Battery failure identification method and system based on magnetic-thermal-electric multi-signal

Abstract

The present application belongs to the technical field of battery detection and identification, in order to solve the problem of inaccurate battery failure identification, a battery failure identification method and system based on magnetic-thermal-electric multi-signal are proposed, in the process of battery charging and discharging, the magnetic field data and temperature signal on the surface of the battery and the current and voltage of the battery are synchronously collected; using the collected charging and discharging current and the calibrated geometric coupling coefficient, the ampere background magnetic field generated by the current is reconstructed, and the intrinsic spin magnetic field signal of the positive active material of the battery is extracted; the magnetic signal health factor, the magnetic differential curve characteristic peak parameter and the magnetic field-current hysteresis area are extracted from the intrinsic spin magnetic field signal; according to the extracted magnetic characteristics from the intrinsic spin magnetic field signal, the current and voltage of the battery, and the extracted thermal characteristics from the temperature signal, a magnetic-thermal-electric multi-physical field characteristic vector is constructed, and a multi-physical field coupling model is used to identify the failure mode of the battery, thereby improving the accuracy, sensitivity and reliability of the battery failure identification.

Description

Technical Field

[0001] This invention belongs to the field of battery detection and identification technology, and particularly relates to a battery failure identification method and system based on magnetic-thermal-electrical multi-signal. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Electrochemical energy storage devices such as lithium-ion batteries, sodium-ion batteries, and potassium-ion batteries have been widely used in portable electronic devices, electric vehicles, and energy storage systems due to their high energy density, long cycle life, and wide range of applications. As battery systems develop towards higher specific energy, higher power, and more complex operating conditions, the coupling relationships between local magnetic response, thermal response, and electrochemical behavior during battery reactions are becoming increasingly complex, placing higher demands on in-situ characterization techniques.

[0004] Existing battery characterization techniques mostly focus on acquiring single electrochemical or physical signals, making it difficult to simultaneously characterize the correlations between multiple sources of information during the reaction process. While in-situ magnetic detection devices can measure magnetic signals during battery reactions, they primarily focus on acquiring the magnetic signals themselves, lacking collaborative processing and quantitative correlation analysis with temperature and electrochemical signals, leading to inaccurate battery failure identification. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, this invention provides a battery failure identification method and system based on magnetic-thermal-electrical multi-signal. It constructs a magnetic-thermal-electrical multi-physics feature vector based on synchronously acquired signals and uses a pre-trained multi-physics coupling model for identification, which significantly improves the accuracy, sensitivity and reliability of battery failure identification, and realizes early, in-situ and accurate failure judgment and warning.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a battery failure identification method based on magnetic-thermal-electrical multi-signal methods, including: During the charging and discharging process of the battery, magnetic field data and temperature signals on the battery surface, as well as the battery current and voltage, are collected simultaneously. By using synchronously acquired charging and discharging currents and calibrated geometric coupling coefficients, the Ampere background magnetic field generated by the current is reconstructed, thereby extracting the intrinsic spin magnetic field signal of the battery positive electrode active material. Extract the magnetic signal health factor, characteristic peak parameters of the magnetic differential curve, and magnetic field-current hysteresis area from the intrinsic spin magnetic field signal; Based on the magnetic features extracted from the intrinsic spin magnetic field signal, the battery's current and voltage, and the thermal features extracted from the temperature signal, a magnetic-thermal-electric multiphysics feature vector is constructed. Using a pre-trained multiphysics coupling model, typical battery failure modes are identified.

[0007] Secondly, the present invention provides a battery failure identification system based on magnetic-thermal-electrical multi-signal, comprising: The synchronous acquisition module is configured to simultaneously acquire magnetic field data and temperature signals on the battery surface, as well as the battery current and voltage, during the battery charging and discharging process. The magnetic signal processing module is configured to: reconstruct the Ampere background magnetic field generated by the current using the synchronously acquired charging and discharging current and the calibrated geometric coupling coefficient, thereby extracting the intrinsic spin magnetic field signal of the battery positive electrode active material; The extraction module is configured to extract the magnetic signal health factor, characteristic peak parameters of the magnetic differential curve, and magnetic field-current hysteresis area from the intrinsic spin magnetic field signal. The identification module is configured to: construct a magnetic-thermal-electric multiphysics feature vector based on the magnetic features extracted from the intrinsic spin magnetic field signal, the battery's current and voltage, and the thermal features extracted from the temperature signal; and identify typical failure modes of the battery using a pre-trained multiphysics coupling model.

[0008] Thirdly, the present invention provides an electronic device including a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the method described in the first aspect.

[0009] Fourthly, the present invention provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the method described in the first aspect.

[0010] Fifthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect.

