A battery electrochemical impedance spectrum-based active thermal runaway early warning and inhibition method

Abstract

The application discloses a kind of active thermal runaway early warning and inhibition method based on battery electrochemical impedance spectrum, specifically related to power battery safety monitoring and thermal runaway prevention and control technical field, obtain the reference electrochemical impedance spectrum under normal service state of battery to be measured, and simultaneously record state of charge, temperature state and connection position;At the end of static charging, parking standby or energy storage cabinet power supply gap applies disturbance signal, obtains real-time electrochemical impedance spectrum, and obtains comparable impedance spectrum by state correction;Again, the comparable impedance spectrum is compared with the reference electrochemical impedance spectrum, extracts high-frequency contact offset feature, mid-frequency interface passivation feature and low-frequency diffusion lag feature, forms hidden heat release pre-state feature;Subsequently, the connection position is combined to determine the electrical connection propagation path and the thermal diffusion propagation path, and the thermal runaway risk state is established, and the application can complete early warning and active inhibition before temperature rise and pressure difference anomaly appear.

Description

Technical Field

[0001] This invention relates to the field of power battery safety monitoring and thermal runaway prevention technology, specifically to an active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy. Background Technology

[0002] In low-power scenarios such as slow charging at night in underground parking garages and long-term standby charging in energy storage cabinets, poor soldering of individual cell tabs, localized wrinkling of the separator, or trace amounts of lithium plating typically do not immediately cause voltage drops and shell temperature rises, but their internal interface impedance has already undergone irreversible shifts. This type of shift is difficult to detect in a timely manner when conventional temperature sensors are spaced far apart and sampling points are far from the defect points. Especially when the vehicle is at the end of a stationary charging phase, the current is low and heat accumulates slowly. Abnormal cells may form localized hotspots during passive equalization or when the vehicle is restarted, inducing a chain reaction of thermal runaway. Existing early warning systems mostly rely on temperature, voltage thresholds, or smoke signals, often only acting after side reactions have already generated significant heat, failing to proactively reduce the intensity of electrochemical reactions in the defective area and block heat diffusion in the early stages of thermal runaway.

[0003] Therefore, how to utilize the interfacial anomalies in the battery electrochemical impedance spectroscopy that appear before the temperature rise to identify rare but seriously harmful hidden internal short circuit pre-states and implement corresponding suppression measures has become an urgent technical problem to be solved. Summary of the Invention

[0004] The purpose of this invention is to provide an active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy, so as to overcome the shortcomings of the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy, comprising:

[0006] S100: Obtain the reference electrochemical impedance spectroscopy of the battery under test under normal service conditions, and simultaneously record the state of charge, temperature state and connection position corresponding to the reference electrochemical impedance spectroscopy.

[0007] S200: When the battery under test is at the end of static charging, in standby mode, or in the energy storage cabinet charging gap, a perturbation signal is applied to the battery under test to obtain a real-time electrochemical impedance spectrum. The real-time electrochemical impedance spectrum is then corrected according to the state of charge and temperature to obtain a comparable impedance spectrum.

[0008] S300, the comparable impedance spectrum is compared with the reference electrochemical impedance spectrum in segments, and high-frequency contact offset features, mid-frequency interface passivation features and low-frequency diffusion hysteresis features are extracted to obtain the hidden exothermic pre-state features of the battery under test.

[0009] S400, based on the concealed heat release pre-state characteristics and the connection position of the battery under test, determine the electrical connection propagation path and thermal diffusion propagation path between the abnormal cell and the adjacent cell, and establish the corresponding thermal runaway initiation risk state.

[0010] S500 generates an active suppression strategy based on the risk state of thermal runaway initiation and performs different operations in order of increasing risk to obtain the suppressed battery state.

[0011] S600: Obtain the electrochemical impedance spectroscopy again for the suppressed battery state, and return the obtained electrochemical impedance spectroscopy to step S300 for verification. If the hidden exothermic pre-state characteristics are weakened, the suppression strategy is maintained. If the hidden exothermic pre-state characteristics are enhanced, a thermal runaway warning is output and a forced power-off isolation is performed.

[0012] Preferably, the acquisition of the reference electrochemical impedance spectroscopy includes:

[0013] After the battery under test has been assembled, maintained and reset, put into operation for the first time or completed a full charge and discharge calibration, it is confirmed that the battery under test has not experienced over-temperature, over-voltage, under-voltage, insulation abnormality, smoke alarm, short circuit protection action, or forced power-off record.

[0014] The connection position is obtained by reading the battery number, module, branch, series-parallel relationship, positive and negative electrode connection direction, adjacent cell position, adjacent position of cooling channel and adjacent position of casing edge of the battery under test;

[0015] An AC perturbation was applied under conditions where the charged state and temperature state were recorded simultaneously, and the voltage and current responses were collected to obtain the initial electrochemical impedance spectrum.

[0016] Isolated jump points caused by relay activation, charging ripple, communication interruption, and external load abrupt changes were eliminated. When the trend of spectral changes was consistent across multiple acquisitions, the initial electrochemical impedance spectrum was determined as the reference electrochemical impedance spectrum.

[0017] Preferably, obtaining comparable impedance spectra includes:

[0018] At the end of static charging, during parking standby, or during the energy storage cabinet recharging interval, the current state of charge and current temperature of the battery under test are read, and a reference electrochemical impedance spectrum with an absolute value of the state of charge difference not exceeding 3% and an absolute value of the temperature difference not exceeding 2 degrees Celsius is selected from the reference data archive as the corresponding object.

[0019] If no corresponding object exists, the average charging current is calculated every second within the 60-second continuous running record. When the absolute value of the difference between the average values ​​of adjacent seconds does not exceed 0.2 amperes, the equalization switch remains open, and the change in cooling pump speed does not exceed 50 revolutions per minute, the 20th to 50th seconds are determined as the time slice that can be collected.

[0020] A perturbation signal is applied within the available acquisition time slice to acquire the real-time electrochemical impedance spectrum of the battery under test. The spectrum is then corrected based on the current state of charge, the current temperature, and the corresponding object to obtain a comparable impedance spectrum.

[0021] Preferably, the disturbance signal is an embedded disturbance signal, and the formation of the embedded disturbance signal includes:

[0022] Adjacent cells are determined based on the connection position. When the battery under test is located in the middle of a series branch, the preceding and following cells are selected. When the battery under test is located at the end of a branch, the nearest cell connected to the same busbar and the nearest cell on the same cooling plate path are selected.

[0023] Read the instantaneous voltage sequence of adjacent cells within the first 10 seconds of the acquisition time slice, remove the average value from the instantaneous voltage sequence and perform frequency domain processing to obtain the external ripple main frequency;

[0024] Delete the frequency points that differ from the main frequency of the external ripple by no more than 8% from the preset impedance acquisition frequency point table to obtain the wrong frequency point table;

[0025] The sinusoidal current components are arranged sequentially according to the frequency point table, with each frequency point lasting for 3 complete cycles. A 0.5-second blank segment is set between different frequency points, and the amplitude of the disturbance current is set to the lower of the rated capacity current of 0.005 times the current and 0.5 amperes.

[0026] Preferably, obtaining real-time electrochemical impedance spectroscopy includes:

[0027] During each frequency misalignment period, the voltage response, current response, and passive response of the battery under test, as well as the adjacent cells, are simultaneously acquired.

[0028] The voltage response, current response, and passive response of the battery under test and adjacent cells are extracted by phase-locked loop at the same frequency to obtain the complex voltage response, complex current response, and complex passive response of the battery under test and adjacent cells.

[0029] The average result of the complex passive response of two adjacent cells is used as the common mode background. When there is only one adjacent cell in the battery under test, the complex passive response of the adjacent cell is used as the common mode background.

[0030] The bulk voltage response is obtained by subtracting the common-mode background from the complex voltage response of the battery under test, and the real-time electrochemical impedance spectroscopy is obtained from the ratio of the bulk voltage response to the complex current response.

[0031] Preferably, the correction of the real-time electrochemical impedance spectroscopy based on the state of charge and temperature includes:

[0032] Select two sets of reference data adjacent to the current state of charge and two sets of reference data adjacent to the current temperature from the reference data archive.

[0033] Based on the impedance difference and state of charge difference between two adjacent sets of reference data under the same temperature condition, the state of charge correction amount is obtained.

[0034] Based on the impedance difference and temperature state difference between two adjacent sets of reference data under the same charging state, the temperature state correction amount is obtained;

[0035] By subtracting the state of charge correction and temperature state correction from the real-time electrochemical impedance spectroscopy, a comparable impedance spectrum is obtained that retains only the shift of the battery under test relative to its own healthy state.

