Nondestructive testing method and device for lithium precipitation of lithium battery based on ultrasonic guided waves

Abstract

The invention relates to the field of battery detection, and discloses a lithium battery lithium precipitation nondestructive testing method and device based on ultrasonic guided waves, and the method comprises the steps: analyzing the lithium precipitation state of a lithium battery to obtain lithium precipitation state characteristics, and calculating the lithium precipitation detection confidence coefficient corresponding to the lithium precipitation state characteristics; analyzing potential lithium precipitation risk characteristics of the lithium battery, querying lithium precipitation induction parameters corresponding to the potential lithium precipitation risk characteristics, and performing guided wave signal reproduction detection on a key electrode area of the lithium battery to obtain signal reproduction data; analyzing lithium precipitation gradient modes of the lithium battery in different charging and discharging stages, and dividing lithium precipitation detection grades corresponding to key electrode areas; and configuring a detection precision guided wave parameter corresponding to the lithium battery to execute detection processing on the key electrode area, positioning a signal abnormal area of the lithium battery in the detection process, calculating a signal feature matching degree corresponding to the signal abnormal area, and generating a lithium precipitation detection result of the lithium battery. The accuracy of lithium separation detection of the lithium battery can be improved.

Description

Technical Field

[0001] The present invention relates to a nondestructive detection method and device for lithium deposition in lithium batteries based on ultrasonic guided waves, and belongs to the field of battery detection. Background Art

[0002] Lithium plating in lithium batteries is a key issue affecting battery safety and cycle life. Lithium plating can cause battery capacity decay and increased internal resistance. In severe cases, it may lead to the risk of thermal runaway. It has important safety warning significance in the fields of new energy vehicles, energy storage power stations, etc.

[0003] At present, the commonly used method for detecting lithium plating in lithium batteries is mainly the disassembly detection method. This method is to judge the degree of lithium plating by physically disassembling the lithium battery, observing the lithium plating morphology on the electrode surface and detecting relevant electrochemical parameters. However, this method will cause irreversible damage to the battery, and the disassembly process is time-consuming, and it is impossible to achieve real-time monitoring of in-service batteries, which leads to a lag in the lithium plating risk warning. Therefore, a non-destructive detection method for lithium plating in lithium batteries based on ultrasonic guided waves is needed to improve the accuracy of lithium plating detection in lithium batteries. Summary of the Invention

[0004] The present invention provides a nondestructive detection method and device for lithium deposition in lithium batteries based on ultrasonic guided waves, the main purpose of which is to improve the accuracy of lithium deposition detection in lithium batteries.

[0005] To achieve the above objectives, the present invention provides a nondestructive detection method for lithium deposition in lithium batteries based on ultrasonic guided waves, comprising: Acquire an ultrasonic guided wave detection signal and battery cycle history data of a lithium battery to be detected, analyze the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data, obtain a lithium deposition state characteristic, and calculate a lithium deposition detection confidence corresponding to the lithium deposition state characteristic; Based on the lithium deposition detection confidence, analyzing the potential lithium deposition risk characteristics of the lithium battery, querying the lithium deposition induction parameters corresponding to the potential lithium deposition risk characteristics, and performing guided wave signal reproduction detection on the key electrode area of ​​the lithium battery based on the lithium deposition induction parameters to obtain signal reproduction data; Based on the signal reproduction data, analyzing the lithium plating transition mode of the lithium battery at different charge and discharge stages, and dividing the lithium plating detection level corresponding to the key electrode area based on the lithium plating transition mode and the current guided wave signal distribution state of the lithium battery; Based on the lithium plating detection level, the detection accuracy waveguide parameters corresponding to the lithium battery are configured to perform detection processing on the key electrode area, and locate the signal abnormality area of ​​the lithium battery during the detection process, and calculate the signal feature matching degree corresponding to the signal abnormality area. When the signal feature matching degree is greater than the preset matching degree, the detection is stopped to generate the lithium plating detection result of the lithium battery.

[0006] Optionally, the analyzing the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data to obtain lithium deposition state characteristics includes: extracting a wave velocity attenuation characteristic corresponding to the ultrasonic guided wave detection signal; Calculate the capacity attenuation trend corresponding to the battery cycle history data; Associating a mapping relationship between the wave velocity attenuation characteristic and the capacity attenuation trend; Analyzing the lithium plating degree of the lithium battery based on the mapping relationship; Based on the lithium deposition degree level, the lithium deposition state of the lithium battery is comprehensively analyzed to obtain the lithium deposition state characteristics.