[0011] The above one or more technical solutions have the following beneficial effects: In this invention, the Ampere background magnetic field is reconstructed using charging and discharging current and geometric coupling coefficient, effectively eliminating current interference magnetic fields and accurately extracting the intrinsic spin magnetic field signal of the positive electrode active material. This significantly improves the purity of the magnetic signal and the detection sensitivity, avoiding interference from background noise in failure judgment. Three core magnetic features—magnetic signal health factors, characteristic peak parameters of the magnetic differential curve, and magnetic field-current hysteresis area—are extracted directionally from the intrinsic spin magnetic field signal, enabling quantitative characterization of battery material degradation, increased polarization, and structural anomalies. This invention constructs a multi-physics field coupled feature vector and recognition model through synchronous acquisition, collaborative processing, and quantitative correlation analysis of multiple magnetic, thermal, and electrical signals. This fundamentally solves the problems of single-signal detection being susceptible to interference, unreliable criteria, and inaccurate identification, significantly improving the accuracy, sensitivity, and reliability of battery failure identification, and achieving early, in-situ, and accurate failure judgment and warning.

[0012] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0013] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0014] Figure 1 This is a schematic diagram of an in-situ battery assembly assembled for real-time in-situ magnetic-thermal-electric joint testing of a TMR chip integrated temperature sensor, as described in an embodiment of the present invention. Figure 2 A schematic diagram of a test device designed for an embodiment of the present invention, which uses a TMR chip-integrated temperature sensor to detect changes in the magnetic-thermal properties of materials inside a pouch cell in real time. In the figure, 1 is the in-situ battery packaging structure, 2 is the TMR chip, 3 is the temperature sensor, 4 is the magnetic-thermal-electric coupling detection module, 5 is the temperature acquisition instrument, 6 is the data logger, 7 is the lifting platform, and 8 is the magnetic field generator. Detailed Implementation

[0015] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0016] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0017] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0018] Example 1 This embodiment achieves real-time in-situ detection of changes in the magneto-thermal characteristics during the electrochemical reaction process in the open state of a commercial battery by synchronously acquiring, preprocessing, time-series registration, and correlation analysis of magnetic and temperature signals.

[0019] like Figure 1 As shown, in this embodiment, a high-density TMR sensor array is constructed on a flexible polyimide (PI) substrate using micro-nano fabrication technology, including: an in-situ battery packaging structure 1, a TMR chip 2, and a temperature sensor 3; in this embodiment, the TMR chip 2, the temperature sensor 3, and the battery testing system are integrated to form a magnetic-thermal-electric coupling detection module 4, and the magnetic-thermal-electric coupling detection module 4 is precisely bonded to the in-situ battery.

[0020] Specifically, a high-sensitivity tunnel magnetoresistive (TMR) chip is selected and soldered onto a custom PCB board. The tightly soldered chip pins are led out with DuPont wires. Both sides of the PCB board are covered with adhesive and polished to be horizontal, ensuring that the chip is on the outside and is fixed parallel by epoxy resin clamps. Thermistor materials and high-precision temperature sensors are integrated on the surface of the TMR chip or in the adjacent position through screen printing, vacuum coating or bonding to form a miniature detection module that integrates magnetic field, temperature and electrical signals.

[0021] Preferably, the thermistor material is an NTC thermistor film or a PTC thermistor polymer material with a thickness ≤50μm to ensure a thermal response time <1s.

[0022] Preferably, the sensitive axis of the high-sensitivity tunnel magnetoresistive chip is oriented in the z-axis direction, and the magnetic field detection range covers the order of 0.1μT to 10mT.

[0023] Preferably, the PCB board adopts a structure of 4 layers or more, with the bottom layer laid with grounded copper foil to achieve electromagnetic shielding and reduce external magnetic noise interference.

[0024] The magnetic-thermal-electric coupling detection module is non-destructively and tightly attached to the surface of the battery under test using a fixture made of high thermal conductivity insulating adhesive or epoxy resin board, ensuring that the gap between the chip sensitive surface and the battery casing is ≤0.1mm to avoid magnetic field attenuation caused by air gaps; at the same time, temperature compensation calibration points are arranged around the module to achieve real-time compensation for ambient temperature drift.

[0025] Preferably, the bonding location is in the middle of the battery or in the area where the electrochemical reaction is most active.

[0026] In this embodiment, the raw magnetic field signal acquired by the TMR array It consists of three superimposed parts: ; Where j is the number of the j-th sensor channel in the TMR array; For charging and discharging current The background magnetic field generated by the Ampere force; The geometric coupling coefficient of the j-th channel to the current varies slowly with time, and its position changes slightly due to deformation such as electrode expansion. Positive electrode active material (such as Fe) 2+ / Fe 3+ The intrinsic magnetic field generated by the change of spin magnetic moment of transition metal ions (such as Ni / Co / Mn) with SOC is the effective signal that needs to be extracted. To mitigate slow zero drift and DC bias, including low-frequency interference such as residual geomagnetic field and sensor zero-point temperature drift.