[0036] Preferably, obtaining the concealed exothermic pre-state characteristics includes:

[0037] Unify the comparable impedance spectra with the reference electrochemical impedance spectra to the same set of frequency points;

[0038] Within the reference electrochemical impedance spectrum, the contact response transition is determined based on the position where the slope difference of the real part of adjacent frequency points changes in the same direction twice. The interface response arc apex is determined based on the frequency point where the absolute value of the imaginary part is the largest after the contact response transition. The diffusion tailing start position is determined based on the position where the phase difference of the three consecutive phases after the interface response arc apex maintains the same sign.

[0039] An adaptive high-frequency band is formed at frequencies above the contact response transition point, an adaptive mid-frequency band is formed at frequencies from the contact response transition point to the start of the diffusion tail, and an adaptive low-frequency band is formed at frequencies below the start of the diffusion tail.

[0040] Segmented comparisons were performed in the adaptive high-frequency band, adaptive mid-frequency band, and adaptive low-frequency band, respectively.

[0041] Preferably, the segmented comparison includes the following steps:

[0042] Within the adaptive high-frequency band, high-frequency contact offset features are extracted based on the real part offset direction of the starting frequency, the phase swing order, and the continuity of adjacent frequency points, and isolated offsets that do not change in the same direction at the preceding and following frequency points are eliminated.

[0043] Within the adaptive mid-frequency band, the low-frequency side movement, half-width variation, and apex retraction state of the interface response arc apex are tracked along the offset direction of the high-frequency contact offset feature. The mid-frequency interface passivation features are extracted when the real offset direction and the imaginary offset direction are consistent at the frequency band junction.

[0044] In the adaptive low-frequency band, the low-frequency tail extension state is tracked from the starting position of the diffusion tail, and the low-frequency diffusion hysteresis feature is extracted when the real part offset direction of at least 4 consecutive frequency points is consistent with the real part offset direction of the mid-frequency interface passivation feature at the junction and the imaginary part offset direction is not reversed.

[0045] The high-frequency contact offset feature, mid-frequency interface passivation feature, and low-frequency diffusion hysteresis feature that appear continuously and whose main offset direction is not reversed are combined into a concealed exothermic pre-state feature.

[0046] Preferably, the establishment of a thermal runaway initiation risk state includes:

[0047] The starting offset direction of the spectral line is obtained by summing the real offsets of the first three frequency points of the adaptive high-frequency band, and the continuous direction at the end of the frequency band is obtained by summing the real offsets of the last three frequency points of the adaptive high-frequency band. The anomaly starting location is determined by combining the phase swing sequence.

[0048] Taking the abnormal starting position as the entry point, the positive electrode tab welding point, negative electrode tab welding point, bus connection end, series and parallel branches, and the end of the adjacent single unit directly connected to the bus are identified as candidate electrical connection paths, and the electrical connection propagation path is obtained according to the continuous direction at the end of the frequency band.

[0049] The direction of the summation of the real offsets of the last two frequency points of the adaptive mid-frequency band and the summation of the real offsets of the first two frequency points of the adaptive low-frequency band is verified. When both are consistent with the continuous direction at the end of the frequency band, the adjacent cell that shares the cooling plate, shell pressing surface or clamping bracket with the battery under test is determined as the heat diffusion propagation path.

[0050] The adjacent cell numbers in the electrical connection propagation path are cross-superimposed with the adjacent cell numbers in the thermal diffusion propagation path, and based on this, a thermal runaway initiation risk state of conduction superposition, thermal runaway initiation risk state of electrical connection traction, or thermal accumulation hysteresis type is established.

[0051] Preferred active suppression strategies and verification include:

[0052] When the risk state of thermal runaway initiation is electrical connection traction type, the charging current of the branch where the battery under test is located is reduced to 50% of the current charging current and maintained for 180 seconds. At the same time, the passive equalization of the battery under test and adjacent cells on the electrical connection propagation path is suspended for no less than 300 seconds.

[0053] When the risk state of thermal runaway initiation is the thermal accumulation hysteresis type, the liquid cooling flow rate or air cooling flow rate in the area corresponding to the thermal diffusion propagation path is adjusted to 120% of the rated value and maintained for 300 seconds, and the cooling range covers the battery under test and the first adjacent cell on the thermal diffusion propagation path.

[0054] When the risk state of thermal runaway initiation is the conduction superposition type, the equalization operation of adjacent cells obtained by cross superposition is turned off and their participation in charging is prohibited. Then the charging input of the abnormal branch where the battery under test is located is cut off.

[0055] After suppression, the electrochemical impedance spectroscopy is acquired again and verified. When the verified hidden exothermic pre-state characteristics are weakened, the active suppression strategy is maintained. When the verified hidden exothermic pre-state characteristics are enhanced, a thermal runaway warning is output, and the charging input is stopped in sequence, the positive terminal relay of the branch is disconnected, and the negative terminal relay of the branch is disconnected after 2 seconds. At the same time, the corresponding area of ​​the heat diffusion propagation path is kept cooled for 600 seconds.

[0056] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0057] This invention differs from existing passive early warning methods that rely on temperature, voltage, or smoke signals. Its core lies in first establishing an individualized reference electrochemical impedance spectroscopy for the battery under test, binding its state of charge, temperature, and connection location. Then, it applies a frequency-misaligned embedded perturbation at the end of static charging, during standby operation, or during the energy storage cabinet recharging interval. Furthermore, it utilizes the passive response of adjacent cells to strip away bus ripple, cooling fluctuations, and environmental drift. This solves the problems of heat not yet being apparent in low-current scenarios, conventional sensors being unable to detect loose electrode connections, localized lithium plating, or micro-damage to the separator. The acquired comparable impedance spectra directly reflect abnormal changes within the battery itself, reducing misjudgments caused by differences in normal operating conditions.

[0058] This invention further differs from schemes that only judge a single impedance change. Its core lies in adaptive segmentation based on the contact response inflection point, interface response apex, and diffusion tail initiation position of the reference electrochemical impedance spectroscopy. It combines high-frequency contact offset, mid-frequency interface passivation, and low-frequency diffusion hysteresis according to frequency band continuity to form concealed pre-heating characteristics. Simultaneously, it determines the electrical connection propagation path and thermal diffusion propagation path based on the connection location, and then performs current reduction, balancing, local cooling, branch disconnection, and adjacent cell isolation in stages according to electrical connection traction type, thermal accumulation hysteresis type, and conduction superposition type. This approach shifts early warning and suppression from post-heating treatment to intervention upon detection of the pre-heating state, and avoids premature recovery or delayed isolation through a closed-loop verification process. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0060] Figure 1 This is a flowchart of an active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to the present invention.

[0061] Figure 2 This is a flowchart of the method for obtaining real-time electrochemical impedance spectroscopy according to the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Example 1, please refer to Figure 1 As shown in this embodiment, an active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy includes:

[0064] S100 acquires the reference electrochemical impedance spectroscopy of the battery under test under normal service conditions, and simultaneously records the state of charge, temperature state and connection position corresponding to the reference electrochemical impedance spectroscopy.

[0065] In this embodiment, since the battery electrochemical impedance spectroscopy (EIS) varies with different states of charge, ambient temperature, aging levels, and connection locations within the module, simply comparing the real-time EIS with a general threshold or standard sample can easily misjudge normal fluctuations as abnormalities or mask early hidden defects within normal differences. Therefore, by first obtaining the reference EIS of the battery under normal service conditions and simultaneously recording its corresponding state of charge, temperature, and connection location, the impedance distribution characteristics of the battery in a healthy state and its operating condition boundaries can be clearly identified. This allows subsequent steps to compare the real-time EIS under the same or correctable conditions, thereby accurately extracting impedance shift characteristics caused by anomalies such as loose tab connections, localized lithium plating, and micro-damage to the separator. Furthermore, the connection location can be used to determine the possible paths of anomalies spreading along electrical connections or thermal conduction directions, providing data for establishing the risk state of thermal runaway and generating active suppression strategies. Specifically, the following steps are included:

[0066] S101. After the battery under test has been assembled, maintained, reset, put into operation for the first time, or undergone a complete charge-discharge calibration, first determine whether the battery under test is in normal service condition. The normal service condition includes: the battery under test has not experienced over-temperature alarm, over-voltage alarm, under-voltage alarm, insulation abnormality alarm, smoke alarm, short circuit protection action, or forced power-off record; at the same time, the charging and discharging current of the battery pack containing the battery under test changes smoothly, without high-power rapid acceleration, fast charging switching, frequent relay switching, or sudden changes in external load.

