A method and apparatus for detecting battery defects

Abstract

The present disclosure provides a battery defect detection method and device, which belongs to the technical field of battery nondestructive testing. The present disclosure fixes the battery to be detected, and applies an external current excitation matched with the type to different types of batteries to be detected. The external current excitation is used to stimulate the battery to generate an electromagnetic response signal representing defects. The topographic signal and the electromagnetic response signal of the battery to be detected processed by the external current excitation are obtained, and the electromagnetic response signal is compensated and corrected by the topographic signal to obtain a compensated electromagnetic response signal. The multi-sensor information of the battery to be detected processed by the external current excitation is collected. Based on the compensated electromagnetic response signal and the multi-sensor information, a multi-dimensional feature matrix is constructed. The multi-dimensional feature matrix is input into a first model which has been pre-trained to obtain a target defect result. In summary, the technical solution provided by the present disclosure can improve the detection accuracy of battery defects and can adapt to various application scenarios.

Description

Technical Field

[0001] This disclosure relates to the field of non-destructive testing technology for batteries, and in particular to a method and apparatus for detecting battery defects. Background Technology

[0002] Currently, lithium-ion batteries, lithium metal batteries, and solid-state batteries are core energy storage devices in the new energy field. Their internal defects directly affect the battery's safety, capacity, and cycle life. However, existing battery testing methods are mostly limited to single detection methods: either using imaging technologies such as ultrasound and X-rays, which can identify macroscopic structural defects but cannot capture the coupling characteristics of electromagnetic effects and microscopic defects within the battery; or using electrochemical impedance spectroscopy, which can perform impedance analysis of electromagnetic effects within the battery, but has a narrow response range, cannot identify mid-to-high frequency domains, and is difficult to locate fine microscopic defects. This results in low accuracy in battery defect detection, making it difficult to adapt to various application scenarios. Summary of the Invention

[0003] This disclosure provides a method and apparatus for detecting battery defects, which to some extent solves the problem that existing battery defect detection methods have low accuracy and are difficult to adapt to various application scenarios.

[0004] According to one aspect of this disclosure, a method for detecting battery defects is provided. The method includes: fixing the battery to be tested and applying an external current excitation matching the type to different types of batteries to be tested; the external current excitation is used to excite an electromagnetic response signal characterizing the defect inside the battery; the types include: lithium-ion batteries, lithium metal batteries, and solid-state batteries; acquiring the topographic signal and electromagnetic response signal of the battery to be tested after external current excitation, and using the topographic signal to compensate and correct the electromagnetic response signal to obtain a compensated electromagnetic response signal; acquiring multi-sensor information of the battery to be tested after external current excitation; the multi-sensor information includes at least one of the following: temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal; constructing a multi-dimensional feature matrix based on the compensated electromagnetic response signal and the multi-sensor information; inputting the multi-dimensional feature matrix into a pre-trained first model to obtain a target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location, and defect degree.

[0005] Furthermore, according to one aspect of the method provided in this disclosure, the battery to be tested is fixed, and an external current excitation matching the type is applied to different types of batteries to be tested, including: fixing the battery to be tested using an in-situ battery clamp; when the battery to be tested is a lithium-ion battery, applying a sinusoidal external current excitation and superimposing it on a first current for charging and discharging the battery to be tested; when the battery to be tested is a lithium metal battery, applying a pulsed external current excitation separately; when the battery to be tested is a solid-state battery, applying a square wave external current excitation and superimposing it on a second current for charging and discharging the battery to be tested; the second current is less than the first current.

[0006] Furthermore, according to one aspect of the method provided in this disclosure, the topographic signal and electromagnetic response signal of the battery under test, after being processed by external current excitation, are acquired, and the electromagnetic response signal is compensated and corrected using the topographic signal to obtain a compensated electromagnetic response signal. This includes: simultaneously scanning the topographic and magnetic signals of the battery under test using a scanning magnetic force microscope to acquire the topographic signal and electromagnetic response signal; the topographic signal includes: surface height, morphological undulation information, and roughness distribution information; the electromagnetic response signal includes: magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution information; amplitude compensation is performed on the magnetic field amplitude based on the surface height; phase correction is performed on the magnetic field phase based on the morphological undulation information; distribution correction is performed on the magnetic field spatial distribution information based on the roughness distribution information; and the amplitude-compensated magnetic field amplitude, the phase-corrected magnetic field phase, and the distribution-corrected magnetic field spatial distribution information are determined as the compensated electromagnetic response signal.

[0007] Furthermore, according to one aspect of the method provided in this disclosure, the scanning magnetic force microscope includes a cobalt-nickel alloy coated microelectromechanical cantilever probe.

[0008] Furthermore, according to one aspect of the method provided in this disclosure, multi-sensor information of the battery under test after being subjected to external current excitation is acquired, including: acquiring temperature signals and fixed pressure signals respectively through temperature sensors and pressure sensors; and simultaneously acquiring battery voltage signals and external excitation current signals through an electrochemical acquisition unit.

[0009] Furthermore, according to one aspect of the method provided in this disclosure, a multi-dimensional feature matrix is ​​constructed based on the compensated electromagnetic response signal and multi-sensor information, including: converting the magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution features in the compensated electromagnetic response signal into magnetic signal feature vectors; normalizing the temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal to obtain sensor feature vectors; and fusing the magnetic signal feature vectors and sensor feature vectors to obtain a multi-dimensional feature matrix.

[0010] Furthermore, according to one aspect of the method provided in this disclosure, the first model includes a convolutional neural network and support vector machine fusion model; inputting a multi-dimensional feature matrix into the pre-trained first model to obtain the target defect result includes: inputting the multi-dimensional feature matrix into the convolutional neural network and support vector machine fusion model, and extracting the spatial features of the magnetic signal using the convolutional layers in the convolutional neural network and support vector machine fusion model; based on the spatial features of the magnetic signal, using the support vector machine classifier in the convolutional neural network and support vector machine fusion model to classify, locate, and determine the degree of the defect, thereby obtaining the target defect result.

[0011] Furthermore, according to one aspect of the method provided in this disclosure, when the spatial characteristics of the magnetic signal are linear magnetic field peak distribution, the defect type is determined to be current collector scratch; when the spatial characteristics of the magnetic signal are dendritic aggregation distribution, the defect type is determined to be lithium dendrite; when the spatial characteristics of the magnetic signal are periodic fluctuation distribution, the defect type is determined to be solid electrolyte interface film abnormality; and when the spatial characteristics of the magnetic signal are sheet-like magnetic field trough distribution, the defect type is determined to be active material shedding.