[0007] Optionally, the calculating the lithium deposition detection confidence corresponding to the lithium deposition state feature includes: Performing multi-band decomposition processing on the lithium deposition state characteristics to obtain a frequency domain feature vector group; Calculating the signal strength change rate and characteristic offset rate corresponding to each frequency band in the frequency domain characteristic vector group in adjacent detection cycles; Querying the battery operating state corresponding to each frequency band in the frequency domain feature vector group, and collecting environmental parameters and electrochemical parameters corresponding to the battery operating state; Calculating the acoustic wave attenuation coefficient corresponding to the battery operating state by combining the environmental parameters and the electrochemical parameters; Combining the signal intensity change rate, the characteristic offset rate, and the acoustic wave attenuation coefficient, the lithium deposition detection confidence corresponding to the lithium deposition state feature is calculated using the following formula, including:

[0008] Among them, A represents the confidence level of lithium deposition detection corresponding to the lithium deposition state characteristics, Indicates the characteristic weight of the lithium deposition state feature in frequency band a, Indicates the rate of change of signal strength of frequency band a at time t, Indicates the attenuation reference value of the a-th frequency band, Indicates the detection cycle time interval, represents the offset suppression gain coefficient, Indicates the frequency offset safety threshold, represents the characteristic offset rate of the ath frequency band at time t, represents the attenuation adjustment coefficient, Indicates the sound wave attenuation coefficient of the ath frequency band, a represents the frequency band index, and q represents the number of frequency bands.

[0009] Optionally, performing guided wave signal reproduction detection on a key electrode region of the lithium battery based on the lithium deposition induction parameter to obtain signal reproduction data includes: Querying the signal reproduction detection standard corresponding to the lithium precipitation induction parameter; Screening the key electrode regions of the lithium battery based on the signal recurrence detection standard; According to the key electrode area, monitoring the guided wave signal characteristics corresponding to each dimension of the key electrode area; Calculating the deviation between the waveguide signal characteristic and a preset signal reference value to obtain a signal characteristic deviation; Based on the signal characteristic deviation, a waveguide signal reproduction detection is performed on the key electrode area of ​​the lithium battery to obtain signal reproduction data.

[0010] Optionally, analyzing the lithium deposition transition mode of the lithium battery at different charge and discharge stages based on the signal reproduction data includes: extracting a recurring signal feature from the signal recurrence data; Determining a lithium plating sensitive frequency band corresponding to the lithium battery according to the reproduced signal characteristics; Quantifying the signal amplitude change rate corresponding to the lithium plating sensitive frequency band; Evaluating local lithium deposition differences corresponding to the lithium battery based on the signal amplitude change rate; Based on the local lithium deposition differences, the lithium deposition transition modes of the lithium battery at different charge and discharge stages are analyzed.

[0011] Optionally, the dividing the lithium deposition detection level corresponding to the key electrode area based on the lithium deposition transition mode and the current guided wave signal distribution state of the lithium battery includes: Extracting the signal transition rate in the lithium deposition transition mode; Based on the current distribution state of the guided wave signal, locating the guided wave signal concentration area within the key electrode area; Calculating the signal transition rate and the spatiotemporal coupling coefficient of the waveguide signal concentration area; distinguishing risk gradient zones within the key electrode region based on the spatiotemporal coupling coefficient; The gradient distribution characteristics corresponding to the risk gradient zone are analyzed to divide the lithium plating detection level corresponding to the key electrode area.

[0012] Optionally, the calculating the spatiotemporal coupling coefficient between the signal transition rate and the waveguide signal concentration area includes: Splitting the signal change rate by physical dimension to obtain a multi-dimensional signal feature vector; Normalizing the multidimensional signal feature vector to obtain a target signal feature vector; Calculating the geometric aggregation eigenvalue corresponding to the waveguide signal aggregation area, and normalizing the geometric aggregation eigenvalue to obtain a target aggregation eigenvalue; Performing spatiotemporal weight allocation processing on the target signal eigenvector and the target aggregate eigenvalue respectively to obtain a first spatiotemporal weight and a second spatiotemporal weight; Combining the target signal eigenvector, the target aggregation eigenvalue, the first spatiotemporal weight, and the second spatiotemporal weight, the following formula is used to calculate the spatiotemporal coupling coefficient between the signal transition rate and the waveguide signal aggregation area, including:

[0013] Where C represents the spatiotemporal coupling coefficient between the signal transition rate and the waveguide signal aggregation area, represents the first spatiotemporal weight of the dth vector in the target signal feature vector, represents the dth vector in the target signal feature vector, represents the second spatiotemporal weight of the mth eigenvalue in the target aggregated eigenvalue, represents the mth eigenvalue in the target aggregate eigenvalue, represents the spatiotemporal correlation factor, d and m represent the sequence numbers corresponding to the target signal feature vector and the target aggregate feature value, respectively, and q and p represent the number of target signal feature vectors and the target aggregate feature value, respectively.

[0014] Optionally, configuring the detection accuracy waveguide parameters corresponding to the lithium battery to perform detection processing on the key electrode area and locate the signal abnormality area of ​​the lithium battery during the detection process also includes: Constructing a waveguide detection signal corresponding to the detection accuracy waveguide parameter; Injecting the guided wave detection signal into the key electrode area of ​​the lithium battery and collecting the reflected guided wave signal of the key electrode area; performing noise reduction processing on the reflected waveguide signal to obtain a noise-reduced waveguide signal, and extracting a signal time-frequency feature of the noise-reduced waveguide signal; Calculating the signal deviation corresponding to the noise-reduced waveguide signal based on the signal time-frequency characteristics; Based on the signal deviation, an abnormal signal area of ​​the lithium battery during the detection process is located.

