decouples the internal impedances of the cell through electrochemical impedance spectroscopy EIS and relaxation time distribution technology, and monitors and predicts the cell with abnormal expansion force based on the relationship between the expansion force and the diffusion impedance Rcp.

[0061] Specifically, in the step (d), the setting of the threshold value corresponding to the abnormal expansion of the battery includes: performing a battery cell aging experiment on the battery cell; exemplarily, the aging experiment includes: normal temperature fast charging cycle, high temperature fast charging cycle, high and low temperature storage, etc.; the battery cell is tested at the first standard SOC and the first standard temperature, and the test includes obtaining a second relationship curve between the hard expansion force of the battery cell and the charge transfer impedance Rct and the diffusion impedance Rcp based on the electrochemical impedance spectrum of the battery cell and relaxation time distribution analysis; according to the second relationship curve, the charge transfer impedance change threshold of the battery cell within an aging cycle and the third diffusion impedance threshold and the fourth diffusion impedance threshold are set.

[0062] So far, through the cell aging experiment, the charge transfer impedance change threshold of the cell in an aging cycle at the first standard SOC and the first standard temperature, as well as the third diffusion impedance threshold and the fourth diffusion impedance threshold are obtained.

[0063] In a specific embodiment of the present application, based on the cell level test, a curve of the relationship between the cell charge transfer impedance Rct, the diffusion impedance Rcp and the cell hard expansion force growth is established, such as Figure 5As shown, it can be found that the hard expansion characteristics of the battery cell are obvious, the charge transfer impedance Rct of the battery cell has not changed significantly, and the diffusion impedance Rcp rises sharply after a plateau period; the appearance of the diffusion impedance Rcp platform indicates that the core inside the battery cell has undergone hard expansion, but due to the battery cell design, the core and the outer layer of the battery cell, such as the steel shell, aluminum-plastic film, and aluminum shell, leave a part of the expansion space, and the core does not directly contact the battery cell shell, so the growth of the diffusion impedance Rcp slows down. When the core expands to a certain extent, the core begins to contact the battery cell shell. Due to the rigid constraint of the shell, the hard expansion of the core is limited, and the internal stress continues to accumulate, resulting in a significant increase in the diffusion impedance Rcp of the battery cell, and the growth trend is significantly accelerated. It is confirmed that the battery cell has undergone obvious hard expansion, and the internal stress of the battery cell has accumulated to a certain extent. The battery cell is risky. Based on the relationship curve between the charge transfer impedance Rct, the diffusion impedance Rcp and the hard expansion force growth of the battery cell, the charge transfer impedance change threshold of the battery cell within an aging cycle, as well as the third diffusion impedance threshold and the fourth diffusion impedance threshold are set, where the value of the diffusion impedance Rcp after the first abnormal growth is the third diffusion impedance threshold ( Figure 5 The value of the diffusion impedance Rcp after the second abnormal increase is the fourth diffusion impedance threshold ( Figure 5 The threshold value is shown in b).

[0064] In a specific implementation of the present application, within an aging cycle, the charge transfer impedance change threshold is set to 0.02mΩ, the third diffusion impedance threshold is 0.5mΩ, and the fourth diffusion impedance threshold is 0.75mΩ; the aging cycle is defined as a battery pack throughput of 300kWh. The throughput of the battery pack refers to the amount of electricity successfully transmitted within a certain period of time, including the input and output power of the battery pack. Of course, in some other embodiments, an aging cycle can be measured in time, such as a week, a month, a quarter, or a year.

[0065] Further, in step (e), within an aging cycle, if the change value ΔRct of the charge transfer impedance of the battery cell does not exceed the charge transfer impedance change threshold, that is, the charge transfer impedance Rct does not change significantly within an aging cycle; and the diffusion impedance Rcp reaches or exceeds the third diffusion impedance threshold, but does not reach the fourth diffusion impedance threshold, that is, the change value ΔRct of the charge transfer impedance ≤ the charge transfer impedance change threshold, and the third diffusion impedance threshold ≤ diffusion impedance Rcp < the fourth diffusion impedance threshold, then the battery cell undergoes a first abnormal expansion and a first intervention operation is performed. Among them, the change value ΔRct of the charge transfer impedance is defined as the charge transfer impedance at the end of the cycle minus the charge transfer impedance at the beginning of the cycle, and then the absolute value is taken, that is, ΔRct = |Rct (end of cycle) - Rct (beginning of cycle) |. Exemplarily, the occurrence of the first abnormal expansion proves that the internal winding core of the battery cell has begun to reach the early stage of hard expansion. At this time, attention should be paid to the battery cell status, abnormal batteries and corresponding battery packs should be marked in the cloud, fast charging of the battery pack should be restricted, and changes in the battery cell charge transfer impedance Rct and diffusion impedance Rcp should be continuously recorded.