[0012] According to another aspect of this disclosure, a battery defect detection device is provided, comprising: an excitation unit for fixing the battery to be tested and applying an external current excitation matching the type to different types of batteries to be tested; the external current excitation is used to excite an electromagnetic response signal characterizing a defect inside the battery; the types include: lithium-ion batteries, lithium metal batteries, and solid-state batteries; an acquisition unit for acquiring the topographic signal and electromagnetic response signal of the battery to be tested after external current excitation, and using the topographic signal to compensate and correct the electromagnetic response signal to obtain a compensated electromagnetic response signal; a collection unit for acquiring multi-sensor information of the battery to be tested after external current excitation; the multi-sensor information includes at least one of the following: temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal; a construction unit for constructing a multi-dimensional feature matrix based on the compensated electromagnetic response signal and the multi-sensor information; and a determination unit for inputting the multi-dimensional feature matrix into a pre-trained first model to obtain a target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location, and defect degree.

[0013] Furthermore, according to another aspect of the apparatus of this disclosure, the excitation unit is specifically configured to: fix the battery to be tested using an in-situ battery clamp; when the battery to be tested is a lithium-ion battery, apply a sinusoidal external current excitation and superimpose it on the first current of the battery's charging and discharging; the first current is greater than the second current; when the battery to be tested is a lithium metal battery, apply a pulsed external current excitation separately; when the battery to be tested is a solid-state battery, apply a square wave external current excitation and superimpose it on the second current of the battery's charging and discharging.

[0014] This disclosure provides a method and apparatus for detecting battery defects. The method involves fixing the battery to be tested and applying an external current excitation matched to the type of the battery; the external current excitation is used to generate an electromagnetic response signal characterizing defects inside the battery; the types include lithium-ion batteries, lithium metal batteries, and solid-state batteries; acquiring the topographic signal and electromagnetic response signal of the battery after external current excitation, and using the topographic signal to compensate and correct the electromagnetic response signal to obtain a compensated electromagnetic response signal; collecting multi-sensor information of the battery after external current excitation; the multi-sensor information includes at least one of the following: temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal; constructing a multi-dimensional feature matrix based on the compensated electromagnetic response signal and multi-sensor information; inputting the multi-dimensional feature matrix into a pre-trained first model to obtain the target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location, and defect severity. In this way, compared to existing single detection methods (ultrasound and X-ray imaging cannot capture the coupling characteristics of electromagnetic effects and micro-defects, and electrochemical impedance spectroscopy cannot cover the mid-to-high frequency domain and is difficult to locate micro-defects), this disclosure, through a progressive logic of applying targeted external current excitation matched to the battery type, using terrain signals to correct electromagnetic response signals, fusing multi-sensor information to construct a feature matrix, and performing fusion analysis through a first model, can accurately solve the pain points of existing technologies. Specifically, it can be understood as follows: targeted excitation solves the problem of the inability to effectively excite mid-to-high frequency domain defect features; terrain correction eliminates signal interference and improves data accuracy; multi-dimensional fusion avoids missed and false detections; and model analysis enables accurate identification and location of electromagnetic effects and micro-defects. In summary, the technical solution provided by this disclosure can improve the detection accuracy of battery defects and can be adapted to various application scenarios.

[0015] It should be understood that both the foregoing general description and the following detailed description are exemplary and intended to provide further illustration of the claimed technology. Attached Figure Description

[0016] The above and other objects, features, and advantages of this disclosure will become more apparent from the more detailed description of the embodiments thereof in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the disclosure and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0017] Figure 1 A schematic flowchart of a battery defect detection method provided in an embodiment of this disclosure;

[0018] Figure 2This is a structural block diagram of a battery defect detection device provided in an embodiment of the present disclosure. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this disclosure more apparent, exemplary embodiments according to this disclosure will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this disclosure, and not all embodiments of this disclosure. It should be understood that this disclosure is not limited to the exemplary embodiments described herein.

[0020] Currently, lithium-ion batteries, lithium metal batteries, and solid-state batteries are core energy storage devices in the new energy field, and their internal defects directly affect battery safety, capacity, and cycle life. However, existing battery testing methods are mostly limited to single detection methods: either using imaging technologies such as ultrasound and X-rays, which can identify macroscopic structural defects but cannot capture the coupling characteristics of electromagnetic effects and microscopic defects inside the battery; or using electrochemical impedance spectroscopy, which can achieve impedance analysis of electromagnetic effects inside the battery, but has a narrow response range, cannot cover the mid-to-high frequency domain, and is difficult to locate fine microscopic defects. This results in low accuracy in battery defect detection, making it difficult to adapt to various application scenarios.

[0021] Therefore, to address the aforementioned technical problems, this disclosure provides a battery defect detection method. Compared to existing single detection methods (ultrasound and X-ray imaging cannot capture the coupling characteristics of electromagnetic effects and micro-defects, and electrochemical impedance spectroscopy cannot cover the mid-to-high frequency domain and is difficult to locate micro-defects), this disclosure, through a progressive logic of applying targeted external current excitation matched to the battery type, using terrain signals to correct electromagnetic response signals, fusing multi-sensor information to construct a feature matrix, and performing fusion analysis through a first model, can accurately solve the pain points of existing technologies. Specifically, this can be understood as follows: targeted excitation solves the problem of the inability to effectively excite mid-to-high frequency domain defect features; terrain correction eliminates signal interference and improves data accuracy; multi-dimensional fusion avoids missed and false detections; and model analysis achieves accurate identification and location of electromagnetic effects and micro-defects. In summary, the technical solution provided by this disclosure can improve the detection accuracy of battery defects and can be adapted to various application scenarios.

[0022] This disclosure provides a method for detecting battery defects. The method includes:

[0023] In step S101, the battery to be tested is fixed, and an external current excitation matching the type is applied to different types of batteries to be tested; the external current excitation is used to excite the generation of electromagnetic response signals characterizing defects inside the battery; the types include: lithium-ion batteries, lithium metal batteries and solid-state batteries;

[0024] In step S102, the terrain signal and electromagnetic response signal of the battery under test after being processed by external current excitation are acquired, and the electromagnetic response signal is compensated and corrected using the terrain signal to obtain the compensated electromagnetic response signal.