[0015] Optionally, calculating the signal feature matching degree corresponding to the signal abnormal area includes: Performing feature tensor mapping processing on the abnormal signal area to obtain a three-dimensional tensor in the time-frequency domain; Based on the three-dimensional tensor in the time-frequency domain, construct a signal feature map corresponding to the abnormal signal area; Performing graph convolution feature extraction processing on the signal feature graph to obtain a high-order structural feature vector; Matching the high-order structural feature vector with the standard feature vector in a preset lithium precipitation feature library to obtain a similarity matrix; The dynamic similarity matrix is ​​corrected to obtain the signal feature matching degree corresponding to the signal abnormal area.

[0016] In order to solve the above problems, the present invention also provides a non-destructive detection device for lithium deposition in lithium batteries based on ultrasonic guided waves, the device comprising: a lithium deposition detection confidence calculation module, configured to obtain an ultrasonic guided wave detection signal and battery cycle history data of a lithium battery to be detected, analyze the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data, obtain a lithium deposition state characteristic, and calculate a lithium deposition detection confidence corresponding to the lithium deposition state characteristic; a reproduction detection processing module, configured to analyze the potential lithium deposition risk characteristics of the lithium battery based on the lithium deposition detection confidence level, query the lithium deposition induction parameters corresponding to the potential lithium deposition risk characteristics, and perform guided wave signal reproduction detection on the key electrode area of ​​the lithium battery based on the lithium deposition induction parameters to obtain signal reproduction data; a lithium deposition detection level classification module, configured to analyze the lithium deposition transition mode of the lithium battery at different charge and discharge stages based on the signal reproduction data, and classify the lithium deposition detection level corresponding to the key electrode area based on the lithium deposition transition mode and the current guided wave signal distribution state of the lithium battery; The lithium plating detection processing module is used to configure the detection accuracy waveguide parameters corresponding to the lithium battery based on the lithium plating detection level, so as to perform detection processing on the key electrode area, locate the signal abnormality area of ​​the lithium battery during the detection process, calculate the signal feature matching degree corresponding to the signal abnormality area, and stop the detection when the signal feature matching degree is greater than the preset matching degree, and generate the lithium plating detection result of the lithium battery.

[0017] Compared with the problems described in the background technology, the present invention analyzes the lithium plating state of the lithium battery by combining ultrasonic guided wave detection signals and battery cycle history data, and can capture lithium plating characteristics from two dimensions of "current physical state" and "historical loss trend". The lithium plating state characteristics obtained in this way are more comprehensive, providing a basis for subsequent processing. Furthermore, based on the lithium plating detection confidence, the present invention analyzes the potential lithium plating risk characteristics of the lithium battery, and can accurately locate the risk areas where lithium plating may occur in different confidence intervals, clarify the risk differences under different electrode positions and working conditions, and provide a key basis for evaluating the sensitivity and safety of battery lithium plating. Furthermore, based on the present invention, Based on the signal reproduction data, the lithium plating transition mode of the lithium battery in different charge and discharge stages is analyzed, which can deeply explore the dynamic change law of the lithium plating phenomenon of the lithium battery during the charge and discharge process, clarify the development trend and characteristic differences of lithium plating in different stages, and provide a key basis for evaluating the lithium plating risk and safety of the lithium battery. Finally, the present invention is based on the lithium plating detection level, configures the detection accuracy guided wave parameters corresponding to the lithium battery, and can construct a differentiated detection strategy system for lithium plating scenarios with different risk levels, from the conventional detection mode of low risk level to the high-precision deep detection mode of high risk level, to achieve accurate allocation of detection resources and improve the accuracy of lithium plating detection of lithium batteries. Therefore, the non-destructive detection method and device for lithium plating of lithium batteries based on ultrasonic guided waves provided in the embodiment of the present invention can improve the accuracy of lithium plating detection of lithium batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic flow chart of a nondestructive testing method for lithium deposition in lithium batteries based on ultrasonic guided waves according to one embodiment of the present invention; Figure 2 A schematic diagram of a recurrence detection process in a nondestructive detection method for lithium deposition in lithium batteries based on ultrasonic guided waves provided in one embodiment of the present invention; Figure 3 A schematic diagram of a module for implementing a nondestructive testing device for lithium deposition in lithium batteries based on ultrasonic guided waves, provided in one embodiment of the present invention.

[0019] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0020] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0021] The embodiments of the present application provide a nondestructive testing method for lithium deposition in lithium batteries based on ultrasonic guided waves. The execution subject of the nondestructive testing method for lithium deposition in lithium batteries based on ultrasonic guided waves includes, but is not limited to, at least one of electronic devices such as a server and a terminal that can be configured to execute the method provided in the embodiments of the present application. In other words, the nondestructive testing method for lithium deposition in lithium batteries based on ultrasonic guided waves can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0022] Reference Figure 1 FIG2 is a flow chart of a nondestructive detection method for lithium deposition in a lithium battery based on ultrasonic guided waves according to an embodiment of the present invention. In this embodiment, the nondestructive detection method for lithium deposition in a lithium battery based on ultrasonic guided waves includes: S1. Obtain an ultrasonic guided wave detection signal and battery cycle history data of a lithium battery to be detected, analyze the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data, obtain a lithium deposition state characteristic, and calculate a lithium deposition detection confidence corresponding to the lithium deposition state characteristic.