[0066] If the change value ΔRct of the charge transfer impedance of the battery cell does not exceed the charge transfer impedance change threshold, and the diffusion impedance Rcp reaches or exceeds the fourth diffusion impedance threshold, that is, the change value ΔRct of the charge transfer impedance ≤ the charge transfer impedance change threshold, and the diffusion impedance Rcp ≥ the fourth diffusion impedance threshold, the battery cell undergoes a second abnormal expansion and a second intervention operation is performed. Exemplarily, the occurrence of the second abnormal expansion proves that the internal winding core of the battery cell has undergone a relatively serious hard expansion and needs to be listed as a risky battery cell, including manual intervention, limiting the charging current, and even initiating a battery pack recall.

[0067] Of course, it is understandable that in addition to the above situations, during an aging cycle, the change value ΔRct of the charge transfer impedance may exceed the charge transfer impedance change threshold. This situation belongs to abnormal aging of the battery cell, and the cause of the fault can be analyzed and corresponding intervention measures can be implemented through other means or methods, but it does not fall within the scope of specific discussion of this application.

[0068] Finally, the fault prediction method of the present application is applicable to different types of battery cells, such as square shell battery cells, cylindrical battery cells, soft pack battery cells, etc.

[0069] This application targets full battery packs on the market, and takes into account the impact of different SOCs, temperatures, and aging levels of battery packs on fault prediction results. On this basis, for different faults, unified thresholds applicable to full battery packs are set, and the normalized impedance information is directly compared with the threshold to predict whether a fault has occurred. In particular, the fault is thermal runaway and abnormal expansion of the battery cell. It can be seen that compared with existing fault prediction methods, it expands the scope of application of fault prediction and simplifies the algorithm flow of the prediction method. It can be applied to fault prediction of full battery packs on the actual market based on cloud data collection.

[0070] It should be noted that although the present invention is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be based on the scope defined by the claims of the present invention.

Claims

1. A method for predicting battery cell failures, comprising the following steps: (a) Based on a full battery pack on the market, reconstruct the electrochemical impedance spectrum of the cells in the full battery pack to obtain electrochemical impedance spectrum data, and record the SOC and temperature corresponding to each battery pack; (b) performing relaxation time distribution analysis on the electrochemical impedance spectroscopy data to obtain a first parameter of a battery cell in each battery pack, wherein the first parameter includes at least one relaxation time and an impedance corresponding to each relaxation time; (c) for each battery pack, based on the recorded SOC and temperature, normalizing the first parameter to obtain a second parameter, wherein the second parameter includes a normalized impedance corresponding to each relaxation time; (d) Setting corresponding thresholds based on different cell fault types; (e) comparing the second parameter with the threshold value to determine the fault type, the faulty battery pack, and the faulty battery cell in the faulty battery pack.

2. The method for predicting battery cell failure according to claim 1, characterized in that: In the step (c), the normalization process includes: Under the first test condition of controlling the temperature to be constant, the battery cell is tested at different SOCs, and the first parameter of the battery cell at different SOCs is obtained based on the electrochemical impedance spectrum of the battery cell and the relaxation time distribution analysis; based on the linear interpolation method, a first functional relationship between the SOC and the first parameter is determined; Under the second test condition of controlling the SOC to be constant, the battery cell is tested at different temperatures, and the first parameter of the battery cell at different temperatures is obtained based on the electrochemical impedance spectrum of the battery cell and the relaxation time distribution analysis; based on the Arrhenius equation, a second functional relationship between the temperature and the first parameter is determined; Based on the first functional relationship and the second functional relationship, the first parameter of the battery cell at the SOC and temperature corresponding to each of the battery packs in the step (c) is normalized to the second parameter at the first standard SOC and the first standard temperature.