[0025] In step S103, multi-sensor information of the battery under test after external current excitation processing is acquired; the multi-sensor information includes at least one of the following: temperature signal, fixed pressure signal, battery voltage signal and external excitation current signal;

[0026] In step S104, a multi-dimensional feature matrix is ​​constructed based on the compensated electromagnetic response signal and multi-sensor information;

[0027] In step S105, the multi-dimensional feature matrix is ​​input into the pre-trained first model to obtain the target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location and defect degree.

[0028] In this disclosure, different types of batteries under test can be understood as energy storage batteries with different electrochemical systems and structures. These different types include lithium-ion batteries, lithium metal batteries, and solid-state batteries. Specifically, lithium-ion batteries rely on the insertion and extraction of lithium ions between the positive and negative electrodes to achieve charging and discharging; lithium metal batteries use metallic lithium as the negative electrode and have higher energy density; solid-state batteries use solid electrolytes, resulting in higher safety and stronger stability.

[0029] In this disclosure, external current excitation can be understood as a medium-high frequency or ultra-high frequency tunable excitation signal used to simulate the working state of the battery under fast charging and high power conditions, and to excite the electromagnetic response related to defects inside the battery.

[0030] In this disclosure, topographic signals can be understood as information on the surface morphology of the battery acquired by scanning magnetic force microscopy. This will be further elaborated below.

[0031] In this disclosure, the electromagnetic response signal can be understood as the magnetic field signal generated inside the battery under high-frequency excitation, which can directly reflect the location, morphology, and severity of internal defects. This will be elaborated upon below.

[0032] In this disclosure, the compensated electromagnetic response signal can be understood as a true and accurate magnetic field signal that has been corrected after eliminating surface morphology interference.

[0033] In this disclosure, multi-sensor information can be understood as multi-channel acquired information capable of reflecting the battery's operating state, environmental state, and excitation state. The multi-sensor information disclosed herein includes at least one of the following: temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal. Specifically, the temperature signal reflects changes in battery temperature, the fixed pressure signal reflects the battery's clamping state, the battery voltage signal reflects the battery's electrochemical state, and the external excitation current signal reflects the excitation input state.

[0034] In this disclosure, the multi-dimensional feature matrix can be understood as a comprehensive feature set formed by integrating magnetic field features, operating condition features, and environmental features.

[0035] In this disclosure, the first model can be understood as a machine learning model that has been pre-trained with a large number of samples and is used for intelligent identification of battery defects.

[0036] In this disclosure, the target defect result can be understood as the final identification result of internal defects in the battery. The target defect result of this disclosure includes: defect type, defect location, and defect severity. Specifically, the defect type includes lithium dendrites, solid electrolyte interface film abnormalities, current collector scratches, and active material shedding; the defect location is the coordinate region of the defect on the battery surface; and the defect severity includes slight, moderate, and severe.

[0037] Specifically, battery defect detection may include the following steps:

[0038] First, the battery to be tested is stably fixed in the in-situ battery fixture, ensuring that the battery testing surface is parallel to the scanning magnetic microscope probe and that the spacing is uniform.

[0039] Next, based on the type of battery to be tested (lithium-ion, lithium metal, or solid-state battery), a medium / ultra-high frequency external current excitation signal matching the battery type is set and output to excite the battery to generate an electromagnetic response related to defects.

[0040] Then, the scanning magnetic force microscope is started to simultaneously acquire topographic signals on the surface of the battery and electromagnetic response signals inside; using the acquired topographic signals, the amplitude, phase and spatial distribution of the electromagnetic response signals are compensated and corrected to obtain the compensated electromagnetic response signals after interference elimination.

[0041] Simultaneously, through temperature sensors, pressure sensors, and electrochemical acquisition units, the battery's temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal are collected to form multi-sensor information.

[0042] Furthermore, the magnetic signal features extracted from the compensated electromagnetic response signal are fused with the operating condition features extracted from multi-sensor information to construct a multi-dimensional feature matrix;

[0043] Finally, the multi-dimensional feature matrix is ​​input into the pre-trained first model. Through feature extraction and classification by the model, the target defect results containing defect type, defect location and defect degree are obtained, and finally the detection report and visualization map are output.

[0044] The following section will detail how to apply external current excitation to different types of batteries, including:

[0045] The battery to be tested is fixed in place using an in-situ battery clamp;

[0046] When the battery under test is a lithium-ion battery, a sinusoidal external current excitation is applied and superimposed on the first current of the battery's charging and discharging.

[0047] When the battery to be tested is a lithium metal battery, a pulsed external current excitation is applied separately.

[0048] When the battery under test is a solid-state battery, a square wave external current excitation is applied and superimposed on the second current of the battery under test during charging and discharging; the second current is less than the first current.

[0049] In this disclosure, the in-situ battery clamp can be understood as a dedicated clamping device that can stably clamp the battery and can simultaneously integrate temperature and pressure acquisition, or can not integrate temperature and pressure acquisition and support simultaneous charging, discharging and detection.

[0050] In this disclosure, the sinusoidal external current excitation can be understood as a high-frequency excitation signal with a continuous and smooth waveform. The first current can be understood as a relatively high-rate charge / discharge current of 1C-3C suitable for lithium-ion batteries.

[0051] In this disclosure, pulsed external current excitation can be understood as an intermittent, high-frequency excitation signal with instantaneous intensity.

[0052] In this disclosure, the square wave external current excitation can be understood as a step-like, rapidly switching high-frequency excitation signal. The second current can be understood as a relatively mild charge / discharge current of 0.5C-2C suitable for solid-state batteries.

[0053] Specifically, the following steps can be included when performing external current excitation:

[0054] First, place the battery to be tested into the original battery fixture and adjust the clamping force of the fixture to make the battery securely fixed and evenly stressed. If the fixture integrates temperature and pressure sensors, ensure that the sensors are in close contact with the battery surface. If not integrated, place the sensors separately.