[0023] The present invention analyzes the lithium deposition state of the lithium battery by combining ultrasonic guided wave detection signals and battery cycle history data, and can capture lithium deposition characteristics from two dimensions: "current physical state" and "historical loss trend". The lithium deposition state characteristics obtained in this way are more comprehensive, providing a basis for subsequent processing.

[0024] The ultrasonic guided wave detection signal is signal data generated when a lithium battery is tested by an ultrasonic guided wave detection device, such as the reflected wave amplitude, guided wave propagation velocity, and signal attenuation coefficient of the lithium deposition area within the lithium battery. These signals can directly reflect the physical state of lithium deposition within the battery. The battery cycle history data is historical records accumulated by the lithium battery during the charge and discharge cycle, such as the number of cycles, the charge and discharge rate of each cycle, the cut-off voltage, and the capacity decay rate. These data can reflect the aging and loss trend of the battery. The lithium deposition state characteristics are characteristic indicators that can reflect the degree, distribution, and morphology of lithium deposition in the lithium battery, such as the area ratio, average thickness, and number of lithium deposition points of the lithium deposition area. The lithium deposition detection confidence is a quantitative indicator that measures the reliability of the lithium deposition state characteristic analysis results. The higher the value, the more reliable the analysis result. Optionally, the ultrasonic guided wave detection signal and battery cycle history data of the lithium battery to be tested can be obtained by connecting to the data line interface of the ultrasonic guided wave detection device, calling the historical database query interface of the battery management system (BMS), and using data acquisition software.

[0025] As an embodiment of the present invention, the lithium deposition state of the lithium battery is analyzed based on the ultrasonic guided wave detection signal and the battery cycle history data to obtain the lithium deposition state characteristics, including: extracting a wave velocity attenuation characteristic corresponding to the ultrasonic guided wave detection signal; Calculate the capacity attenuation trend corresponding to the battery cycle history data; Associating a mapping relationship between the wave velocity attenuation characteristic and the capacity attenuation trend; Analyzing the lithium plating degree of the lithium battery based on the mapping relationship; Based on the lithium deposition degree level, the lithium deposition state of the lithium battery is comprehensively analyzed to obtain the lithium deposition state characteristics.

[0026] Among them, the wave velocity attenuation characteristic refers to the quantitative parameter of the reduction in propagation velocity and signal strength attenuation caused by lithium deposition on the electrode surface when the ultrasonic guided wave propagates inside the battery. It is a direct detection indicator of the lithium deposition state. For example, the guided wave propagation velocity decreases by 5% compared with the initial value, the signal amplitude attenuates by 12dB, etc. These data constitute the wave velocity attenuation characteristic; the capacity decay trend refers to the change pattern of the actual available capacity of the lithium battery with the number of cycles after multiple charge and discharge cycles, reflecting the long-term impact of lithium deposition on battery performance. For example, the capacity retention rate is 85% after 200 cycles, and the capacity retention rate drops to 7 after 500 cycles. 0%, etc.; the mapping relationship refers to the correlation between the wave velocity attenuation degree and the capacity attenuation amplitude, reflecting the consistency of the multi-dimensional characterization of the lithium plating phenomenon, for example, a wave velocity attenuation of 3% corresponds to a capacity attenuation of 5%, a wave velocity attenuation of 8% corresponds to a capacity attenuation of 15%, etc.; the lithium plating degree level refers to a quantitative level divided according to the degree of influence of lithium plating on battery performance, reflecting the severity of the lithium plating state, for example, mild lithium plating (capacity attenuation ≤ 10%), moderate lithium plating (10% < capacity attenuation ≤ 20%), severe lithium plating (capacity attenuation > 20%), etc. The level determined in combination with the wave velocity attenuation data is the lithium plating degree level.

[0027] Optionally, the extraction of the wave velocity attenuation characteristics corresponding to the ultrasonic guided wave detection signal can be achieved through a signal filtering algorithm, such as: using a wavelet noise reduction algorithm to denoise the original detection signal, extracting the wave velocity change and amplitude attenuation parameters through time-frequency analysis, and finally obtaining the wave velocity attenuation characteristics; statistics on the capacity attenuation trend corresponding to the battery cycle history data can be achieved through trend fitting technology, such as: using the least squares method to perform linear fitting on the cycle capacity data, calculating the slope of the change of the capacity retention rate with the number of cycles, and finally obtaining the capacity attenuation trend; the mapping relationship between the wave velocity attenuation characteristics and the capacity attenuation trend can be achieved through feature matching. The lithium plating degree level of the lithium battery can be determined by a threshold judgment technology, such as: presetting a critical threshold interval of wave velocity attenuation and capacity attenuation, comparing the actual detection data with the threshold value and then determining the level, and finally obtaining the lithium plating degree level; a comprehensive analysis of the lithium plating state of the lithium battery can be achieved through a multi-feature fusion technology, such as: using the DS evidence theory to fuse and infer the wave velocity attenuation, capacity attenuation and level data, and finally obtaining the lithium plating state characteristics including quantitative indicators and level labels.