[0027] Also synchronized is the current signal I(t) from the battery testing system, which is strictly aligned on the time axis (TTL hardware synchronization, accuracy <1ms).

[0028] Before the start of each charge-discharge cycle, the reference values ​​of each channel are collected when the current I=0 (static state). : ; in, For the first There are 1 sampling time points, where N represents the total number of sampling time points; Indicates the j-th sensor channel. Each sampling time The original magnetic field signal collected. Among them, For the first The measured static reference value of the channel (unit: T) is obtained by measuring the current. Under static conditions of the battery cell The average value of the original magnetic field signal at each sampling time is used to subtract the DC bias in the subsequent background stripping step; N is the total number of sampling times.

[0029] From the original magnetic field signal Subtract the corresponding baseline value from the middle Obtain the debiasing signal : ; For zero drift (such as thermal drift) that changes slowly over time, the reference value is updated once during each cycle interval (the rest period between charge / discharge transitions, typically 5–30 seconds) to achieve dynamic tracking. After this step, The item was effectively suppressed, and the residual amount was less than 0.1% of the full scale.

[0030] During the first cycle of a new battery, the rate of change of magnetic susceptibility of the cathode material is selected to satisfy... The calibration window is defined as the SOC range within which the relative rate of change of magnetic susceptibility does not exceed 5% of the maximum magnetic susceptibility. Indicates the maximum magnetic susceptibility. This represents the relative rate of change of magnetic susceptibility.

[0031] Calculate the initial coupling coefficient within this calibration window. : ; in, For the bias removal signal within the calibration window; To calibrate the current within the calibration window, the initial coupling coefficient matrix {k} is obtained by averaging over multiple time points. j (0)}

[0032] Using synchronously acquired charging and discharging current I(t) and coupling coefficient Reconstruct the background magnetic field generated by the current and subtract it from the signal: ; in, This is the positive intrinsic magnetic signal estimated in the current step; As the online real-time estimated signal for the current cycle, it has two specific applications: First, it can be displayed in real time via LabVIEW during the experiment, allowing operators to monitor signal quality and determine whether each channel is functioning correctly; second, it serves as the input for adaptive coupling coefficient updates, specifically, after each cycle, the signal is displayed in the SOC platform calibration window. Using approximate constancy as a criterion, the coupling coefficients are re-estimated using the least squares method. This drives the iterative convergence of the coefficients, which, after adaptive multi-loop updates, Approaching the final decoupling result step by step The relationship between the two is as follows: That is, the two tend to be consistent after the coupling coefficients have fully converged; Indicates the j-th sensor channel The debiasing signal at each moment; express The charging and discharging current at any given moment.

[0033] After each loop, the calibration window (SOC platform segment) for the current loop is re-found, and the coupling coefficient for this loop is calculated. And updated using an exponentially weighted average: ; in, The exponentially weighted average forgetting factor is dimensionless and has a range of values. A typical value of 0.90 is used to balance the weights between historical estimates and new estimates; For the first The historical cumulative coefficient, already smoothed by exponential weighted average at the start of the next cycle, is expressed in T / A; superscript. Indicates the first The loop continues. For the first The coupling coefficients estimated in the second iteration within the calibration window using the least squares method are in units of T / A and have not yet undergone exponential weighted averaging.

[0034] The three symbols mentioned above are obtained in different ways: For the first The coupling coefficient, estimated within the calibration window using the least squares method in the next iteration, is calculated using the following formula: ; The value only reflects the estimation result of the current cycle and contains single-measurement noise, so it cannot be directly used for decoupling. This is the debiasing signal.

[0035] For the first The historical cumulative coefficients, which have already undergone exponentially weighted average smoothing at the start of the next cycle, are the previous... The accuracy of the results stored in the calibration file after each iteration increases with the number of iterations. After this loop ends, the new estimated value will be... Compared with historical values According to forgetting factor The update coefficients obtained after weighted fusion are calculated using the following formula: The coefficients will be written into the calibration file as input coefficients for the next cycle.

[0036] After the above three steps, the final output is the decoupled pure spin magnetic field signal of each channel. : ; in, For the first Channel sampling time The intrinsic magnetic field generated by the spin magnetic moment of the sensed positive electrode active material is the target extraction signal; Indicates the sampling time of the j-th sensor channel. The original magnetic field signal collected; This represents the reference value for the j-th sensor channel; express Current at any given moment; for Moment coupling coefficient.

[0037] The pure spin magnetic field signal directly reflects the intrinsic spatiotemporal magnetic variation of the positive electrode active material, and is subsequently fed into three analytical branches: one for analyzing the pure spin magnetic field signal. As SOC evolves, dB / dV characteristic peaks are extracted for SOC / SOH estimation; the RDT (relaxation differential temperature) characteristics of the temperature sensor are combined to characterize the heat generation distribution; and the electrode expansion mechanical response is characterized by combining the signal from the capacitive displacement sensor.