3. The method for predicting battery cell failure according to claim 1 or 2, characterized in that: The second parameter at least includes charge transfer impedance and diffusion impedance, wherein the charge transfer impedance is the sum of the positive electrode charge transfer impedance and the negative electrode charge transfer impedance.

4. The method for predicting battery cell failure according to claim 3, characterized in that: The failure types include thermal runaway and abnormal expansion of battery cells.

5. The method for predicting battery cell failure according to claim 4, characterized in that: In the step (d), the setting of the threshold corresponding to thermal runaway includes: Performing a severe operating condition test on the battery cell to degrade the battery cell to a boundary state of thermal runaway; Testing the battery cell at the first standard SOC and the first standard temperature, wherein the testing includes performing a relaxation time distribution analysis based on an electrochemical impedance spectrum of the battery cell to obtain a first relationship curve between different degradation degrees and the charge transfer impedance and the diffusion impedance; A first charge transfer impedance threshold and a first diffusion impedance threshold are set based on the first relationship curve.

6. The method for predicting battery cell failure according to claim 5, characterized in that: The first charge transfer impedance threshold is 0.05 mΩ; the first diffusion impedance threshold is 0.8 mΩ.

7. The method for predicting battery cell failure according to claim 5 or 6, characterized in that: In the step (d), the threshold setting also includes: based on the redundancy of the thermal runaway fault detection method, setting a second charge transfer impedance threshold and a second diffusion impedance threshold, the second charge transfer impedance threshold is equal to N1*first charge transfer impedance threshold, the second diffusion impedance threshold is equal to N2*first diffusion impedance threshold, wherein N1 and N2 are both greater than 1.

8. The method for predicting battery cell failure according to claim 7, characterized in that: In the step (e), if the charge transfer impedance of the battery cell reaches or exceeds the first charge transfer impedance threshold but does not reach the second charge transfer impedance threshold; or, if the diffusion impedance reaches or exceeds the first diffusion impedance threshold but does not reach the second diffusion impedance threshold, then the battery cell has a first thermal runaway risk, and a first thermal runaway intervention operation is performed; If the charge transfer impedance of the battery cell reaches or exceeds the second charge transfer impedance threshold; or the diffusion impedance reaches or exceeds the second diffusion impedance threshold, the battery cell has a second thermal runaway risk, and a second thermal runaway intervention operation is performed.

9. The method for predicting battery cell failure according to claim 4, characterized in that: In the step (d), the setting of the threshold corresponding to the abnormal battery expansion includes: Performing a cell aging test on the cell; Testing the battery cell at the first standard SOC and the first standard temperature, wherein the testing includes performing a relaxation time distribution analysis based on an electrochemical impedance spectrum of the battery cell to obtain a second relationship curve between the hard expansion force of the battery cell and the charge transfer impedance and the diffusion impedance; According to the second relationship curve, a charge transfer impedance change threshold value, a third diffusion impedance threshold value, and a fourth diffusion impedance threshold value of the battery cell within an aging cycle are set.

10. The method for predicting battery cell failure according to claim 9, characterized in that: The charge transfer impedance change threshold is 0.02mΩ; the third diffusion impedance threshold is 0.5mΩ; the fourth diffusion impedance threshold is 0.75mΩ; and one aging cycle is defined as a battery pack throughput of 300kWh.

11. The method for predicting battery cell failure according to claim 9 or 10, characterized in that: In the one aging cycle, if the change value of the charge transfer impedance of the battery cell does not exceed the charge transfer impedance change threshold, and the diffusion impedance reaches or exceeds the third diffusion impedance threshold but does not reach the fourth diffusion impedance threshold, the battery cell undergoes a first abnormal expansion and a first intervention operation is performed; wherein the change value of the charge transfer impedance is defined as the charge transfer impedance at the end of the cycle minus the charge transfer impedance at the beginning of the cycle, taking the absolute value; If the change value of the charge transfer impedance of the battery cell does not exceed the charge transfer impedance change threshold, and the diffusion impedance reaches or exceeds the fourth diffusion impedance threshold, the battery cell undergoes a second abnormal expansion, and a second intervention operation is performed.

Images