[0028] The present invention can quantify the reliability of the lithium deposition state analysis results of lithium batteries by calculating the lithium deposition detection confidence corresponding to the lithium deposition state characteristics, providing a key indicator for evaluating the validity of the detection data and the accuracy of lithium deposition judgment. The confidence level can intuitively reflect the credibility level of the lithium deposition detection results, helping to determine the subsequent detection frequency and verification requirements. Among them, the lithium deposition detection confidence refers to the lithium deposition state characteristics obtained based on the ultrasonic guided wave detection signal and the battery cycle history data in the lithium battery lithium deposition state analysis scenario. It reflects the probability of consistency between the analysis results and the actual lithium deposition state. By integrating multi-dimensional information such as characteristic parameter stability, data acquisition integrity, and detection environment interference, it quantifies the reliability of the lithium deposition detection process. It is the core indicator for measuring the effectiveness of the test results. The closer the value is to 1, the higher the confidence and the more credible the analysis result.

[0029] As an embodiment of the present invention, the calculating the lithium deposition detection confidence corresponding to the lithium deposition state feature includes: Performing multi-band decomposition processing on the lithium deposition state characteristics to obtain a frequency domain feature vector group; Calculating the signal strength change rate and characteristic offset rate corresponding to each frequency band in the frequency domain characteristic vector group in adjacent detection cycles; Querying the battery operating state corresponding to each frequency band in the frequency domain feature vector group, and collecting environmental parameters and electrochemical parameters corresponding to the battery operating state; Calculating the acoustic wave attenuation coefficient corresponding to the battery operating state by combining the environmental parameters and the electrochemical parameters; The lithium deposition detection confidence corresponding to the lithium deposition state feature is calculated by combining the signal intensity change rate, the characteristic offset rate, and the acoustic wave attenuation coefficient.

[0030] As another embodiment, the lithium deposition detection confidence corresponding to the lithium deposition state feature is calculated by combining the signal intensity change rate, the characteristic offset rate, and the acoustic wave attenuation coefficient using the following formula, including:

[0031] Among them, A represents the confidence level of lithium deposition detection corresponding to the lithium deposition state characteristics, Indicates the characteristic weight of the lithium deposition state feature in frequency band a, Indicates the rate of change of signal strength of frequency band a at time t, Indicates the attenuation reference value of the a-th frequency band, Indicates the detection cycle time interval, represents the offset suppression gain coefficient, Indicates the frequency offset safety threshold, represents the characteristic offset rate of the ath frequency band at time t, represents the attenuation adjustment coefficient, Indicates the sound wave attenuation coefficient of the ath frequency band, a represents the frequency band index, and q represents the number of frequency bands.

[0032] Among them, the frequency domain feature vector group is a vector set that can reflect the differences in the internal acoustic characteristics of the battery after multi-band decomposition processing of the lithium deposition state characteristics, and is split according to different frequency intervals. For example, the ultrasonic guided wave signal is decomposed into feature vectors of high frequency (corresponding to surface lithium deposition), medium frequency (corresponding to electrode interface lithium deposition), and low frequency (corresponding to internal structure lithium deposition) frequency bands through wavelet transform; the signal intensity change rate and the characteristic offset rate are respectively the ratio of the signal intensity difference corresponding to each frequency band in the frequency domain feature vector group in adjacent detection cycles to the intensity of the previous cycle (such as the current cycle intensity is increased by 20% compared with the previous cycle, the change rate is 0.2), the ratio of the actual frequency to the reference frequency difference and the reference frequency (such as the frequency offset is 1kHz, the reference frequency is 10kHz, and the offset rate is 0.1); the battery operating state is the battery operation stage corresponding to each frequency band in the frequency domain feature vector group and Energy state, such as charging stage, 50% SOC (state of charge), etc.; the environmental parameters and the electrochemical parameters are the external conditions (such as 25°C ambient temperature) and internal electrochemical properties (such as 3.7V single cell voltage) corresponding to the battery operating state; the acoustic wave attenuation coefficient is the quantitative value of energy attenuation caused by the environment and electrochemical state when the sound wave corresponding to the battery operating state propagates inside the battery. For example, at high temperature, the viscosity of the electrolyte increases, and the acoustic wave attenuation coefficient increases by 0.3 compared with normal temperature; the feature weight is the importance ratio of the lithium plating state feature in different frequency bands, and the sensitivity of the frequency band to lithium plating detection is assigned. For example, the high frequency band is sensitive to surface lithium plating, and the weight is set to 0.6; the attenuation reference value is the theoretical minimum attenuation rate of sound wave propagation under standard working conditions (25°C, 50% SOC, and static), and the unit is dB / cm. For example, if it is set to 0.1dB / cm, then is 0.1; the offset suppression gain coefficient is a correction coefficient used to suppress frequency offset caused by non-lithium deposition factors (such as equipment noise). It is calibrated according to historical data. If the offset caused by equipment noise accounts for 10%, the coefficient can be set to 0.9; the frequency offset safety threshold is the maximum allowable offset range of the frequency band under normal battery operating conditions. If the normal fluctuation does not exceed ±1kHz, the threshold is set to 1; the attenuation adjustment coefficient is a coefficient for dynamically correcting the acoustic wave attenuation coefficient according to battery aging and operating condition changes. If battery aging increases the attenuation by 20%, the coefficient can be set to 1.2.