[0038] The system power-on sequence is strictly defined: first, power is supplied to the LDO bias power supply of the TMR array, and the ripple is allowed to stabilize. The battery testing system remains open-circuited to ensure that the bias setup process is not interfered with by the charging and discharging current. Then, the data acquisition card (DAQ) is powered on, and the readings of all TMR channels and temperature sensor channels are checked to ensure they are within reasonable ranges; abnormal channels are marked and removed. After the experimental parameters are configured in the LabVIEW software on the main control PC, the system enters the baseline acquisition state. The battery cells are allowed to stand in the shielding cover for ≥30 minutes (OCV stable, temperature uniform) before... Under these conditions, all TMR channels were acquired at full speed for 60 seconds, and the mean value of each channel was taken to obtain the zero-field reference. and temperature reference Save it to the calibration file.

[0039] Strict time synchronization of the three signals is achieved through a hardware trigger chain. Using the battery testing system as the master clock, LabVIEW pre-sets the DAQ to a state awaiting external triggering, clears the internal sampling buffer, and resets the counter. LabVIEW then sends a start command to the battery testing system. Simultaneously with executing the first sampling point, the testing system outputs a TTL rising edge pulse (1ms width) from the auxiliary I / O port. This pulse is directly connected to the external trigger input of the DAQ. Upon triggering, the DAQ immediately begins synchronous periodic sampling of the TMR voltage and temperature signals, and LabVIEW records the PC timestamp of the trigger moment as a unified time signature. All three data streams (magnetic signal, temperature signal, and electrochemical signal) are aligned with this trigger time, achieving a synchronization accuracy better than ±1ms. Regarding sampling rates, the TMR magnetic signal and temperature signal are uniformly acquired by DAQ, with the former ranging from 20 to 100 Hz and the latter from 1 to 10 Hz. The electrochemical signal... , The data is recorded by the battery testing system's built-in sampling rate (1-10Hz) and then transmitted to LabVIEW via TCP / GPIB for aggregation.

[0040] After initiating the battery charge-discharge cycle, the change in magnetic field strength along the z-axis is measured in real time throughout the process. The TMR chip, based on the tunneling magnetoresistance effect, transmits the magnetic field... Converted into resistance change This leads to the output voltage signal. Through calibration coefficients Converted to magnetic field strength, the formula is: Simultaneously, the battery surface temperature is collected via an integrated temperature sensor. With a resolution of 0.01℃ and a sampling interval consistent with the magnetic field signal, the battery testing system continuously outputs and records the charging and discharging current. With voltage Three signals (magnetic field) ,temperature Current / Voltage The data is uploaded to the computer in real time via a data reader card, enabling the synchronous acquisition of magnetic-thermal-electric multi-physics field coupling data.

[0041] To eliminate the temperature cross-sensitivity of the TMR chip, real-time software compensation is employed. TMR chips exhibit two types of temperature cross-sensitivity: zero-point bias drift with temperature and sensitivity coefficient. Because of temperature variations, complete compensation is performed in two steps.

[0042] The first step is zero-point temperature drift compensation, at the reference temperature. Record the zero-point output voltage of each TMR chip in a zero magnetic field environment at (25℃) The output voltage after compensation in the experiment for: ; in, The zero-point temperature drift coefficient is obtained by linear fitting of multi-temperature point calibration (5℃ step, 0℃~60℃) of the constant temperature chamber; The measured original output voltage of the TMR chip (unit: V).

[0043] The second step is sensitivity temperature drift compensation. The change with temperature is expressed as: ; in, The sensitivity temperature coefficient is obtained by temperature scanning calibration under a known magnetic field of a Helmholtz coil; To be at the reference temperature The TMR sensitivity coefficient under calibration (unit: T / V); Let t be the temperature at time t.

[0044] The first step, zero-point temperature drift compensation, targets additive interference, specifically the spurious output voltage generated by the TMR chip due to temperature changes under zero magnetic field conditions. This item is independent of the external magnetic field and is directly superimposed on the original output.

[0045] After compensation, the voltage with zero-point drift removed is obtained. : ; at this time It still contains multiplicative errors, that is... Sensitivity coefficient used when converting to magnetic field It deviates from the calibrated value with temperature. If a fixed one is used directly Conversion, when the temperature deviates from the reference value, a systematic proportional error is introduced, the magnitude of which is... ,exist ℃ ℃ - ¹ Under typical conditions, the scaling error is approximately 4%.