[0067] S102, the battery management system reads the battery number, module, branch, series-parallel connection, positive and negative electrode connection direction, adjacent cell positions, position near the cooling channel, and position near the edge of the casing or heat source area of ​​the battery under test. This connection position is not simply used to mark the battery number, but is used to subsequently determine the direction in which abnormal impedance changes may diffuse through the tabs, busbars, casing, cooling plate, or adjacent cells.

[0068] For example, if the battery under test is located in the middle of the module, its heat is more likely to accumulate in the adjacent cells on the left and right; if the battery under test is close to the end of the bus, its contact resistance change is more likely to be related to loose tab connection or solder joint degradation; if the battery under test is close to the cooling inlet, its temperature state is naturally different from that of the battery far from the cooling inlet, and a separate benchmark needs to be established.

[0069] S103: Before acquiring the reference electrochemical impedance spectroscopy, read the current state of charge (SOC) of the battery under test. The SOC can be obtained by the battery management system based on current integration, voltage correction, or open-circuit voltage estimation. The purpose of acquiring the SOC is that the charge transfer process, diffusion process, and interface film state of the battery are different in high, intermediate, and low SOC states, resulting in different impedance spectra.

[0070] Therefore, subsequent real-time impedance spectra cannot be directly compared with impedance spectra under arbitrary states. Instead, reference electrochemical impedance spectra under the same or similar charge states should be used for comparison, or the real-time impedance spectra should be corrected using the charge state.

[0071] S104 simultaneously acquires a reference electrochemical impedance spectroscopy (TIS) spectrum and reads the temperature state of the battery under test. The temperature state includes at least the surface temperature of the battery under test, and may further include the module ambient temperature, cooling plate temperature, air inlet temperature, air outlet temperature, or liquid cooling inlet temperature.

[0072] The purpose of collecting temperature data is to address the high sensitivity of battery impedance to temperature. At low temperatures, the electrolyte's ion transport capacity decreases, leading to an overall increase in impedance; at high temperatures, interfacial reactions intensify, potentially causing a decrease in impedance. Without recording temperature data, the increased impedance caused by low temperatures might be misinterpreted as a pre-short circuit state or a loose connection at the tabs, while the decreased impedance caused by high temperatures might be mistakenly interpreted as a normal battery condition.

[0073] S105: After the battery under test is in a stable service state, the impedance acquisition unit applies a small AC disturbance signal to the battery under test. This disturbance signal should be smaller than the amplitude that would affect the normal charge and discharge control of the battery, so as not to cause significant temperature rise, significant change in state of charge, or protection action.

[0074] The impedance acquisition unit simultaneously acquires the voltage and current responses of the battery under test under the influence of a disturbance signal. Based on the amplitude and phase relationships between the voltage and current responses, the electrochemical impedance spectra of the battery under test are obtained at different frequency bands. These different frequency bands can at least reflect the contact connection state, interfacial reaction state, and ion diffusion state, enabling the reference electrochemical impedance spectrum to cover the main characteristic regions required for subsequent early identification of thermal runaway.

[0075] S106: After obtaining the initial electrochemical impedance spectroscopy (EIS), it undergoes noise reduction and validity assessment. Specifically, this includes: eliminating anomalies caused by relay activation, motor controller interference, charging pile ripple, communication interruptions, loose sampling lines, or sudden changes in external load; determining whether impedance changes in adjacent frequency bands are continuous during the same acquisition process; and determining whether the impedance spectrum morphology obtained from multiple repeated acquisitions is consistent. If an isolated jump point, abnormal phase reversal, or significant inconsistency with previous or subsequent sampling results appears in a particular impedance spectrum, it is not used as the reference EIS spectrum; instead, it re-enters the stabilization waiting and repeated acquisition process. This avoids mistakenly storing external noise as a battery health status.

[0076] S107. When multiple impedance spectra obtained from the battery under test under normal service conditions show a consistent trend, and the corresponding state of charge and temperature are within a recordable and traceable range, this impedance spectrum is determined as the reference electrochemical impedance spectrum of the battery under test. This reference electrochemical impedance spectrum is not a standard curve shared by all batteries, but rather an individualized health reference formed specifically for the battery under test. This preserves manufacturing differences, assembly differences, welding differences, and installation location differences, making subsequent judgments more suitable for the battery under test itself.

[0077] S108 binds and stores the reference electrochemical impedance spectroscopy (EIS), state of charge (SOC), temperature state, connection location, battery number, acquisition time, acquisition conditions, and validity judgment results to form a reference data file for the battery under test. This reference data file is used in subsequent steps. After obtaining the real-time EIS, the system retrieves the corresponding reference EIS based on the battery number and connection location, and then matches or corrects it based on the real-time SOC and temperature state to determine whether the shift in the real-time impedance spectrum originates from an internal battery anomaly, rather than from changes in normal operating conditions.

[0078] In this invention, the output of step S100 is: obtaining the reference electrochemical impedance spectroscopy (AIS) of the battery under test, and obtaining the state of charge, temperature state, and connection location associated with the reference AIS. This output will serve as the basis for subsequent real-time impedance spectroscopy correction, anomaly feature extraction, establishment of thermal runaway initiation risk states, and generation of active suppression strategies.

[0079] Please see Figure 2As shown in S200, when the battery under test is at the end of static charging, in standby mode, or in the energy storage cabinet charging gap, a perturbation signal is applied to the battery under test to obtain a real-time electrochemical impedance spectrum. The real-time electrochemical impedance spectrum is then corrected according to the state of charge and temperature to obtain a comparable impedance spectrum.

[0080] Understandably, it is crucial to actively acquire the real-time internal state of the battery under test under conditions of low current and low temperature rise, which are most easily overlooked by routine monitoring, and convert this state into a data basis that can be effectively compared with the reference electrochemical impedance spectroscopy (EIS). During periods of static charging, standby operation, or recharging in an energy storage cabinet, the external voltage, current, and temperature of the battery typically change slowly. Issues such as loose electrode connections, localized lithium plating, micro-damage to the separator, or micro-short circuits may not necessarily manifest as significant temperature rises or voltage differences, but they will first cause impedance changes in interface reactions, contact connections, and ion diffusion processes. Therefore, by applying a perturbation signal, the voltage and current responses of the battery under test can be actively excited without disrupting normal operation, resulting in a real-time EIS. Simultaneously, since the impedance spectrum itself is affected by the state of charge and temperature, directly comparing the real-time EIS with the reference EIS can easily lead to misinterpreting normal differences in charge or temperature as precursors to thermal runaway. Therefore, it is necessary to correct the real-time EIS based on the aforementioned recorded state of charge and temperature, converting it to a comparable operating condition corresponding to the reference EIS.

[0081] S210, first determine whether the data acquisition timing falls within the end of static charging, standby mode, or energy storage cabinet recharging interval. The end of static charging is defined as the charging current being within 5% of the rated charging current for 120 consecutive seconds, and the voltage of the battery under test changing by no more than 3 millivolts within 120 seconds; standby mode is defined as the main circuit having no drive discharge command and the bus current not exceeding 1% of the rated capacity for 180 consecutive seconds; the energy storage cabinet recharging interval is defined as an idle period of no less than 30 seconds between recharging commands.

[0082] Once any condition is met, the current state of charge (SOC) and temperature are read and compared with the SOC and temperature bound to the reference electrochemical impedance spectroscopy (EIS). The comparison method is as follows: a reference EIS with an absolute SOC difference not exceeding 3% and an absolute temperature difference not exceeding 2 degrees Celsius is selected from the reference data archive as the direct comparison object; if no corresponding data is available, the acquisition action is postponed until the charging current stabilizes or the charging control is idle. The time slice selection method is as follows: within a continuous 60-second recording, the average charging current is calculated every second. The absolute difference between adjacent second averages does not exceed 0.2 amperes, the equalization switch remains open, and the cooling pump speed change does not exceed 50 revolutions per minute. The 20th to 50th seconds of this 60-second period are then designated as the available acquisition time slice. With this processing, disturbance signals will not be superimposed on charging switching, equalization pulses, or cooling start-stop processes.