[0055] Subsequently, the corresponding external current excitation parameters are set according to the battery type. For lithium-ion batteries, the excitation module outputs a sine wave signal of 10kHz~500kHz, and the current amplitude is adjusted to 0.1A~10A. At the same time, this sine wave excitation signal is superimposed on the first charge and discharge current of 1C~3C to ensure that the excitation and charge and discharge are synchronized. For lithium metal batteries, the excitation module outputs a pulse wave signal of 50kHz~800kHz with a duty cycle of 30%~50%, and the current amplitude is adjusted to 0.01A~5A. An independent excitation mode is adopted, which is not superimposed on the charge and discharge current to avoid affecting the stability of the lithium metal negative electrode. For solid-state batteries, the excitation module outputs a square wave signal of 100kHz~1GHz, and the current amplitude is adjusted to 0.5A~20A. This signal is superimposed on the second charge and discharge current of 0.5C~2C to adapt to the ion conduction characteristics of solid electrolytes.

[0056] The following will explain in detail how to obtain terrain signals and electromagnetic response signals, and how to perform compensation correction to obtain the compensated electromagnetic response signal, including:

[0057] Using a scanning magnetic force microscope, the battery under test was simultaneously scanned for topographic and magnetic signals, and topographic and electromagnetic response signals were acquired. The topographic signals included surface height, morphological undulation information, and roughness distribution information. The electromagnetic response signals included magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution information.

[0058] Amplitude compensation for the magnetic field is performed based on the surface height.

[0059] Phase correction of the magnetic field phase is performed based on topographic undulation information;

[0060] Based on the roughness distribution information, the spatial distribution information of the magnetic field is corrected.

[0061] The magnetic field amplitude after amplitude compensation, the magnetic field phase after phase correction, and the magnetic field spatial distribution information after distribution correction are determined as the compensated electromagnetic response signal.

[0062] In this disclosure, a scanning magnetic field microscope (SMD) can be understood as a high-precision instrument with nanometer-level spatial resolution capable of simultaneously acquiring surface morphology and magnetic field distribution. The SMD of this disclosure includes a cobalt-nickel alloy-coated MEMS cantilever probe, specifically a Co-Ni alloy-coated MEMS cantilever probe with high sensitivity and high magnetic field resolution, enabling high-precision synchronous acquisition of topographic and magnetic signals.

[0063] In this disclosure, the topographic signals obtained by scanning magnetic force microscopy include: surface height, morphological undulation information, and roughness distribution information. Specifically, the surface height is used to characterize the height variation of the battery surface, the morphological undulation is used to characterize the surface smoothness, and the roughness is used to characterize the surface micro-roughness.

[0064] In this disclosure, the electromagnetic response signal obtained by scanning magnetic force microscopy includes: magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution information. Specifically, the magnetic field amplitude reflects the magnitude of the magnetic field strength, the magnetic field phase reflects the signal timing and response characteristics, and the magnetic field spatial distribution reflects the distribution pattern of the magnetic field within the detection area.

[0065] In this disclosure, amplitude compensation can be understood as correcting the magnetic field amplitude based on the difference in surface height, thereby eliminating signal deviation caused by height changes.

[0066] In this disclosure, phase correction can be understood as correcting the magnetic field phase based on surface morphology fluctuations to eliminate phase shifts caused by morphology.

[0067] In this disclosure, distribution correction can be understood as smoothing and correcting the spatial distribution of the magnetic field based on roughness, thereby improving the authenticity and reliability of the distribution characteristics.

[0068] Specifically, obtaining the compensated electromagnetic response signal may include the following steps:

[0069] First, the scanning magnetic force microscope (SMP) is activated, and the cobalt-nickel alloy coated microelectromechanical cantilever probe is adjusted to a height of 50–120 nm above the battery detection surface. The scanning rate is set to 0.5–5 μm / s, the scanning range to 5 μm × 5 μm–200 μm × 200 μm, and the resolution to 512 × 512–1024 × 1024 pixels. Simultaneous scanning mode is then initiated. During scanning, the probe senses the changes in the battery surface height in real time, acquiring surface height data, morphological undulation curves, and roughness distribution data to form a complete topographic signal. Simultaneously, the probe senses the magnetic field generated inside the battery due to external current excitation through magnetic coupling, acquiring magnetic field amplitude, phase, and spatial distribution data to form the raw electromagnetic response signal. Then, the signal correction stage begins, based on the acquired surface height... The magnetic field amplitude is compensated using a linear correction algorithm based on surface height data. This eliminates the attenuation or enhancement deviations caused by different surface heights, ensuring that the amplitude data accurately reflects the internal magnetic field strength. Based on topographic undulation information, a phase compensation algorithm corrects the magnetic field phase, eliminating signal phase shifts caused by surface undulations and ensuring consistency between the phase signal and the actual temporal characteristics of the magnetic field. Based on roughness distribution information, a Gaussian smoothing algorithm corrects the spatial distribution of the magnetic field, filtering out stray signals from surface micro-roughness and highlighting the true shape of the magnetic field distribution. Finally, the amplitude-compensated magnetic field amplitude, the phase-corrected magnetic field phase, and the spatially corrected magnetic field distribution information are integrated to form a compensated electromagnetic response signal that eliminates surface topographic interference, used for subsequent feature extraction.

[0070] The following will explain in detail how to obtain multi-sensor information, including:

[0071] Temperature signals and fixed pressure signals are acquired using temperature sensors and pressure sensors, respectively.

[0072] The battery voltage signal and the external excitation current signal are simultaneously acquired through the electrochemical acquisition unit.

[0073] In this disclosure, a temperature sensor can be understood as a sensing element used to acquire the battery's operating temperature in real time, and the obtained temperature signal can be the real-time temperature value of the battery during excitation and charging / discharging processes.

[0074] In this disclosure, the pressure sensor can be understood as a sensing element that collects the pressure applied to the battery by the clamp, and the obtained fixed pressure signal can be understood as the clamping pressure that keeps the battery in a stable state.

[0075] In this disclosure, the electrochemical acquisition unit can be understood as a high-precision acquisition module for acquiring battery voltage and excitation current, and the obtained battery voltage signal and external excitation current signal can be the real-time operating voltage of the battery and the actual input excitation current.