[0046] The second step, sensitivity temperature drift compensation, eliminates the aforementioned multiplicative error, converting the output of the first step... Divide by the actual sensitivity after temperature correction The final compensation result was obtained. : ; The order of the two steps cannot be reversed: if the second step of sensitivity correction is performed first and then the zero point is subtracted, the zero point drift term will be affected. It will be incorrectly divided by Magnification or reduction introduces additional errors. After two steps are cascaded, both additive zero-point error and multiplicative sensitivity error are eliminated, resulting in a better output. This is the compensated magnetic induction intensity that only reflects the changes in the actual external magnetic field, and can be directly used for subsequent signal decoupling and state estimation.

[0047] Combining the two compensation steps yields the temperature-corrected magnetic flux density. Full expression: ; in, The measured original output voltage of the TMR chip; To be at the reference temperature The TMR sensitivity coefficient under calibration; Reference temperature Zero-point output voltage of the TMR chip in a zero magnetic field environment; Temperature coefficient of sensitivity; This is the zero-point temperature drift coefficient.

[0048] After the three signals are aligned on a unified time axis, they form a multimodal state vector at each moment, while the two-dimensional spatial distribution of the TMR array is also considered. The inverse problem of the Biot-Savart law is used to invert the magnetic moment distribution inside the battery cell. This allows for the spatial localization of cathode material degradation. All data, including raw signals, decoupled signals, temperature fields, and displacement signals, is grouped and stored in a file according to cyclic numbering. At the end of the experiment, LabVIEW synchronously sends a stop command to the battery testing system and DAQ, and closes the file after confirming that the final data packet has been completely written.

[0049] On the computer side, a LabVIEW or Python+MATLAB hybrid programming environment is used to filter (low-pass + moving average), calibrate, transform, and perform coupling calculations on the acquired raw data. A LabVIEW program is written based on the output voltage-magnetic field relationship of the TMR chip to synchronously display the chip voltage and magnetic field strength curves. At the start of the test, battery electrochemistry, sensor chip power supply, and temperature measurement are simultaneously initiated to ensure real-time synchronization of electrical-magnetic-thermal data.

[0050] After temperature compensation and signal decoupling, the final output pure spin magnetic field signals of each channel are obtained as follows: First, the raw TMR output voltage acquired by DAQ is... Substituting into the temperature drift compensation formula, we obtain the temperature-corrected magnetic flux density. : ; The above formula is the complete expression for temperature drift compensation, with subscripts... Indicates the first The passage. Then... As input, replacing the original magnetic field signal Following a three-step process of background stripping → dynamic decoupling → adaptive correction, the final result is... : ; Here The intrinsic magnetic field in the signal decomposition model mentioned above The meaning is consistent; it is the final extraction result after temperature correction to further remove Ampere force background and zero drift, and it is the basis for all subsequent state estimations.

[0051] Using a single index in the signal decoupling process ( The channels of the array are sequentially numbered to simplify the formula. In the spatial averaging step, to reflect the two-dimensional spatial structure of the array, the channels are numbered... and row / column position (in For line numbers, (Column number) through mapping relationship Establish a correspondence, that is, the first The first channel is equivalent to the one located at the first... Line 1 The decoupled signal of the sensor array is denoted as: The two representations describe the same signal from the same sensor, with identical numerical values, differing only in their symbolic form. Therefore, the spatial averaging formula is: ; In the formula, , These represent the number of rows and columns of the TMR array, respectively. Total number of channels; The average magnetic flux density across the entire array is a scalar representing the overall positive electrode magnetization state of the cell, compared to any single channel. The relationship is that they are the arithmetic mean of all channels, and both have the same dimension (unit: T), but... It suppresses spatial non-uniformity noise, resulting in a higher signal-to-noise ratio.

[0052] Regarding SOC estimation, the Fe²⁺ content in the cathode active material (taking LFP as an example) during the lithium insertion / extraction process... + / Fe³ + Valence state switching causes a monotonic change in magnetic susceptibility, and the spatial average magnetic flux density of the entire array... There is a stable mapping relationship between it and SOC.

[0053] In the first 0.1C slow charge-discharge calibration experiment, the true SOC value was accurately established using the coulomb counting method. : ； in, The start time of the experiment ( The initial state of charge (dimensionless) of the charge experiment was determined. Discharge experiment ; The rated capacity of the battery is expressed in A·s, or coulombs, and is equal to the rated capacity in mAh multiplied by 3.6, which serves as the normalization benchmark. Indicates time The current; This indicates the Coulomb efficiency.

[0054] by For the tag, the average magnetic flux density of the entire array space and temperature As input, a calibration curve is fitted using polynomial regression: ; coefficient Determined by least squares method (goodness of fit) In subsequent loops, the real-time full-array spatial average magnetic flux density is directly substituted. and temperature Estimating SOC does not rely on current integration, thus fundamentally avoiding the cumulative error of coulomb counting.