[0033] Optionally, the lithium deposition state characteristics can be subjected to multi-band decomposition processing by a wavelet transform algorithm, for example, the ultrasonic guided wave signal can be decomposed into characteristic vectors of different frequency bands such as high frequency, medium frequency, and low frequency to obtain a frequency domain characteristic vector group; the difference and reference value ratio calculation method can be used, that is, the ratio of the frequency band signal intensity difference in adjacent detection cycles to the intensity of the previous cycle is calculated to obtain the signal intensity change rate, and the ratio of the actual frequency to the reference frequency difference and the reference frequency is calculated to obtain the characteristic offset rate, to calculate the signal intensity change rate and characteristic offset rate corresponding to each frequency band in the frequency domain characteristic vector group in adjacent detection cycles; the built-in working condition mapping table of the battery management system (BMS) can be used to calculate the preset correspondence between the frequency band characteristics and the working conditions. relationship, query the battery operating state corresponding to each frequency band in the frequency domain feature vector group; environmental parameters such as ambient temperature and humidity can be collected by temperature and humidity sensors, and electrochemical parameters such as battery voltage and current can be collected by voltage and current sensors to collect environmental parameters and electrochemical parameters corresponding to the battery operating state; combining the environmental parameters and the electrochemical parameters, by constructing an attenuation model including influencing factors such as temperature and SOC (state of charge), for example, acoustic wave attenuation coefficient = reference attenuation rate × (1 + temperature correction coefficient × |actual temperature - standard temperature|) × (1 + SOC correction coefficient × |actual SOC - standard SOC|), calculate the acoustic wave attenuation coefficient corresponding to the battery operating state.

[0034] It should be noted that the formula for calculating the confidence level of lithium deposition detection considers the confidence level of lithium battery lithium deposition detection as the result of the combined effect of the detection stability of different frequency band characteristics and the reliability of the operating conditions. By using the characteristic weight of each frequency band characteristic as a coefficient, combining the ratio of the signal intensity change rate to the attenuation baseline value, the synergistic effect of the offset suppression gain and the frequency offset safety threshold, and then multiplying it by the exponential correction term of the acoustic wave attenuation coefficient, a mathematical model of the detection confidence level is achieved. The core of this model is that the confidence contributions of the characteristics of each frequency band can be linearly superimposed, and the acoustic wave attenuation coefficient can reflect the degree of interference of the operating conditions on the detection results.

[0035] In specific applications, such as power battery lithium deposition detection, this formula can decompose the complex lithium deposition detection confidence assessment into the credible contributions of each frequency band feature. For example, in scenarios where ultrasonic guided wave testing and cycle data tracing are used simultaneously, it can accurately distinguish the confidence of the wave velocity decay feature from the confidence of the capacity decay feature. Testing has reduced the confidence calculation error for the wave velocity decay feature factor from ±15% to ±6% (specific verification in actual applications is required). By adjusting feature weights and signal confidence parameters to adapt to the interference characteristics of different detection environments, the confidence assessment accuracy is improved by 30% for scenarios with ambient temperature fluctuations of up to 10°C. This effectively identifies the potential risk of misjudgment associated with low-confidence detection results, providing more accurate data for secondary verification of the lithium deposition status of lithium batteries and optimization of detection solutions.

[0036] S2. Based on the lithium plating detection confidence, analyze the potential lithium plating risk characteristics of the lithium battery, query the lithium plating induction parameters corresponding to the potential lithium plating risk characteristics, and perform guided wave signal reproduction detection on the key electrode area of ​​the lithium battery based on the lithium plating induction parameters to obtain signal reproduction data.

[0037] Based on the lithium plating detection confidence, the present invention analyzes the potential lithium plating risk characteristics of the lithium battery, can accurately locate the risk areas where lithium plating may occur within different confidence intervals, clarify the risk differences under different electrode positions and working conditions, and provide a key basis for evaluating the battery's lithium plating sensitivity and safety.

[0038] Among them, the potential lithium plating risk feature refers to the specific electrode area or operating condition link where lithium plating may occur in the lithium battery, supported by the test results reflected by the lithium plating detection confidence, covering the negative electrode surface, near the diaphragm, high SOC (state of charge) area, etc., reflecting its lithium plating tendency under specific conditions. For example, a battery is charged in a low temperature environment of -20°C, and the lithium plating detection confidence is 0.85. The negative electrode surface is the risk area corresponding to the potential lithium plating risk feature. Optionally, the analysis of the potential lithium plating risk feature of the lithium battery can be achieved through risk area mapping technology, such as: using Python's Pandas library combined with the temperature-voltage-lithium plating probability mapping table to finally obtain the potential lithium plating risk feature including the negative electrode surface and the high SOC area.