[0076] Specifically, obtaining information from multiple sensors can include the following steps:

[0077] First, the temperature sensor is fixedly attached to the battery surface near the detection area, ensuring tight contact between the sensor and the battery surface to avoid air gaps affecting temperature measurement accuracy. The temperature sensor's acquisition frequency is set to 100Hz~1kHz to collect real-time temperature change data of the battery during external current excitation and charging / discharging processes, forming a temperature signal and recording the temperature fluctuation range and rate of change. Next, a pressure sensor is installed on the clamping end of the battery clamp, indirectly contacting the battery surface, to collect data on the fixed pressure applied by the clamp to the battery. The acquisition frequency is the same as that of the temperature sensor, forming a fixed pressure signal, ensuring the pressure is stable within the preset range of 0.5~2MPa. If pressure fluctuations occur, feedback is sent to the clamp adjustment mechanism for correction. The voltage acquisition terminal of the electrochemical acquisition unit is connected to the positive and negative terminals of the battery, and the current acquisition terminal is connected in series in the external current excitation circuit to synchronously acquire the real-time operating voltage of the battery and the actual input external excitation current data. The acquisition frequency is also set to 100Hz~1kHz to ensure that the acquired voltage, current signals are synchronized with the magnetic signals, temperature signals, and pressure signals. Finally, the acquired temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal are summarized and filtered to remove stray interference signals and form complete multi-sensor information for subsequent feature fusion.

[0078] The following will explain in detail how to obtain a multi-dimensional feature matrix, including:

[0079] The magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution characteristics in the compensated electromagnetic response signal are converted into magnetic signal feature vectors.

[0080] The temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal are normalized to obtain the sensor feature vector;

[0081] The feature vectors of the magnetic signal and the sensor are fused to obtain a multi-dimensional feature matrix.

[0082] In this disclosure, the magnetic signal feature vector can be understood as a one-dimensional feature sequence formed by digitizing features such as magnetic field amplitude, phase, and spatial distribution.

[0083] In this disclosure, the sensor feature vector can be understood as a feature sequence formed after normalizing multiple operating condition signals.

[0084] In this disclosure, feature fusion can be understood as splicing magnetic signal features and sensor features according to dimensions to form a comprehensive feature matrix containing defect information and operating condition information.

[0085] Specifically, obtaining a multi-dimensional feature matrix may include the following steps:

[0086] First, feature extraction is performed on the compensated electromagnetic response signal. Feature parameters such as peak value, mean, variance, and peak-to-peak value are extracted from the magnetic field amplitude; feature parameters such as phase shift, phase change rate, and phase difference are extracted from the magnetic field phase; and feature parameters such as distribution entropy, clustering coefficient, and feature region area are extracted from the magnetic field spatial distribution. All extracted magnetic signal feature parameters are digitized and arranged in a preset order to form a one-dimensional magnetic signal feature vector. The vector dimension is determined based on the number of extracted feature parameters to ensure complete coverage of magnetic field-related defect features. Subsequently, multi-sensor information is preprocessed using a min-max normalization algorithm to integrate temperature signals, fixed pressure signals, and battery voltage signals. The values ​​of the voltage signal and the external excitation current signal are mapped to the [0,1] interval to eliminate the influence of differences in the dimensions of different signals. After normalization, the mean, fluctuation amplitude, and trend of each signal are extracted and arranged in order to form a one-dimensional sensor feature vector. Finally, feature fusion is performed by feature splicing. The magnetic signal feature vector and the sensor feature vector are spliced ​​in order of dimension to form a two-dimensional multi-dimensional feature matrix. The rows of the matrix correspond to different feature parameters, and the columns correspond to different acquisition times or detection areas. This ensures that the matrix contains both the magnetic signal features of internal battery defects and the sensor features of the working state, providing comprehensive and accurate feature input for subsequent model recognition.

[0087] The first model disclosed herein includes a convolutional neural network and support vector machine fusion model. The following details how to utilize the first model to obtain the target defect result, including:

[0088] The multi-dimensional feature matrix is ​​input into the convolutional neural network support vector machine fusion model, and the convolutional layer in the convolutional neural network support vector machine fusion model is used to extract the spatial features of the magnetic signal.

[0089] Based on the spatial characteristics of magnetic signals, the support vector machine classifier in the convolutional neural network-support vector machine fusion model is used to classify, locate and determine the degree of defects, and obtain the target defect results.

[0090] In this disclosure, the convolutional neural network-support vector machine fusion model can be understood as an improved fusion model that combines the spatial feature extraction capability of convolutional neural networks with the high-precision classification capability of support vector machines.

[0091] In this disclosure, a convolutional layer can be understood as a network structure used to automatically extract features such as spatial distribution, peaks, fluctuations, and textures of magnetic signals.

[0092] In this disclosure, the spatial characteristics of magnetic signals can be understood as the linear, dendritic, periodic, or sheet-like distribution patterns of the magnetic field in space.

[0093] In this disclosure, the support vector machine classifier can be understood as a model structure used to accurately identify and classify defect types, locations, and severity.

[0094] Specifically, determining the target defect result may include the following steps:

[0095] First, the constructed multi-dimensional feature matrix is ​​adjusted to match the input format of the convolutional neural network / support vector machine fusion model, ensuring that the matrix dimension matches the number of neurons in the model's input layer. The adjusted multi-dimensional feature matrix is ​​then input into the model, passing through a convolutional layer. This layer uses a preset convolution kernel (e.g., 3×3 or 5×5) to perform convolution operations on the feature matrix, automatically extracting the spatial distribution features of the magnetic signal (e.g., linear peaks, dendritic clusters, periodic fluctuations, patchy troughs, etc.) while retaining the operating condition information from the sensor features. After convolution, a pooling layer performs dimensionality reduction using the max pooling algorithm to retain key features, remove redundant information, and reduce the model's computational load. The dimensionality-reduced feature vector is then input into a fully connected layer for feature integration and optimization, resulting in a comprehensive feature vector that integrates the spatial features of the magnetic signal and the sensor's operating condition features. The feature vector is then input into the support vector machine classifier in the model. The classifier matches and discriminates the feature vector based on pre-trained standard defect sample parameters. First, it identifies the defect type (current collector scratches, lithium dendrites, solid electrolyte interface film abnormalities, active material shedding). Then, based on the coordinate information of the spatial distribution of the magnetic signal, it locates the specific position of the defect within the battery detection area (accurate to the nanometer level). Finally, it combines at least one of the following: magnetic signal amplitude deviation (magnetic signal amplitude relative to a standard defect-free battery), electrochemical parameter change rate, etc., to assess the severity of the defect, which can be classified into three levels: slight (amplitude deviation ≤ 10%), moderate (10% < amplitude deviation ≤ 30%), and severe (amplitude deviation > 30%). Finally, the defect type, defect location, and defect severity are integrated to form the target defect result.