[0055] To further improve robustness under different operating conditions, a Kalman filter is used to fuse the magnetic signal and the Coulomb count, and the state equation is... With observation equation They are respectively: ; ; in, , These represent process noise and observation noise, respectively. The Kalman gain is adaptively adjusted based on the variance ratio of the two types of noise, and the SOC estimation error can be controlled within ±2%. b0-b4 are calibration fit coefficients determined using least squares polynomial regression based on the initial 0.1C calibration experimental data.

[0056] For SOH estimation, temperature gradient is considered. Based on current / voltage characteristics, a correlation matrix of magnetic, thermal, and electrical fields is constructed. With cyclic aging, the structural degradation of the cathode material leads to changes in the state of charge (SOC) at the same point of charge. The absolute value decreases, defining the magnetic signal health factor. : ; in, For the first Average magnetic flux density under full charge state in the second cycle; For the first Average magnetic flux density under full charge state in the next cycle.

[0057] Further calculation of the differential characteristics of the decoupled magnetic signal, with the magnetic differential curve plotted on the voltage axis. (i.e., one of the analytical indicators listed in the figure) After Savitzky-Golay smoothing, the peak position shift of its characteristic peak. Peak height attenuation Directly reflects the degree of phase transition and the decrease in utilization rate of active materials; magnetic field-current hysteresis area The degree of intensification of electrochemical polarization is defined as: ; This area monotonically increases with aging and exhibits a strong linear correlation with capacity decay. A multi-dimensional feature vector is constructed based on these characteristics. : ; in, ; Indicates the peak position shift of the characteristic peak; Indicates the peak height decay of the characteristic peak; The magnetic field-current hysteresis area; This indicates the deviation of the highest surface temperature from the initial value.

[0058] Using Support Vector Regression (SVR) or Random Forest (RF) models, Input, RPT measured capacity Train the SOH estimation model using labels: ; After each RPT (Real-Time Testing) is completed, the new measured capacity is written to the training set, and the model is continuously updated online, achieving adaptive improvement in SOH estimation accuracy as data accumulates. All analysis metrics can be automatically reported and stored, and remote monitoring is supported.

[0059] Systematic testing was conducted on batteries with different positive and negative electrode material systems under various failure states, establishing a multimodal characteristic database. The database covers the following five typical failure modes, each corresponding to characteristics in three dimensions: magnetic, thermal, and electrochemical: Regarding lithium plating, at the end of charging (SOC>80%), a specific TMR channel exhibits an abnormal magnetic signal abrupt change (amplitude>3σ, where σ is the normal cycle standard deviation) that does not monotonically change with SOC. The corresponding region shows a localized temperature increase (ΔT>1.5℃), while the coulombic efficiency also decreases. A downward trend is observed (normal LFP battery CE > 99.5%, drops below 99% after lithium plating). Regarding gas production, the TMR channel in the positive electrode region... The overall degradation rate is abnormally accelerated, with peak amplitudes exceeding normal degradation trends at high SOC. Battery swelling also leads to decreased sensor adhesion and introduces systemic magnetic field shift. SEI film thickening manifests as... - The SOC curve shows an overall shift, with a reduction in the dB / dV peak area, accompanied by increased temperature rise (increased internal resistance) and higher low-frequency impedance during cycling. The structural degradation of the active material is manifested in the dB / dV curve as an irreversible positive shift of the characteristic peak and a decrease in peak height, with a rapid and synchronous capacity decay. In the early stages of an internal short circuit, this is characterized by abrupt changes in the TMR channel signal near the short circuit location, abnormal local temperature rise rates (>5℃ / min), and accelerated self-discharge during the resting period.

[0060] The decoupled multidimensional feature vectors are then input into a classifier trained on the aforementioned database: ; in, As a health factor for magnetic signals, , These represent the peak position shift and peak height attenuation of the characteristic peak in dB / dV, respectively. This represents the deviation of the highest surface temperature from the initial value. The above feature vectors are classified, and failure type labels (normal, lithium plating, gas production, SEI thickening, structural degradation, internal short circuit) and confidence probabilities for each category are output. An alarm is triggered when the confidence of any failure type exceeds a preset threshold (e.g., 0.8).

[0061] Taking lithium plating detection as an example, the complete judgment logic is explained as follows: ① An abnormal sudden change in the TMR signal at the end of charging (e.g., >3σ) is detected; ② A local temperature rise occurs in the corresponding area of ​​the temperature sensor (e.g., ΔT > 1.5℃); ③ The coulombic efficiency CE(n) shows a decreasing trend. When all three conditions are met simultaneously, it is judged as a lithium plating event, and it is recommended to reduce the charging rate or reduce the charging cutoff SOC to achieve a closed-loop response from detection to warning.

[0062] like Figure 2 As shown, the test device for real-time in-situ detection of changes in the magnetothermal properties of materials inside a bag battery using a TMR chip-integrated temperature sensor includes a temperature acquisition instrument 5, a data logger 6, a lifting platform 7, and a magnetic field generator 8.