[0083] S220 does not use fixed frequency sweep perturbation within the time slice; instead, it first determines adjacent cells based on their connection positions. If the battery under test is located in the middle of a series branch, the preceding and following cells are selected as adjacent cells; if the battery under test is located at the end of a branch, the nearest cell connected to the same busbar and the nearest cell on the same cooling plate path are selected as adjacent cells. The sampling controller reads the instantaneous voltage sequence of adjacent cells within the first 10 seconds of the time slice, with a sampling interval of 10 milliseconds, to obtain the voltage fluctuation sequence of adjacent cells. Discrete Fourier transform is performed on the voltage fluctuation sequence of adjacent cells to obtain the set of external ripple dominant frequencies. The Discrete Fourier transform is performed according to the following formula: Where xn is the nth sampled value after removing the average value from the adjacent individual cell voltage fluctuation sequences, and N is the number of sampling points. Let Xk be the kernel function of the Fourier transform, and Xk be the amplitude and phase results at the corresponding frequency point. The top three frequencies in terms of amplitude are taken as the dominant frequencies of the external ripple. Then, frequencies that differ from the dominant frequency of the external ripple by no more than 8% are deleted from the preset impedance acquisition frequency table, resulting in an error frequency table. The preset impedance acquisition frequency table includes 0.1 Hz, 0.2 Hz, 0.5 Hz, 1 Hz, 2 Hz, 5 Hz, 10 Hz, 20 Hz, 50 Hz, 100 Hz, 200 Hz, 500 Hz, and 1000 Hz. An embedded perturbation signal is formed based on the error frequency table. The embedded perturbation signal is a superimposed sequence of multiple sinusoidal current components arranged in chronological order, with each frequency point lasting for three complete cycles, and a 0.5-second gap between different frequencies. The disturbance current amplitude is set to the lower of the rated capacity current of 0.005 times and 0.5 amps to ensure that the disturbance signal does not trigger the switching of charge and discharge control, nor does it change the recording accuracy of the state of charge of the battery under test.

[0084] S230, the embedded disturbance signal is connected to the branch where the battery under test is located through the sampling controller. During the period of each frequency point, the voltage response, current response, and passive response of the battery under test and adjacent cells are synchronously collected. The voltage response of the battery under test is denoted as Vt, the current response as It, and the passive voltage responses of adjacent cells as Va and Vb, respectively. First, phase-locked loop extraction is performed on Vt, It, Va, and Vb respectively. The phase-locked loop extraction is performed according to the following formula: , , Where yn is the processed response sequence, f is the current frequency, fs is the sampling frequency, and Ay is the complex response at the current frequency. This yields the complex voltage response AVt, complex current response AIt, and complex passive responses AVA and AVb of adjacent cells of the battery under test. Then, a common-mode background Bc is constructed as follows: Bc = (AVa + AVb) / 2.

[0085] When the battery under test has only one adjacent cell, Bc is taken as the complex passive response of that adjacent cell. The common-mode background is removed from the complex voltage response of the battery under test to obtain the bulk voltage response Vbatt, calculated as: Vbatt = AVt - Bc. The real-time electrochemical impedance spectroscopy is then calculated using the bulk voltage response and the complex current response, calculated as: Zreal(f) = Vbatt / AIt. Zreal(f) consists of the real part, imaginary part, amplitude, and phase. This processing ensures that the common components corresponding to bus disturbances, low-frequency charger oscillations, and synchronous drift of ambient temperature are carried by the passive response of the adjacent cell and then subtracted from the response of the battery under test, retaining the response components caused by the internal contact connections, interface reactions, and diffusion processes of the battery under test.

[0086] S240, based on the current state of charge and temperature, select the corresponding state segment from the reference electrochemical impedance spectroscopy. The state segment includes two sets of reference data adjacent to the current state of charge and two sets of reference data adjacent to the current temperature; when only one set of adjacent data exists, that set is used as the correction reference. When performing state transition correction on the real-time electrochemical impedance spectroscopy, first calculate the state of charge correction and the temperature correction. The state of charge correction is determined according to the difference between two adjacent sets of reference data at the same temperature: The temperature state correction is determined based on the difference between two adjacent sets of reference data under the same charging state: Where Zbase is the reference electrochemical impedance spectroscopy, S1 and S2 are adjacent charged states, T1 and T2 are adjacent temperature states, Scur is the current charged state, Tcur is the current temperature state, and Sref and Tref are the charged state and temperature state bound to the selected reference electrochemical impedance spectroscopy. The corrected comparable impedance spectrum is obtained according to the following formula: When the reference data file contains data that is completely consistent with the current state of charge and the current temperature, Csoc(f) and Ctemp(f) are both set to 0, and Zcmp(f) is equal to the result of real-time electrochemical impedance spectroscopy after subtracting common-mode background.

[0087] After completing the state transition correction, the comparable impedance spectra are saved separately for high-frequency, mid-frequency, and low-frequency bands. The high-frequency band is 100 Hz to 1000 Hz, used for subsequent identification of contact connection offset; the mid-frequency band is 1 Hz to 100 Hz, used for subsequent identification of interface passivation offset; and the low-frequency band is 0.1 Hz to 1 Hz, used for subsequent identification of diffusion hysteresis offset.

[0088] Since the comparable impedance spectrum has removed external disturbances shared by adjacent cells and completed migration corrections through state of charge and temperature, the impedance shift extracted in subsequent steps can point to changes in the battery under test relative to its own health state, rather than spurious shifts caused by micro-fluctuations in the charging end current, temperature differences at the cooling inlet, or differences in installation location. This implementation method can obtain a stable comparable impedance spectrum before the surface temperature of the battery under test rises or the voltage drops abruptly, providing input for subsequently establishing the risk state of thermal runaway initiation.

[0089] S300, the comparable impedance spectrum is compared with the reference electrochemical impedance spectrum in segments, and high-frequency contact offset features, mid-frequency interface passivation features and low-frequency diffusion hysteresis features are extracted to obtain the hidden exothermic pre-state features of the battery under test.

[0090] After obtaining the comparable impedance spectrum, the comparable impedance spectrum and the reference electrochemical impedance spectrum are compared segment by segment according to steps S310 to S340. In this embodiment, both the comparable impedance spectrum and the reference electrochemical impedance spectrum are arranged from high to low frequency, and each frequency point includes a real part, an imaginary part, an amplitude, and a phase. To facilitate subsequent processing, the two spectra are first unified to the same set of frequency points; when there are frequency points in the comparable impedance spectrum that are inconsistent with the reference electrochemical impedance spectrum, linear interpolation of two adjacent frequency points is used to fill in the gaps. The interpolation result is obtained by multiplying the distance between the frequency point to be filled and the low-frequency side frequency point by the impedance difference between the high-frequency side and the low-frequency side, and then adding the impedance value of the low-frequency side.

[0091] S310, first determine the contact response inflection point, interface response apex, and diffusion tail inflection point within the reference electrochemical impedance spectroscopy. The contact response inflection point is determined by the slope of the real part change: calculate the real part slope by dividing the difference between the real parts of two adjacent frequency points by the frequency interval, then calculate the difference between adjacent real part slopes, and the position where the difference first changes in the same direction twice consecutively is determined as the contact response inflection point.

[0092] The apex of the interface response arc is determined according to the absolute value of the imaginary part: after the contact response transition, the frequency point with the largest absolute value of the imaginary part is selected as the apex of the interface response arc; if the absolute values ​​of the imaginary parts of two adjacent frequency points are the same, the frequency point with the lower frequency is selected as the apex of the interface response arc.

[0093] The starting position of the diffusion tail is determined by phase change: after the apex of the interface response arc, the phase difference between adjacent frequency points is calculated sequentially, and the position where three consecutive phase differences maintain the same sign is determined as the starting position of the diffusion tail. An adaptive high-frequency band is formed from the frequency above the contact response inflection point, an adaptive mid-frequency band is formed from the contact response inflection point to the starting position of the diffusion tail, and an adaptive low-frequency band is formed from the frequency below the starting position of the diffusion tail.

[0094] S320 extracts high-frequency contact offset features within the adaptive high-frequency band. First, it calculates the difference between the real part of the starting frequency of the comparable impedance spectrum and the real part of the starting frequency of the reference electrochemical impedance spectrum. A positive difference is recorded as a positive starting offset, and a negative difference as a negative starting offset. Then, it calculates the phase change direction for each frequency point in descending order of frequency. If the phase difference between the comparable impedance spectrum and the reference electrochemical impedance spectrum first increases and then decreases, it is recorded as a phase oscillation; if the phase difference changes directly in one direction, no phase oscillation is recorded.

[0095] The continuity of adjacent frequency points is then calculated as follows: if the offset direction of the current frequency point is the same as that of the previous frequency point, it is recorded as 1; if they are different, it is recorded as 0. If the number of consecutive times recorded as 1 within the adaptive high-frequency band is no less than 3, the offset is considered to originate from the battery under test. If only a single frequency point shows an offset, and neither the preceding nor following frequency points changes in the same direction, then that frequency point is discarded as an isolated offset caused by the instantaneous contact of the sampling fixture. After the starting offset direction, phase swing sequence, and continuity of adjacent frequency points are jointly confirmed, the high-frequency contact offset characteristics are output. The high-frequency contact offset characteristics include at least the offset direction, the number of consecutive frequency points, and the number of phase swings.