[0096] The following will elaborate on the different defect types, including:

[0097] When the spatial characteristics of the magnetic signal are linear magnetic field peak distribution, the defect type is determined to be a current collector scratch;

[0098] When the spatial characteristics of the magnetic signal are dendritic clusters, the defect type is determined to be lithium dendrites;

[0099] When the spatial characteristics of the magnetic signal are periodic fluctuations, the defect type is determined to be an anomaly in the solid electrolyte interface film.

[0100] When the spatial characteristics of the magnetic signal are a sheet-like distribution of magnetic field valleys, the defect type is determined to be the shedding of active material.

[0101] In this disclosure, the linear magnetic field peak distribution can be understood as a continuous, linearly extending peak region in space, with uniform peak line width (50~500nm), peak intensity 1.5~3 times higher than the surrounding normal region, and peak extension direction consistent with the extension direction of the current collector scratch, without obvious branches or random distribution. Furthermore, because the conductivity of the scratched area differs from the normal area when the current collector has a scratch defect, under external current excitation, this area will generate a local current concentration phenomenon, which can then form a linearly distributed magnetic field peak. Therefore, by identifying this linear magnetic field peak distribution, the defect type can be accurately determined to be a current collector scratch, and the size and depth of the scratch can be further determined based on the width and intensity of the peak.

[0102] In this disclosure, the dendritic aggregation distribution can be understood as the magnetic field amplitude exhibiting a tree-like shape in space, with a main peak region (corresponding to the root of the lithium dendrite), from which multiple branching secondary peaks extend. The branches intertwine and are randomly distributed, with the peak intensity gradually weakening from the main root to the branch ends, and the overall distribution area exhibiting an irregular tree-like morphology. Furthermore, since lithium dendrites are dendritic-grown metallic lithium structures, their conductivity is much higher than that of the surrounding electrolyte. Under external current excitation, lithium dendrites become local current channels, generating concentrated magnetic field signals and exhibiting the characteristics of a dendritic aggregation distribution. Therefore, this characteristic can be used to directly determine the defect type as lithium dendrites, and the growth degree of lithium dendrites can be assessed based on the number, length, and peak intensity of the branches.

[0103] In this disclosure, the periodic fluctuation distribution can be understood as the magnetic field amplitude exhibiting regular, periodic fluctuations in space, with a stable fluctuation frequency (consistent with the thickness fluctuation period of the solid electrolyte interfacial film), a small amplitude fluctuation range (deviation ≤20%), and a continuous planar distribution of the fluctuation region without obvious peaks or troughs. Furthermore, because the thickness of the solid electrolyte interfacial film (SEI film) exhibits periodic non-uniform distribution when it is abnormal, the resistance at the interface shows periodic changes. Under external current excitation, the magnetic field signal will fluctuate periodically with the resistance change, forming the characteristic of a periodic fluctuation distribution. Therefore, this characteristic can be used to determine that the defect type is a solid electrolyte interfacial film abnormality, and the degree of SEI film abnormality and thickness gradient can be judged based on the fluctuation frequency and amplitude deviation.

[0104] In this disclosure, the sheet-like magnetic field trough distribution can be understood as a continuous, sheet-like trough region in space where the magnetic field amplitude is clearly defined. The magnetic field amplitude within this region is reduced by more than 30% compared to the normal region, and the shape of the trough region is consistent with the shape of the area where the active material has detached, without significant peak interference. Furthermore, because the area where the active material has detached loses its electrical and magnetic response capabilities, it cannot generate a normal magnetic field signal under external current excitation, resulting in a sheet-like trough in the magnetic field amplitude. Therefore, by identifying this sheet-like magnetic field trough distribution, the defect type can be determined to be active material detachment, and the extent and severity of the active material detachment can be assessed based on the area of ​​the trough region and the degree of amplitude reduction.

[0105] Exemplary examples also include embodiments for defect detection of lithium-ion batteries, including:

[0106] A ternary lithium-ion battery is selected, fixed in the in-situ battery clamp, and a sinusoidal high-frequency excitation is applied and superimposed on a 2C charge / discharge current.

[0107] Topographic signals and electromagnetic response signals are acquired simultaneously using a scanning magnetic microscope, and the magnetic field amplitude, phase, and distribution are corrected using surface height, undulation, and roughness.

[0108] Simultaneously, temperature, pressure, voltage, and current information are collected;

[0109] Construct a multi-dimensional feature matrix and input it into the CNN-SVM model;

[0110] Based on the periodic fluctuation distribution characteristics, the model identifies solid electrolyte interface membrane anomalies and outputs the defect location and moderate defect level.

[0111] Exemplary examples also include embodiments for defect detection in solid-state batteries, including:

[0112] A sulfide solid-state battery was selected, clamped in the in-situ battery fixture, and a square wave high-frequency excitation was applied and superimposed on a 1C charge / discharge current.

[0113] Topographic and magnetic signals were acquired and corrected using a scanning magnetic microscope;

[0114] Simultaneously acquire information from multiple sensors;

[0115] Construct the feature matrix and input it into the model;

[0116] Based on the characteristics of the linear magnetic field peak distribution, the model identifies the current collector scratches and outputs the defect location, size, and minor defect level.

[0117] Exemplary examples also include embodiments for defect detection in lithium metal batteries, including:

[0118] A lithium metal battery is selected, fixed in place by a battery clamp, and independently excited by a pulse wave.

[0119] Topographic and magnetic signal acquisition and correction were performed using a scanning magnetic microscope.

[0120] Collect information from multiple sensors and construct a multi-dimensional feature matrix;

[0121] After inputting the CNN-SVM model, the model identifies lithium dendrite defects based on the dendritic clustering distribution characteristics and outputs the defect location, size, and moderate defect level.

[0122] This disclosure also provides a battery defect detection device. Figure 2 This is a structural block diagram of a battery defect detection device provided in an embodiment of the present disclosure, such as... Figure 2 As shown, the battery defect detection device 200 includes:

[0123] The excitation unit 201 is used to fix the battery under test and apply an external current excitation that matches the type to different types of batteries under test; the external current excitation is used to excite the generation of electromagnetic response signals that characterize defects inside the battery; the types include: lithium-ion batteries, lithium metal batteries and solid-state batteries;

[0124] The acquisition unit 202 is used to acquire the terrain signal and electromagnetic response signal of the battery under test after being processed by external current excitation, and to use the terrain signal to compensate and correct the electromagnetic response signal to obtain a compensated electromagnetic response signal.