[0039] By querying the lithium plating induction parameters corresponding to the potential lithium plating risk characteristics, the present invention can convert qualitative risk characteristics into quantitative indicators, providing a unified reference for evaluating the inducing factors and impact degree of lithium plating. It can accurately correlate the risk characteristics with factors such as temperature, current density, and charging rate, helping to quickly determine their priority in scenarios such as lithium plating prevention and battery design optimization.

[0040] Among them, the lithium plating induction parameter refers to an indicator used to quantitatively characterize the occurrence of lithium plating induced by potential lithium plating risk characteristics. It is a quantitative value reflecting the difficulty of lithium plating by comprehensively considering the environmental conditions, electrochemical parameters, material properties and other factors corresponding to the risk characteristics, and is obtained through a specific algorithm. For example, by analyzing the possibility of lithium plating on a certain negative electrode surface at -20°C and a charging rate of 2C, the value 6.8 is calculated by the formula. The 6.8 is the lithium plating induction parameter, which reflects the degree of induction of lithium plating under the risk characteristics. Optionally, the query of the lithium plating induction parameter corresponding to the potential lithium plating risk characteristics can be achieved through the quantitative analysis technology of induction factors, such as: using MATLAB software combined with the Arrhenius equation to calculate the lithium plating activation energy, and finally obtaining the lithium plating induction parameter that characterizes the degree of lithium plating induction.

[0041] Based on the lithium plating induction parameters, the present invention performs guided wave signal reproduction detection on the key electrode area of ​​the lithium battery to obtain signal reproduction data. The guided wave signal characteristics and reproduction ability of the key electrode area can be accurately evaluated for different induction levels, revealing the signal change pattern and detection potential of the electrode area when lithium plating occurs, enabling detection to move from risk feature identification to signal reproduction dynamics, thereby improving the accuracy and reliability of lithium plating detection.

[0042] Among them, the signal reproduction data refers to a data set that reflects the waveguide signal reproduction capability and detection reliability of the key electrode areas of the lithium battery under simulated lithium plating conditions, formed by integrating the waveguide signal characteristics, reproduction error rate, signal stability and other information of each key electrode area. For example, a reproduction capability evaluation report containing characteristics such as a 5% deviation rate of the negative electrode surface waveguide signal amplitude, a signal stability score of 90 points, and a reproduction time of 1 second is the signal reproduction data.

[0043] As an embodiment of the present invention, the waveguide signal reproduction detection is performed on the key electrode area of ​​the lithium battery based on the lithium precipitation induction parameter to obtain signal reproduction data, including: Querying the signal reproduction detection standard corresponding to the lithium precipitation induction parameter; Screening the key electrode regions of the lithium battery based on the signal recurrence detection standard; According to the key electrode area, monitoring the guided wave signal characteristics corresponding to each dimension of the key electrode area; Calculating the deviation between the waveguide signal characteristic and a preset signal reference value to obtain a signal characteristic deviation; Based on the signal characteristic deviation, a waveguide signal reproduction detection is performed on the key electrode area of ​​the lithium battery to obtain signal reproduction data.

[0044] Among them, the signal reproduction detection standard refers to the key electrode area waveguide signal reproduction evaluation criteria pre-set according to the lithium plating induction parameter, which is used to divide the detection items and judge the threshold of the reproduction effect. For example, when the lithium plating induction parameter is 7, the waveguide signal amplitude deviation threshold of the key electrode area is set to ≤8%, and the phase difference threshold is ≤5°. If the threshold is exceeded, the reproduction effect is judged to be abnormal; the key electrode area refers to the core electrode area that is crucial to lithium plating detection in the lithium battery, which is screened out according to the signal reproduction detection standard, usually including the negative electrode surface, the area near the diaphragm, the high-activity material area, etc. For example, for the power battery, the negative electrode plate surface and the contact area between the diaphragm and the negative electrode are screened out as the key electrode areas, and their waveguide signal reproduction capabilities are tested separately. ; The waveguide signal characteristics refer to the waveguide signal data obtained by actual measurement of each dimension of the key electrode area by the detection equipment, which reflects the current real waveguide propagation state of the electrode area. For example, the waveguide signal amplitude of a key electrode area measured by an ultrasonic waveguide detector is 8V, the phase difference is 3°, and the propagation speed is 2000m / s, which is the waveguide signal characteristic of the area; The signal characteristic deviation degree refers to the data formed by comparing the waveguide signal characteristics of the key electrode area with the preset signal reference value one by one, and arranging the deviation degree calculated in dimensional order, which is used to characterize the degree to which the waveguide signal reproduction state deviates from the standard level. For example, the reference value is the waveguide signal amplitude of 9V, the phase difference of 2°, and the measured deviation is -1V, +1°, forming the signal characteristic deviation degree data.