[0096] S330, based on the offset direction corresponding to the high-frequency contact offset characteristics, enters the adaptive mid-frequency band. First, using the interface response arc apex of the reference electrochemical impedance spectrum as a reference, calculate the arc apex position of the comparable impedance spectra near the same frequency point. The arc apex position is obtained according to the principle of maximizing the absolute value of the imaginary part. If the arc apex of the comparable impedance spectrum shifts towards the lower frequency side, it is recorded as arc apex lag; if the arc apex shifts towards the higher frequency side, it is recorded as arc apex advance. Arc apex broadening is calculated based on half-arc width, which is the frequency span between the high-frequency and low-frequency points when the absolute value of the imaginary part reaches half of the absolute value of the arc apex's imaginary part. The broadening amount is obtained by subtracting the reference half-arc width from the comparable impedance spectrum's half-arc width. Arc apex retraction is calculated based on the difference in the absolute value of the imaginary part of the arc apex; when the absolute value of the imaginary part of the comparable impedance spectrum's arc apex is less than the absolute value of the reference arc apex's imaginary part, it is recorded as retraction. Subsequently, the spectral line connection between the end of the adaptive high-frequency band and the beginning of the adaptive mid-frequency band is examined. When the real part offset direction of the two frequency points before and after the connection is consistent, and the imaginary part offset direction is also consistent, it is determined that the arc apex change and the high-frequency contact offset feature remain continuous. After this continuity is established, the mid-frequency interface passivation feature is output. The mid-frequency interface passivation feature includes the arc apex hysteresis direction, half-arc width change, and arc apex retraction state.

[0097] S340 uses the mid-frequency interface passivation feature as the low-frequency comparison entry point, entering the adaptive low-frequency band. Starting from the beginning of the diffusion tail, the real and imaginary shifts of the comparable impedance spectrum relative to the reference electrochemical impedance spectrum are calculated in descending order of frequency. If the real shift direction of at least four consecutive frequency points in the low-frequency band is consistent with the real shift direction of the mid-frequency interface passivation feature at the junction, and the imaginary shift direction does not reverse, it is recorded as a continued extension of the diffusion tail. Then, the low-frequency tail slope is calculated by dividing the imaginary difference between the end of the low-frequency band and the beginning of the diffusion tail by the real difference; the low-frequency tail slope of the comparable impedance spectrum is subtracted from the low-frequency tail slope of the reference electrochemical impedance spectrum to obtain the low-frequency diffusion hysteresis feature.

[0098] After completing the above processing, the high-frequency contact offset feature, the mid-frequency interface passivation feature, and the low-frequency diffusion hysteresis feature are combined in frequency band order. During combination, the terminal frequency of the high-frequency contact offset feature, the starting frequency of the mid-frequency interface passivation feature, and the entry frequency of the low-frequency diffusion hysteresis feature must be continuous in frequency order, and their main offset directions must not be reversed.

[0099] When this combination relationship is satisfied, the hidden exothermic pre-state characteristics of the battery under test are obtained. This characteristic indicates that the contact connection anomaly first appears in the high-frequency band, then causes the interface reaction to be blocked, and further propagates to the low-frequency diffusion process. This makes the establishment of the subsequent risk state not dependent on a single frequency point change, but on the continuous evolution relationship across frequency bands.

[0100] S400, based on the concealed heat release pre-state characteristics and the connection position of the battery under test, determine the electrical connection propagation path and thermal diffusion propagation path between the abnormal cell and the adjacent cells, and establish the corresponding thermal runaway initiation risk state.

[0101] In this invention, the concealed heat release pre-state characteristics include high-frequency contact offset characteristics, mid-frequency interface passivation characteristics, and low-frequency diffusion hysteresis characteristics; the connection positions include the positive and negative electrode connection directions of the battery under test, the busbar connection end, the arrangement order of adjacent cells, the common cooling plate, the shell pressing surface, and the corresponding relationship of the clamping bracket.

[0102] S410, first read the spectral line start-up offset direction and the band end continuous direction from the high-frequency contact offset characteristics. The spectral line start-up offset direction is obtained by summing the real part offsets of the first 3 frequency points of the adaptive high-frequency band and taking the sign, and is denoted as D0; the band end continuous direction is obtained by summing the real part offsets of the last 3 frequency points of the adaptive high-frequency band and taking the sign, and is denoted as D1.

[0103] The real part offset is the real part of the comparable impedance spectrum minus the real part of the reference electrochemical impedance spectrum. A positive summation is recorded as 1, a negative summation as -1, and a zero summation as 0. The phase oscillation sequence is then read. If the phase difference oscillates towards the negative phase side first and then back to the reference phase side along the high-to-low frequency direction, it is recorded as P = 1; if it oscillates towards the positive phase side first and then back to the reference phase side, it is recorded as P = -1; and if there is no phase oscillation, it is recorded as P = 0.

[0104] The connection position is then oriented. When the positive end is located in the direction of increasing sequence number of adjacent units, the positive end orientation code Cp is 1 and the negative end orientation code Cn is -1; when the positive end is located in the direction of decreasing sequence number of adjacent units, Cp is -1 and Cn is 1; the orientation code Cc on the shell bonding side is 0.

[0105] Calculate the positive electrode side orientation value Ap, the negative electrode side orientation value An, and the shell-fitting side orientation value Ac respectively. Ap is equal to the sum of D0 multiplied by Cp, D1 multiplied by Cp, and P; An is equal to the sum of D0 multiplied by Cn, D1 multiplied by Cn, and P inverted; Ac is 2 when D0 and D1 have opposite signs, Ac is 1 when D0 and D1 have the same sign and P is 0, and Ac is 0 in all other cases.

[0106] Take the position with the highest azimuth value as the starting position of the anomaly; if Ap and An are equal, select the side where the bus connection end is consistent with the actual current entry end of the battery under test; if Ac is equal to any end side, select the side where the casing is attached, because anomalies on the casing side are easily manifested as the starting point offset and the end continuous direction canceling each other out in the high frequency band.

[0107] S420: Establish candidate electrical connection paths with the starting point of the anomaly as the entry point. When the starting point of the anomaly is on the positive side, the positive electrode tab welding point, the positive busbar connection end, and the end of the adjacent cell directly connected to the positive busbar are taken as the first-level nodes. When the starting point of the anomaly is on the negative side, the negative electrode tab welding point, the negative busbar connection end, and the end of the adjacent cell directly connected to the negative busbar are taken as the first-level nodes. When the starting point of the anomaly is on the shell contact side, first take the tab welding point closest to the shell contact surface, and then enter the busbar connection end corresponding to that tab.

[0108] The path propagation direction is determined by D1. When D1 is 1, propagation proceeds along the direction of increasing sequence number of adjacent cells; when D1 is -1, propagation proceeds along the direction of decreasing sequence number of adjacent cells; and when D1 is 0, propagation proceeds along the actual current direction. For each stage advancement, only busbars, tab welding points, and series / parallel branches with continuous metallic connections to the current node are selected, without crossing insulating partitions or gaps between cells without metallic connections. The electrical connection propagation path is formed by sequentially arranging the entry node, the first-stage node, and the first adjacent cell node. This path reflects the direction in which contact connection anomalies may first be drawn to adjacent cells via metallic conductors.

[0109] S430 performs reverse verification of the mid-frequency interface passivation characteristics and low-frequency diffusion hysteresis characteristics. First, the real part offset summation sign of the last two frequency points of the adaptive mid-frequency band is calculated, denoted as Dm; then, the real part offset summation sign of the first two frequency points of the adaptive low-frequency band is calculated, denoted as Dl. When both Dm and Dl are consistent with D1, it is determined that no directional rotation has occurred at the frequency band junction; when Dm or Dl is opposite to D1, it is determined that the aforementioned electrical connection propagation path only represents contact traction and does not enter the thermal diffusion candidate path.

[0110] When no directional rotation occurs, adjacent cells sharing a cooling plate, casing clamping surface, or clamping bracket with the battery under test are read from the connection position and arranged according to the order of physical contact with the anomaly initiation location to form a candidate thermal diffusion path. The starting point of the candidate thermal diffusion path is the casing contact surface of the battery under test or the clamping surface near the anomaly initiation location, and the ending point is the adjacent cell of the first shared heat dissipation entity. This yields the thermal diffusion propagation path.