[0125] The acquisition unit 203 is used to acquire multi-sensor information of the battery under test after external current excitation processing; the multi-sensor information includes at least one of the following: temperature signal, fixed pressure signal, battery voltage signal and external excitation current signal;

[0126] The building unit 204 is used to construct a multi-dimensional feature matrix based on the compensated electromagnetic response signal and multi-sensor information;

[0127] The determination unit 205 is used to input the multi-dimensional feature matrix into the pre-trained first model to obtain the target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location and defect degree.

[0128] In one exemplary embodiment, the excitation unit 201 is specifically used to: fix the battery to be tested using an in-situ battery clamp; when the battery to be tested is a lithium-ion battery, apply a sinusoidal external current excitation and superimpose it on the first current of the battery charging and discharging; when the battery to be tested is a lithium metal battery, apply a pulsed external current excitation separately; when the battery to be tested is a solid-state battery, apply a square wave external current excitation and superimpose it on the second current of the battery charging and discharging; the second current is less than the first current.

[0129] In one exemplary embodiment, the acquisition unit 202 is specifically used to: simultaneously scan the topography and magnetic signals of the battery under test using a scanning magnetic force microscope, and acquire topography signals and electromagnetic response signals; the topography signals include: surface height, topographic undulation information, and roughness distribution information; the electromagnetic response signals include: magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution information; based on the surface height, perform amplitude compensation on the magnetic field amplitude; based on the topographic undulation information, perform phase correction on the magnetic field phase; based on the roughness distribution information, perform distribution correction on the magnetic field spatial distribution information; and determine the amplitude-compensated magnetic field amplitude, the phase-corrected magnetic field phase, and the distribution-corrected magnetic field spatial distribution information as the compensated electromagnetic response signals.

[0130] In one exemplary embodiment, the acquisition unit 202 is specifically used for: a scanning magnetic force microscope including a cobalt-nickel alloy coated microelectromechanical cantilever probe.

[0131] In one exemplary embodiment, the acquisition unit 203 is specifically used to: acquire temperature signals and fixed pressure signals respectively through a temperature sensor and a pressure sensor; and simultaneously acquire battery voltage signals and external excitation current signals through an electrochemical acquisition unit.

[0132] In one exemplary embodiment, the construction unit 204 is specifically used to: convert the magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution characteristics in the compensated electromagnetic response signal into a magnetic signal feature vector; normalize the temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal to obtain a sensor feature vector; and fuse the magnetic signal feature vector and the sensor feature vector to obtain a multi-dimensional feature matrix.

[0133] In one exemplary embodiment, the determining unit 205 is specifically used to: input a multi-dimensional feature matrix into a convolutional neural network support vector machine fusion model, and extract the spatial features of the magnetic signal using the convolutional layer in the convolutional neural network support vector machine fusion model; based on the spatial features of the magnetic signal, use the support vector machine classifier in the convolutional neural network support vector machine fusion model to classify, locate and determine the degree of defects, and obtain the target defect result.

[0134] In one exemplary embodiment, the determining unit 205 is specifically used to: determine the defect type as current collector scratch when the spatial characteristics of the magnetic signal are linear magnetic field peak distribution; determine the defect type as lithium dendrite when the spatial characteristics of the magnetic signal are dendritic aggregation distribution; determine the defect type as solid electrolyte interface film abnormality when the spatial characteristics of the magnetic signal are periodic fluctuation distribution; and determine the defect type as active material shedding when the spatial characteristics of the magnetic signal are sheet-like magnetic field trough distribution.

[0135] In summary, this disclosure provides a method and apparatus for detecting battery defects. This disclosure involves fixing the battery to be tested and applying an external current excitation matched to the type of the battery; the external current excitation is used to generate an electromagnetic response signal characterizing defects inside the battery; the types include: lithium-ion batteries, lithium metal batteries, and solid-state batteries; acquiring the topographic signal and electromagnetic response signal of the battery after external current excitation, and using the topographic signal to compensate and correct the electromagnetic response signal to obtain a compensated electromagnetic response signal; collecting multi-sensor information of the battery after external current excitation; the multi-sensor information includes at least one of the following: temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal; constructing a multi-dimensional feature matrix based on the compensated electromagnetic response signal and multi-sensor information; inputting the multi-dimensional feature matrix into a pre-trained first model to obtain the target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location, and defect severity. In this way, compared to existing single detection methods (ultrasound and X-ray imaging cannot capture the coupling characteristics of electromagnetic effects and micro-defects, and electrochemical impedance spectroscopy cannot cover the mid-to-high frequency domain and is difficult to locate micro-defects), this disclosure, through a progressive logic of applying targeted external current excitation matched to the battery type, using terrain signals to correct electromagnetic response signals, fusing multi-sensor information to construct a feature matrix, and performing fusion analysis through a first model, can accurately solve the pain points of existing technologies. Specifically, it can be understood as follows: targeted excitation solves the problem of the inability to effectively excite mid-to-high frequency domain defect features; terrain correction eliminates signal interference and improves data accuracy; multi-dimensional fusion avoids missed and false detections; and model analysis enables accurate identification and location of electromagnetic effects and micro-defects. In summary, the technical solution provided by this disclosure can improve the detection accuracy of battery defects and can be adapted to various application scenarios.

[0136] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed 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 disclosure.

[0137] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0138] The block diagrams of devices, apparatuses, devices, and systems disclosed herein are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0139] Additionally, as used herein, the "or" used in a list of items beginning with "at least one" indicates a separate list, such that a list of, for example, "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not imply that the described example is preferred or better than other examples.