[0063] As one implementation method, the specific steps of the assembly method for an in-situ pouch cell with a TMR chip integrated temperature sensor for real-time in-situ magnetic-thermal joint testing are as follows: Place a 2cm×3cm transition metal compound electrode material in the positive electrode tank, and then lay the separator, lithium sheet counter electrode, spring sheet and gasket in sequence. After adding electrolyte, close the positive and negative electrode tanks, tighten the screws and apply silicone sealant. Insert the battery connector so that the electrode leads come into contact with the silver-plated wires and are led out from the electrode interface to complete the in-situ battery encapsulation. The z-axis sensor chip and the platinum resistance temperature sensor are integrated to form a magnetic-thermal detection module. The module is precisely fixed to the front detection area of ​​the in-situ pouch cell using a non-magnetic epoxy resin clamp, so that the sensitive surface of the TMR chip is in close contact with the aluminum-plastic film surface of the battery and the platinum resistance temperature sensor is in close contact with the battery surface without gaps. A diameter wiring hole is opened on the outside of the clamp to allow the TMR chip shielding wire and the temperature sensor wiring to pass through.

[0064] As one implementation method, the specific steps of the test method for real-time in-situ detection of synchronous changes in the magnetism and temperature of an in-situ pouch cell using a temperature sensor integrated with a TMR chip are as follows: Cut and fold the edges of the packaged in-situ pouch cell, connect the nickel-plated copper wires on the outside of the cell to the terminals of the electrochemical test cabinet, and use gaskets to fix the cell to the aluminum alloy non-magnetic lifting platform. Adjust the lifting platform to keep the cell horizontal and ensure that the cell does not shake or shift during the test. Connect the shielded differential signal line of the TMR chip of the magnetic-thermal detection module to the Keithley 2182 nanovoltmeter, and connect the platinum resistance temperature sensor to the precision temperature acquisition instrument 5. Connect the Keithley 2182 nanovoltmeter and the temperature acquisition instrument 5 to the same external computer through the RS232 interface. Use the computer-side synchronization calibration software to complete the clock synchronization of the three test instruments and achieve millisecond-level signal acquisition synchronization. Test parameters were set using electrochemical testing software, and the acquisition programs of Keithley 2182 nanovoltmeter and temperature acquisition instrument 5 were triggered simultaneously. The sampling frequency was set to 5Hz to achieve synchronous acquisition and real-time storage of electrochemical, magnetic, and temperature signals. The acquired magnetic signal differential voltage value was converted into magnetic field strength using computer-based data analysis software, and a three-dimensional change curve of magnetic field strength-temperature-battery capacity was plotted to analyze the coupling change law of magnetic-thermal characteristics during battery charging and discharging.

[0065] Example 2 The purpose of this embodiment is to provide a battery failure identification system based on magnetic-thermal-electrical multi-signal, including: The synchronous acquisition module is configured to simultaneously acquire magnetic field data and temperature signals on the battery surface, as well as the battery current and voltage, during the battery charging and discharging process. The magnetic signal processing module is configured to: reconstruct the Ampere background magnetic field generated by the current using the synchronously acquired charging and discharging current and the calibrated geometric coupling coefficient, thereby extracting the intrinsic spin magnetic field signal of the battery positive electrode active material; The extraction module is configured to extract the magnetic signal health factor, characteristic peak parameters of the magnetic differential curve, and magnetic field-current hysteresis area from the intrinsic spin magnetic field signal. The identification module is configured to: construct a magnetic-thermal-electric multiphysics feature vector based on the magnetic features extracted from the intrinsic spin magnetic field signal, the battery's current and voltage, and the thermal features extracted from the temperature signal; and identify typical failure modes of the battery using a pre-trained multiphysics coupling model.

[0066] In further embodiments, the following is also provided: An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When executed by the processor, the computer instructions perform the method described in Embodiment 1. For brevity, further details are omitted here.

[0067] It should be understood that in this embodiment, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0068] Memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of memory may also include non-volatile random access memory. For example, memory may also store information about the device type.

[0069] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the method described in Embodiment 1.

[0070] The method in Embodiment 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The software modules can reside in readily available storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not provided here.

[0071] A computer program product includes a computer program that, when executed by a processor, implements the method described in Embodiment 1.

[0072] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which execute in a device on a target real or virtual processor to perform the processes / methods described above. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided among program modules as needed. The machine-executable instructions for the program modules can execute within a local or distributed device. In a distributed device, the program modules can reside in both local and remote storage media.