[0111] S440, the electrical connection propagation path and the thermal diffusion propagation path are cross-superimposed. The adjacent cell numbers included in the electrical connection propagation path are formed into a set Pe, and the adjacent cell numbers included in the thermal diffusion propagation path are formed into a set Pt. The cross set Pg is obtained according to Pg = Pe ∩ Pt.

[0112] When Pg contains at least one adjacent cell number, the battery under test is established as a conduction-superimposed thermal runaway initiation risk state, and the adjacent cells in Pg are prioritized for suppression. When Pg is empty and Pe contains adjacent cell numbers, an electrical connection-driven thermal runaway initiation risk state is established. When Pg is empty and Pt contains adjacent cell numbers, a thermal accumulation-hysteresis-driven thermal runaway initiation risk state is established. This processing correlates the continuous evolution of the impedance spectrum across frequency bands with the actual connection locations within the battery pack, so that the risk state no longer only indicates the anomaly of the battery under test itself, but also provides the electrical connection direction and thermal diffusion direction of the anomalous cell propagating to adjacent cells.

[0113] S500 generates an active suppression strategy based on the risk state of thermal runaway initiation and performs different operations in order of increasing risk to obtain the suppressed battery state.

[0114] After establishing the initial risk state of thermal runaway, the process enters the stage of forming and implementing active suppression strategies. In this embodiment, the initial risk states of thermal runaway include electrically connected induced thermal runaway, thermal accumulation delayed thermal runaway, and conduction-superimposed thermal runaway. The active suppression strategy does not involve a uniform power cut-off after a sudden temperature rise. Instead, based on the aforementioned electrical connection propagation path and heat diffusion propagation path, it sequentially applies operations such as current reduction, equalization pause, branch disconnection, local cooling enhancement, and isolation of adjacent cells to the battery under test and its adjacent cells, thereby suppressing the abnormal reaction before it generates significant heat release.

[0115] During execution, the system first reads the thermal runaway nascent risk state, anomaly initiation location, electrical connection propagation path, thermal diffusion propagation path, and concealed exothermic pre-state characteristics of the battery under test. The high-frequency contact offset characteristic within the concealed exothermic pre-state characteristics is used to determine whether the electrical connection anomaly is still propagating along the busbar or tab welding point; the mid-frequency interface passivation characteristic is used to determine whether interface side reactions have been weakened; and the low-frequency diffusion hysteresis characteristic is used to determine whether the internal mass transfer hindrance of the battery under test is still ongoing.

[0116] Based on the above information, the active suppression strategy is divided into three execution levels: Level 1 corresponds to the risk state of thermal runaway initiation caused by electrical connection traction, Level 2 corresponds to the risk state of thermal runaway initiation caused by thermal accumulation hysteresis, and Level 3 corresponds to the risk state of thermal runaway initiation caused by conduction superposition. If two risk states exist simultaneously in the same judgment, the higher-level one is executed, and the corresponding operation for the lower-level one is retained.

[0117] When the risk state of thermal runaway initiation is electrical connection traction type, the first level of suppression is executed first. The charging equipment or energy storage converter is controlled to reduce the charging current of the branch where the battery under test is located to 50% of the current charging current and maintain it for 180 seconds; if the current state is in a standby state, the branch where the battery under test is located is prohibited from entering pulse charging.

[0118] Subsequently, passive equalization of the battery under test and adjacent cells along the electrical connection propagation path is paused for at least 300 seconds. This process aims to prevent the equalization resistor or switch from introducing additional heat near the abnormal initiation location, while simultaneously reducing current stress at the tab welding points and busbar connections. After the first stage of suppression is completed, the terminal voltage, branch current, surface temperature of the battery under test, and the terminal voltage of adjacent cells are recorded to establish the first suppression state.

[0119] When the risk state of thermal runaway initiation is a delayed thermal accumulation type, a second-level suppression is implemented on the basis of the first-level suppression. First, the branch charging current is kept in the state after the first-level suppression, and then the cooling intensity of the region corresponding to the thermal diffusion propagation path is increased.

[0120] If liquid cooling is used, the coolant flow rate of the coolant sharing the cooling plate with the battery under test is adjusted to 120% of the rated flow rate and maintained for 300 seconds. If air cooling is used, the airflow rate of the air duct containing the battery under test is adjusted to 120% of the rated airflow rate and maintained for 300 seconds. The cooling target covers not only the battery under test but also the first adjacent cell in the heat diffusion propagation path. If the casing contact side is identified as the abnormal starting point, priority is given to cooling the cooling area in contact with the casing pressing surface. If the clamping bracket is included in the heat diffusion propagation path, priority is given to cooling the adjacent cells on both sides of the clamping bracket. After the second level of suppression is completed, the temperature difference between the battery under test and the adjacent cells in the heat diffusion propagation path, the cooling execution time, and the cooling area number are recorded to form the second suppression state.

[0121] When the risk state of thermal runaway initiation is of the conduction-superimposed type, the third-level suppression is implemented after the first and second-level suppression. First, adjacent cells resulting from the intersection of electrical connection propagation paths and thermal diffusion propagation paths are prioritized for protection. The equalization operation of these adjacent cells is kept off, and their participation in charging is prohibited. Then, the charging input of the abnormal branch containing the battery under test is cut off, causing the battery under test to be removed from the external charging circuit. If the battery pack structure allows for the disconnection of a single branch, only the branch containing the battery under test is disconnected; if a single branch cannot be disconnected independently, the charging circuit of the module to which the battery under test belongs is disconnected. After disconnection, local cooling is maintained for 300 seconds, and adjacent cells resulting from the intersection are marked as isolated objects.

[0122] The isolated object no longer receives equalization compensation from the same bus direction of the battery under test, nor is it used as a voltage reference cell for subsequent power replenishment control. This forms the third suppression state.

[0123] During the suppression process at each level, the suppression is performed progressively from lowest to highest risk, rather than directly entering the highest level of power-off isolation. If, after Level 1 suppression, the terminal voltage of the battery under test does not continuously decrease within 180 seconds, and the surface temperature does not rise by more than 1 degree Celsius within 180 seconds, then Level 1 suppression is maintained and the subsequent impedance verification is initiated. If the terminal voltage continuously decreases by more than 5 millivolts, or the surface temperature rises by more than 1 degree Celsius within 180 seconds, then Level 2 suppression is initiated. If, after Level 2 suppression, the temperature difference between the battery under test and adjacent cells on the heat diffusion propagation path still increases by more than 1 degree Celsius within 300 seconds, then Level 3 suppression is initiated. The above values ​​are set based on low-power charging scenarios to ensure that the suppression action occurs before smoke, valve ejection, and significant thermal shock.

[0124] The suppressed battery state includes the branch connection state, charging allowed state, equalization allowed state, cooling execution state, adjacent cell isolation state, and suppression execution record of the battery under test. The branch connection state indicates whether the battery under test is still connected to the charging circuit; the charging allowed state indicates whether the battery under test is allowed to continue charging; the equalization allowed state indicates whether the battery under test and adjacent cells on the propagation path are allowed to be passively equalized; the cooling execution state indicates the cooling area, cooling duration, and cooling intensity; and the adjacent cell isolation state indicates whether adjacent cells on the cross-over path have been removed from the charging reference. This suppressed battery state serves as the input for subsequent acquisition of electrochemical impedance spectroscopy and return to segmented comparison, enabling the active suppression action to form a closed loop with impedance verification.

[0125] S600: Obtain the electrochemical impedance spectroscopy again for the suppressed battery state, and return the obtained electrochemical impedance spectroscopy to step S300 for verification. If the hidden exothermic pre-state characteristics are weakened, the suppression strategy is maintained. If the hidden exothermic pre-state characteristics are enhanced, a thermal runaway warning is output and a forced power-off isolation is performed.

[0126] After the active suppression strategy is completed, the process proceeds to the S600 verification stage. In this invention, the suppressed battery state includes branch connection state, charging allowed state, equalization allowed state, cooling execution state, adjacent cell isolation state, and suppression execution record. The purpose of S600 is to determine whether the active suppression strategy has reduced the pre-heating characteristics, avoiding premature release of suppression based solely on temporary voltage or temperature stability, and also avoiding delays in forced power-off isolation when impedance anomalies continue to increase.

[0127] S610, after the completion of Level 1, Level 2, or Level 3 suppression, does not immediately perform impedance verification; instead, it first reads the battery status after suppression. If the battery under test is still in the charging circuit, the charging current must remain at the same control level for 60 consecutive seconds, and the absolute value of the average current difference between adjacent seconds must not exceed 0.2 amperes. If the battery under test has exited the charging circuit, the residual current on the bus must not exceed 0.1 amperes for 60 consecutive seconds. The cooling execution state must remain at the same level for at least 60 seconds, and the equalization allowable state for the battery under test and adjacent cells on the propagation path must be closed.