[0140] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0141] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0142] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0143] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A method of detecting a defect of a battery, characterized by, The method includes: The battery to be tested is fixed, and an external current excitation matching the type is applied to the battery of different types; the external current excitation is used to excite the generation of electromagnetic response signals characterizing defects inside the battery; the types include: lithium-ion batteries, lithium metal batteries, and solid-state batteries; The terrain signal and the electromagnetic response signal of the battery under test after being processed by the external current excitation are acquired, and the electromagnetic response signal is compensated and corrected using the terrain signal to obtain a compensated electromagnetic response signal. The system acquires multi-sensor information of the battery under test after processing with the external current excitation; the multi-sensor information includes: temperature signal, fixed pressure signal, battery voltage signal, and external excitation current signal; Based on the compensated electromagnetic response signal and the multi-sensor information, a multi-dimensional feature matrix is ​​constructed; The multi-dimensional feature matrix is ​​input into a pre-trained first model to obtain the target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location, and defect severity; The step of acquiring the terrain signal and the electromagnetic response signal of the battery under test after being processed by the external current excitation, and using the terrain signal to compensate and correct the electromagnetic response signal to obtain a compensated electromagnetic response signal, includes: Using a scanning magnetic force microscope, the battery under test is simultaneously scanned for topographic and magnetic signals to acquire the topographic signal and the electromagnetic response signal. The topographic signal includes: surface height, morphological undulation information, and roughness distribution information. The electromagnetic response signal includes: magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution information. The magnetic field amplitude is compensated based on the surface height. Based on the morphological undulation information, the phase of the magnetic field is corrected. Based on the roughness distribution information, the spatial distribution information of the magnetic field is corrected. The magnetic field amplitude after amplitude compensation, the magnetic field phase after phase correction, and the magnetic field spatial distribution information after distribution correction are determined as the compensated electromagnetic response signal; The process of fixing the battery under test and applying an external current excitation matching the type to different types of the battery under test includes: The battery to be tested is fixed using an in-situ battery clamp; When the battery under test is a lithium-ion battery, a sinusoidal external current excitation is applied and superimposed on the first current of the charging and discharging of the battery under test; When the battery to be tested is a lithium metal battery, a pulsed external current excitation is applied separately. When the battery under test is a solid-state battery, a square wave external current excitation is applied and superimposed on the second current of the battery under test during charging and discharging; the second current is less than the first current.

2. The method of claim 1, wherein, The scanning magnetic microscope includes a cobalt-nickel alloy coated microelectromechanical cantilever probe.

3. The method according to claim 1, characterized in that, The acquisition of multi-sensor information of the battery under test after being processed by the external current excitation includes: The temperature signal and the fixed pressure signal are acquired by a temperature sensor and a pressure sensor, respectively. The battery voltage signal and the external excitation current signal are simultaneously acquired through the electrochemical acquisition unit.

4. The method according to claim 1, characterized in that, The construction of a multi-dimensional feature matrix based on the compensated electromagnetic response signal and the multi-sensor information includes: The magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution characteristics in the compensated electromagnetic response signal are converted into magnetic signal feature vectors. The temperature signal, the fixed pressure signal, the battery voltage signal, and the external excitation current signal are normalized to obtain the sensor feature vector; The magnetic signal feature vector and the sensor feature vector are fused to obtain the multi-dimensional feature matrix.

5. The method according to claim 1, characterized in that, The first model includes a convolutional neural network and support vector machine fusion model; the step of inputting the multi-dimensional feature matrix into the pre-trained first model to obtain the target defect result includes: The multi-dimensional feature matrix is ​​input into the convolutional neural network support vector machine fusion model, and the convolutional layers in the convolutional neural network support vector machine fusion model are used to extract the spatial features of the magnetic signal. Based on the spatial features of the magnetic signal, the support vector machine classifier in the convolutional neural network support vector machine fusion model is used to classify, locate and determine the degree of defects, so as to obtain the target defect result.

6. The method according to claim 5, characterized in that, When the spatial characteristics of the magnetic signal are linear magnetic field peak distribution, the defect type is determined to be a current collector scratch; When the spatial characteristics of the magnetic signal are dendritic clusters, the defect type is determined to be lithium dendrites; When the spatial characteristics of the magnetic signal are periodic fluctuations, the defect type is determined to be an anomaly in the solid electrolyte interface film. When the spatial characteristics of the magnetic signal are a sheet-like magnetic field trough distribution, the defect type is determined to be the shedding of active material.

7. A battery defect detection device, characterized in that, The device includes: An excitation unit is used to fix the battery under test and apply an external current excitation matching the type to different types of the battery under test; the external current excitation is used to excite the generation of an electromagnetic response signal characterizing defects inside the battery; the types include: lithium-ion battery, lithium metal battery and solid-state battery; The acquisition unit is used to acquire the terrain signal and the electromagnetic response signal of the battery under test after being processed by the external current excitation, and to use the terrain signal to compensate and correct the electromagnetic response signal to obtain a compensated electromagnetic response signal. The acquisition unit is used to acquire multi-sensor information of the battery under test after the external current excitation processing; the multi-sensor information includes: temperature signal, fixed pressure signal, battery voltage signal and external excitation current signal; A construction unit is used to construct a multi-dimensional feature matrix based on the compensated electromagnetic response signal and the multi-sensor information; A determining unit is used to input the multi-dimensional feature matrix into a pre-trained first model to obtain the target defect result; the first model is used to perform fusion defect analysis on the multi-dimensional feature matrix; the target defect result includes: defect type, defect location, and defect severity; Specifically, the acquisition unit is used for: Using a scanning magnetic force microscope, the battery under test is simultaneously scanned for topographic and magnetic signals to acquire the topographic signal and the electromagnetic response signal. The topographic signal includes: surface height, morphological undulation information, and roughness distribution information. The electromagnetic response signal includes: magnetic field amplitude, magnetic field phase, and magnetic field spatial distribution information. The magnetic field amplitude is compensated based on the surface height. Based on the morphological undulation information, the phase of the magnetic field is corrected. Based on the roughness distribution information, the spatial distribution information of the magnetic field is corrected. The magnetic field amplitude after amplitude compensation, the magnetic field phase after phase correction, and the magnetic field spatial distribution information after distribution correction are determined as the compensated electromagnetic response signal; The excitation unit is specifically used for: The battery to be tested is fixed using an in-situ battery clamp; When the battery under test is a lithium-ion battery, a sinusoidal external current excitation is applied and superimposed on the first current of the charging and discharging of the battery under test; the first current is greater than the second current. When the battery to be tested is a lithium metal battery, a pulsed external current excitation is applied separately. When the battery under test is a solid-state battery, a square wave external current excitation is applied and superimposed on the second current of the battery under test during charging and discharging.

Images