[0073] The computer program code used to implement the methods of the present invention may be written in one or more programming languages. This computer program code may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the computer or other programmable data processing device, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a computer, partially on a computer, as a stand-alone software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0074] In the context of this invention, computer program code or related data may be carried by any suitable carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals may include electrical, optical, radio, sound, or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0075] Those skilled in the art will recognize that the units and algorithm steps described in conjunction with the embodiments herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0076] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A battery failure identification method based on magnetic-thermal-electrical multi-signal, characterized in that, include: During the charging and discharging process of the battery, magnetic field data and temperature signals on the battery surface, as well as the battery current and voltage, are collected simultaneously. By using synchronously acquired charging and discharging currents and calibrated geometric coupling coefficients, the Ampere background magnetic field generated by the current is reconstructed, thereby extracting the intrinsic spin magnetic field signal of the battery positive electrode active material. The magnetic signal health factor, characteristic peak parameters of the magnetic differential curve, and magnetic field-current hysteresis area are extracted from the intrinsic spin magnetic field signal; wherein, the magnetic signal health factor is: the ratio of the average magnetic flux density in the nth cycle full charge state to the average magnetic flux density in the first cycle full charge state; The characteristic peak parameters of the magnetic differential curve include: peak position shift and peak height attenuation of the characteristic peak on the differential curve obtained by differentiating the spin magnetic field signal with respect to voltage during battery charging and discharging. The magnetic field-current hysteresis area is the area enclosed by the hysteresis curve plotted with the average magnetic induction intensity as the vertical axis and the current as the horizontal axis during the battery charge-discharge cycle. Based on the magnetic features extracted from the intrinsic spin magnetic field signal, the battery's current and voltage, and the thermal features extracted from the temperature signal, a magnetic-thermal-electric multiphysics feature vector is constructed. Using a pre-trained multiphysics coupling model, typical battery failure modes are identified.

2. The battery failure identification method based on magnetic-thermal-electrical multi-signal as described in claim 1, characterized in that, By utilizing synchronously acquired charging and discharging currents and calibrated geometric coupling coefficients, the Ampere background magnetic field generated by the current is reconstructed, thereby extracting the intrinsic spin magnetic field signal of the battery's positive electrode active material. Specifically, the extraction of the intrinsic spin magnetic field signal involves: The magnetic field reference values ​​of each sensor channel are collected and determined under static zero current conditions, and the original magnetic field signals are debiased. The initial geometric coupling coefficients are calculated within the calibration window and dynamically updated using an exponentially weighted average during the loop. The intrinsic spin magnetic field signal of the positive electrode active material is obtained by subtracting the reconstructed Ampere background magnetic field from the debiased magnetic field signal.

3. The battery failure identification method based on magnetic-thermal-electrical multi-signal as described in claim 2, characterized in that, It also includes performing temperature compensation on the magnetic field signal, which includes zero-point temperature drift compensation and sensitivity temperature drift compensation in sequence. After compensation, a magnetic induction intensity signal that only reflects the changes in the external real magnetic field is obtained. The intrinsic spin magnetic field signal of the positive electrode active material of the battery is extracted by using the temperature-compensated magnetic induction intensity signal as the original magnetic field signal.

4. The battery failure identification method based on magnetic-thermal-electrical multi-signal as described in claim 1, characterized in that, The magnetic-thermal-electric multiphysics feature vector includes the magnetic signal health factor, the peak position shift of the characteristic peak, the peak height attenuation, the magnetic field-current hysteresis area, and the deviation of the highest surface temperature from the initial value. The magnetic-thermal-electric multiphysics feature vector is input into the multiphysics coupling model to obtain the typical failure modes of the battery. The multiphysics coupling model is trained by a support vector regression model or a random forest model.

5. The battery failure identification method based on magnetic-thermal-electrical multi-signal as described in claim 1, characterized in that, It also includes battery SOC estimation based on intrinsic spin magnetic field signal and temperature signal: battery SOC estimation adopts magnetic field-temperature SOC polynomial fitting and combines Kalman filtering to fuse current integral signal.

6. A battery failure identification system based on magnetic-thermal-electrical multi-signal, employing the battery failure identification method based on magnetic-thermal-electrical multi-signal as described in any one of claims 1-5, characterized in that, include: The synchronous acquisition module is configured to simultaneously acquire magnetic field data and temperature signals on the battery surface, as well as the battery current and voltage, during the battery charging and discharging process. The magnetic signal processing module is configured to: reconstruct the Ampere background magnetic field generated by the current using the synchronously acquired charging and discharging current and the calibrated geometric coupling coefficient, thereby extracting the intrinsic spin magnetic field signal of the battery positive electrode active material; The extraction module is configured to extract the magnetic signal health factor, characteristic peak parameters of the magnetic differential curve, and magnetic field-current hysteresis area from the intrinsic spin magnetic field signal. The identification module is configured to: construct a magnetic-thermal-electric multiphysics feature vector based on the magnetic features extracted from the intrinsic spin magnetic field signal, the battery's current and voltage, and the thermal features extracted from the temperature signal; and identify typical failure modes of the battery using a pre-trained multiphysics coupling model.

7. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, perform the method described in any one of claims 1-5.

9. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the method described in any one of claims 1-5.

Images