[0128] After the above conditions are met, the next 10 seconds to 50 seconds will be used as the verification acquisition window. The period before the 10th second is considered the electrical transient dissipation time after the suppression action, and no acquisition will be performed after the 50th second to avoid the cooling adjustment changing the temperature state again. This ensures that the electrochemical impedance spectroscopy acquired again will not contain non-bulk responses caused by branch disconnection, equalization switching, or changes in cooling level.

[0129] In the S620, within the verification acquisition window, the embedded disturbance signal acquisition method from S200 is used. For the battery state after Level 1 and Level 2 suppression, the disturbance signal is applied to the branch where the battery under test is located; for the battery state after Level 3 suppression, the disturbance signal is only applied to the low-energy detection circuit that allows safe sampling, without restoring the charging-allowed state of the battery under test.

[0130] After acquiring the complex voltage response, complex current response, and passive response of adjacent cells of the battery under test, a second electrochemical impedance spectroscopy (EIS) is obtained according to the common-mode background stripping method in S200. The second EIS is denoted as Zagain(f), and is calculated as: Zagain(f) = [AVt2 - Bc2] / AIt2; where AVt2 is the complex voltage response of the battery under test within the verification acquisition window, Bc2 is the common-mode background formed by the passive responses of adjacent cells, AIt2 is the complex current response within the verification acquisition window, and f is the frequency. Subsequently, state transition corrections are performed based on the current state of charge and current temperature state within the verification acquisition window to obtain the verification comparable impedance spectrum Zreview(f), calculated as: Zreview(f) = Zagain(f) - Csoc2(f) - Ctemp2(f); Csoc2(f) is the state of charge correction amount at the verification time, and Ctemp2(f) is the temperature state correction amount at the verification time; their calculation methods are consistent with S240. The resulting Zreview(f) serves as the input for returning to S300 for segmented comparison.

[0131] S630: Input Zreview(f) and the reference electrochemical impedance spectroscopy into S310 to S340. Re-extract high-frequency contact offset features, mid-frequency interface passivation features and low-frequency diffusion hysteresis features according to the contact response transition, interface response arc apex and diffusion tail start position to obtain the verified hidden exothermic pre-state features.

[0132] To determine whether the pre-exothermic features of the concealed exothermic spectrum are weakened or enhanced, the pre-exothermic features of the concealed exothermic spectrum obtained from S300 before active suppression are read first. The comparable impedance spectrum before active suppression is denoted as Zbefore(f), the comparable impedance spectrum after verification is denoted as Zreview(f), and the reference electrochemical impedance spectrum is denoted as Zbase(f).

[0133] The high-frequency review ratio Rh is calculated according to the following formula: Where Fh is the adaptive high-frequency band set, and Re represents the real part. Rh is used to characterize the degree of preservation of high-frequency contact migration features after suppression.

[0134] The intermediate frequency reconciliation ratio Rm is calculated according to the following formula: Where W1 represents the amount of suppression of the change in the width of the first half of the arc, L1 represents the amount of shift in the frequency index of the first arc apex, and G1 represents the amount of suppression of the shrinkage of the first arc apex; W2, L2, and G2 are the corresponding values ​​during the verification. Rm is used to characterize the change in the passivation characteristics of the intermediate frequency interface after suppression.

[0135] The low-frequency verification ratio Rl is calculated according to the following formula: Where Q1 is the difference in low-frequency tail slope before suppression, and Q2 is the difference in low-frequency tail slope during verification. Rl is used to characterize the change in low-frequency diffusion hysteresis characteristics after suppression.

[0136] The comprehensive review value R is calculated according to the following formula: R = 0.4 × Rh + 0.35 × Rm + 0.25 × Rl. When any denominator is 0, the corresponding item is not included in the calculation, and the weights of the remaining items are reset to the original weight ratio.

[0137] In S640, when R does not exceed 0.85, and the number of consecutive frequency points of the high-frequency contact offset characteristic does not increase, the arc apex of the mid-frequency interface passivation characteristic does not continue to move towards the low-frequency side, and the number of consecutive trailing frequency points of the low-frequency diffusion hysteresis characteristic does not increase, the concealed heat dissipation pre-state characteristic is determined to be weakened. At this time, the already executed suppression strategy is maintained, the suspended equalization allowance state is not restored, the charging current is not increased, the adjacent cell isolation state is not canceled, and after 300 seconds, it re-enters S610 for the next round of verification. Only when R does not exceed 0.85 in two consecutive rounds of verification is it allowed to switch to manual confirmation or low current maintenance state.

[0138] When R reaches 1.10, or the number of consecutive frequency points of high-frequency contact offset characteristics increases by more than 2 compared to before suppression, or the mid-frequency interface response apex continues to shift to the low-frequency side by more than 1 frequency point, or the number of consecutive frequency points of low-frequency diffusion tail increases by more than 2 compared to before suppression, the hidden exothermic pre-state characteristics are determined to be enhanced. At this time, a thermal runaway warning is output, and the warning content includes the battery number under test, the location of the abnormality initiation, the electrical connection propagation path, the thermal diffusion propagation path, the risk status of thermal runaway initiation, and the comprehensive verification value R.

[0139] After the thermal runaway warning is issued, a forced power-off isolation is executed. First, the charging input of the branch containing the battery under test is stopped, then the positive relay of the branch is disconnected, and after a 2-second interval, the negative relay of the branch is disconnected. If the branch containing the battery under test cannot be disconnected independently, the charging circuit of the module to which the battery under test belongs is disconnected. After disconnection, the area corresponding to the thermal propagation path continues to cool for 600 seconds, and the equilibrium allowable state of the battery under test and its overlapping adjacent cells is kept off. The overlapping adjacent cells are no longer used as reference objects for charging, and the battery number and the verified hidden heat release pre-state characteristics are written into the fault record. This forms the battery state after forced power-off isolation, so that the thermal runaway warning and isolation actions are triggered by the impedance characteristic closed loop, rather than waiting for a temperature surge or smoke signal before taking action.

[0140] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for active thermal runaway early warning and suppression based on battery electrochemical impedance spectroscopy, characterized in that, include: S100: Obtain the reference electrochemical impedance spectroscopy of the battery under test under normal service conditions, and simultaneously record the state of charge, temperature state and connection position corresponding to the reference electrochemical impedance spectroscopy. S200: When the battery under test is at the end of static charging, in standby mode, or in the energy storage cabinet charging gap, a perturbation signal is applied to the battery under test to obtain a real-time electrochemical impedance spectrum. The real-time electrochemical impedance spectrum is then corrected according to the state of charge and temperature to obtain a comparable impedance spectrum. S300, the comparable impedance spectrum is compared with the reference electrochemical impedance spectrum in segments, and high-frequency contact offset features, mid-frequency interface passivation features and low-frequency diffusion hysteresis features are extracted to obtain the hidden exothermic pre-state features of the battery under test. S400, based on the concealed heat release pre-state characteristics and the connection position of the battery under test, determine the electrical connection propagation path and thermal diffusion propagation path between the abnormal cell and the adjacent cell, and establish the corresponding thermal runaway initiation risk state. S500 generates an active suppression strategy based on the risk state of thermal runaway initiation and performs different operations in order of increasing risk to obtain the suppressed battery state. S600: Obtain the electrochemical impedance spectroscopy again for the suppressed battery state, and return the obtained electrochemical impedance spectroscopy to step S300 for verification. If the hidden exothermic pre-state characteristics are weakened, the suppression strategy is maintained. If the hidden exothermic pre-state characteristics are enhanced, a thermal runaway warning is output and a forced power-off isolation is performed.

2. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 1, characterized in that, The acquisition of the reference electrochemical impedance spectroscopy includes: After the battery under test has been assembled, maintained and reset, put into operation for the first time or completed a full charge and discharge calibration, it is confirmed that the battery under test has not experienced over-temperature, over-voltage, under-voltage, insulation abnormality, smoke alarm, short circuit protection action, or forced power-off record. The connection position is obtained by reading the battery number, module, branch, series-parallel relationship, positive and negative electrode connection direction, adjacent cell position, adjacent position of cooling channel and adjacent position of casing edge of the battery under test; An AC perturbation was applied under conditions where the charged state and temperature state were recorded simultaneously, and the voltage and current responses were collected to obtain the initial electrochemical impedance spectrum. Isolated jump points caused by relay activation, charging ripple, communication interruption, and external load abrupt changes were eliminated. When the trend of spectral changes was consistent across multiple acquisitions, the initial electrochemical impedance spectrum was determined as the reference electrochemical impedance spectrum.

3. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 2, characterized in that, The comparable impedance spectra obtained include: At the end of static charging, during parking standby, or during the energy storage cabinet recharging interval, the current state of charge and current temperature of the battery under test are read, and a reference electrochemical impedance spectrum with an absolute value of the state of charge difference not exceeding 3% and an absolute value of the temperature difference not exceeding 2 degrees Celsius is selected from the reference data archive as the corresponding object. If no corresponding object exists, the average charging current is calculated every second within the 60-second continuous running record. When the absolute value of the difference between the average values ​​of adjacent seconds does not exceed 0.2 amperes, the equalization switch remains open, and the change in cooling pump speed does not exceed 50 revolutions per minute, the 20th to 50th seconds are determined as the time slice that can be collected. A perturbation signal is applied within the available acquisition time slice to acquire the real-time electrochemical impedance spectrum of the battery under test. The spectrum is then corrected based on the current state of charge, the current temperature, and the corresponding object to obtain a comparable impedance spectrum.

4. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 3, characterized in that, The disturbance signal is an embedded disturbance signal, and the formation of the embedded disturbance signal includes: Adjacent cells are determined based on the connection position. When the battery under test is located in the middle of a series branch, the preceding and following cells are selected. When the battery under test is located at the end of a branch, the nearest cell connected to the same busbar and the nearest cell on the same cooling plate path are selected. Read the instantaneous voltage sequence of adjacent cells within the first 10 seconds of the acquisition time slice, remove the average value from the instantaneous voltage sequence and perform frequency domain processing to obtain the external ripple main frequency; Delete the frequency points that differ from the main frequency of the external ripple by no more than 8% from the preset impedance acquisition frequency point table to obtain the wrong frequency point table; The sinusoidal current components are arranged sequentially according to the frequency point table, with each frequency point lasting for 3 complete cycles. A 0.5-second blank segment is set between different frequency points, and the amplitude of the disturbance current is set to the lower value between the rated capacity corresponding to 0.005 times the current and 0.5 amperes.

5. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 4, characterized in that, The acquisition of real-time electrochemical impedance spectroscopy includes: During each frequency misalignment period, the voltage response, current response, and passive response of the battery under test, as well as the adjacent cells, are simultaneously acquired. The voltage response, current response, and passive response of the battery under test and adjacent cells are extracted by phase-locked loop at the same frequency to obtain the complex voltage response, complex current response, and complex passive response of the battery under test and adjacent cells. The average result of the complex passive response of two adjacent cells is used as the common mode background. When there is only one adjacent cell in the battery under test, the complex passive response of the adjacent cell is used as the common mode background. The bulk voltage response is obtained by subtracting the common-mode background from the complex voltage response of the battery under test, and the real-time electrochemical impedance spectroscopy is obtained from the ratio of the bulk voltage response to the complex current response.

6. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 5, characterized in that, Correction of real-time electrochemical impedance spectroscopy based on charge state and temperature state includes: Select two sets of reference data adjacent to the current state of charge and two sets of reference data adjacent to the current temperature from the reference data archive. Based on the impedance difference and state of charge difference between two adjacent sets of reference data under the same temperature condition, the state of charge correction amount is obtained. Based on the impedance difference and temperature state difference between two adjacent sets of reference data under the same charging state, the temperature state correction amount is obtained; By subtracting the state of charge correction and temperature state correction from the real-time electrochemical impedance spectroscopy, a comparable impedance spectrum is obtained that retains only the shift of the battery under test relative to its own healthy state.

7. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 6, characterized in that, The acquisition of concealed exothermic pre-state characteristics includes: Unify the comparable impedance spectra with the reference electrochemical impedance spectra to the same set of frequency points; Within the reference electrochemical impedance spectrum, the contact response transition is determined based on the position where the slope difference of the real part of adjacent frequency points changes in the same direction twice. The interface response arc apex is determined based on the frequency point where the absolute value of the imaginary part is the largest after the contact response transition. The diffusion tailing start position is determined based on the position where the phase difference of the three consecutive phases after the interface response arc apex maintains the same sign. An adaptive high-frequency band is formed at frequencies above the contact response transition point, an adaptive mid-frequency band is formed at frequencies from the contact response transition point to the start of the diffusion tail, and an adaptive low-frequency band is formed at frequencies below the start of the diffusion tail. Segmented comparisons were performed in the adaptive high-frequency band, adaptive mid-frequency band, and adaptive low-frequency band, respectively.

8. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 7, characterized in that, Segment comparison includes the following steps: Within the adaptive high-frequency band, high-frequency contact offset features are extracted based on the real part offset direction of the starting frequency, the phase swing order, and the continuity of adjacent frequency points, and isolated offsets that do not change in the same direction at the preceding and following frequency points are eliminated. Within the adaptive mid-frequency band, the low-frequency side movement, half-width variation, and apex retraction state of the interface response arc apex are tracked along the offset direction of the high-frequency contact offset feature. The mid-frequency interface passivation features are extracted when the real offset direction and the imaginary offset direction are consistent at the frequency band junction. In the adaptive low-frequency band, the low-frequency tail extension state is tracked from the starting position of the diffusion tail, and the low-frequency diffusion hysteresis feature is extracted when the real part offset direction of at least 4 consecutive frequency points is consistent with the real part offset direction of the mid-frequency interface passivation feature at the junction and the imaginary part offset direction is not reversed. The high-frequency contact offset feature, mid-frequency interface passivation feature, and low-frequency diffusion hysteresis feature that appear continuously and whose main offset direction is not reversed are combined into a concealed exothermic pre-state feature.

9. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 8, characterized in that, The establishment of a thermal runaway initiation risk state includes: The starting offset direction of the spectral line is obtained by summing the real offsets of the first three frequency points of the adaptive high-frequency band, and the continuous direction at the end of the frequency band is obtained by summing the real offsets of the last three frequency points of the adaptive high-frequency band. The anomaly starting location is determined by combining the phase swing sequence. Taking the abnormal starting position as the entry point, the positive electrode tab welding point, negative electrode tab welding point, bus connection end, series and parallel branches, and the end of the adjacent single unit directly connected to the bus are identified as candidate electrical connection paths, and the electrical connection propagation path is obtained according to the continuous direction at the end of the frequency band. The direction of the summation of the real offsets of the last two frequency points of the adaptive mid-frequency band and the summation of the real offsets of the first two frequency points of the adaptive low-frequency band is verified. When both are consistent with the continuous direction at the end of the frequency band, the adjacent cell that shares the cooling plate, shell pressing surface or clamping bracket with the battery under test is determined as the heat diffusion propagation path. The adjacent cell numbers in the electrical connection propagation path are cross-superimposed with the adjacent cell numbers in the thermal diffusion propagation path, and based on this, a thermal runaway initiation risk state of conduction superposition, thermal runaway initiation risk state of electrical connection traction, or thermal accumulation hysteresis type is established.

10. The active thermal runaway early warning and suppression method based on battery electrochemical impedance spectroscopy according to claim 9, characterized in that, Active suppression strategies and reviews include: When the risk state of thermal runaway initiation is electrical connection traction type, the charging current of the branch where the battery under test is located is reduced to 50% of the current charging current and maintained for 180 seconds. At the same time, the passive equalization of the battery under test and adjacent cells on the electrical connection propagation path is suspended for no less than 300 seconds. When the risk state of thermal runaway initiation is the thermal accumulation hysteresis type, the liquid cooling flow rate or air cooling flow rate in the area corresponding to the thermal diffusion propagation path is adjusted to 120% of the rated value and maintained for 300 seconds, and the cooling range covers the battery under test and the first adjacent cell on the thermal diffusion propagation path. When the risk state of thermal runaway initiation is the conduction superposition type, the equalization operation of adjacent cells obtained by cross superposition is turned off and their participation in charging is prohibited. Then the charging input of the abnormal branch where the battery under test is located is cut off. After suppression, the electrochemical impedance spectroscopy is acquired again and verified. When the verified hidden exothermic pre-state characteristics are weakened, the active suppression strategy is maintained. When the verified hidden exothermic pre-state characteristics are enhanced, a thermal runaway warning is output, and the charging input is stopped in sequence, the positive terminal relay of the branch is disconnected, and the negative terminal relay of the branch is disconnected after 2 seconds. At the same time, the corresponding area of ​​the heat diffusion propagation path is kept cooled for 600 seconds.

Images