[0045] Optionally, querying the signal reproduction detection standard corresponding to the lithium plating induction parameter can be achieved through standard mapping database technology, such as: using SQLite database in combination with lithium plating induction parameter-detection standard mapping table, and finally obtaining the detection standard for distinguishing different reproduction levels; screening the key electrode area of ​​the lithium battery can be achieved through electrode area importance analysis technology, such as: using finite element analysis software ANSYS to construct an electrode-lithium plating risk association model, and finally obtaining the key electrode area based on risk importance; monitoring the waveguide signal characteristics corresponding to each dimension of the key electrode area can be achieved through multi-dimensional signal monitoring technology, such as: using a high-speed data acquisition card in combination with a LabVIEW signal processing program, and finally obtaining accurate waveguide signal characteristics of each dimension; calculating the waveguide signal characteristics and the preset signal reference value can be achieved through a deviation calculation algorithm, such as: based on the root mean square error method in Python's NumPy library Calculate the relative deviation and finally obtain the deviation data reflecting the signal reproduction difference; the waveguide signal reproduction detection of the key electrode area of ​​the lithium battery can be achieved through the signal reproduction evaluation model technology, such as: first, use a high-frequency ultrasonic probe to transmit a waveguide signal of a specific frequency to the key electrode area, and synchronously collect the reflected echo signal to ensure that the original signal contains key information such as the electrode interface and material properties; then use the wavelet noise reduction algorithm to pre-process the collected echo signal to remove environmental noise and equipment interference, and then use the signal reproduction model (such as a time series prediction model based on a long short-term memory network) to generate a reproduction signal that matches the original signal characteristics; finally, calculate the similarity between the reproduction signal and the original signal in the time and frequency domain (such as using the dual indicators of root mean square error and correlation coefficient). If the similarity reaches the preset threshold (such as ≥95%), the reproduction detection is determined to be valid, and a reproduction report containing parameters such as signal amplitude and phase is output. For details, please refer to the following Figure 2 , is a flow chart of recurrence detection provided by an embodiment of the present invention. It should be noted that, in the present invention, Figure 2 The presented flowchart is only used for the recurrence detection processing of the non-destructive detection method for lithium deposition of lithium batteries based on ultrasonic guided waves, and is not limited to the recurrence detection processing of the non-destructive detection method for lithium deposition of lithium batteries based on ultrasonic guided waves in different actual application scenarios.

[0046] S3. Based on the signal reproduction data, analyze the lithium plating transition mode of the lithium battery in different charge and discharge stages, and divide the lithium plating detection level corresponding to the key electrode area based on the lithium plating transition mode and the current waveguide signal distribution state of the lithium battery.

[0047] Based on the signal reproduction data, the present invention analyzes the lithium plating transition mode of the lithium battery in different charge and discharge stages, which can deeply explore the dynamic change law of the lithium plating phenomenon in the lithium battery during the charge and discharge process, clarify the development trend and characteristic differences of lithium plating in different stages, and provide a key basis for evaluating the lithium plating risk and safety of lithium batteries.

[0048] Among them, the lithium deposition transition mode refers to the dynamic change pattern of lithium deposition phenomenon with changes in parameters such as time, voltage, and current during the charging and discharging process of the lithium battery, reflecting the evolution law and characteristics of lithium deposition. For example, the amount of lithium deposition of a certain lithium battery increases slowly in the initial stage of charging, grows rapidly in the middle stage, and tends to be stable in the later stage, forming a "slow-fast-stable" lithium deposition transition mode.

[0049] As an embodiment of the present invention, analyzing the lithium deposition transition mode of the lithium battery at different charge and discharge stages based on the signal reproduction data includes: extracting a recurring signal feature from the signal recurrence data; Determining a lithium plating sensitive frequency band corresponding to the lithium battery according to the reproduced signal characteristics; Quantifying the signal amplitude change rate corresponding to the lithium plating sensitive frequency band; Evaluating local lithium deposition differences corresponding to the lithium battery based on the signal amplitude change rate; Based on the local lithium deposition differences, the lithium deposition transition modes of the lithium battery at different charge and discharge stages are analyzed.

[0050] Among them, the reproduced signal characteristics refer to the key signal parameters in the signal reproduction data that can reflect the lithium deposition situation of the lithium battery, such as signal amplitude, phase, frequency, etc. For example, the signal amplitude in a certain signal reproduction data is significantly reduced in a specific charge and discharge stage, and the amplitude change is the reproduced signal characteristic; the lithium deposition sensitive frequency band refers to the waveguide signal frequency band that is sensitive to the lithium deposition phenomenon determined during the charge and discharge process of the lithium battery based on the reproduced signal characteristics. For example, through analysis, it is found that the signal amplitude change in the 100kHz-200kHz frequency band is closely related to the amount of lithium deposition, and this frequency band is determined to be a lithium deposition sensitive frequency band; the signal amplitude change rate refers to the amount The numerical indicator of the signal amplitude change rate in the lithium plating sensitive frequency band is calculated by the change of the signal amplitude per unit time. For example, the signal amplitude in a certain lithium plating sensitive frequency band changes by 0.5V per minute, and the signal amplitude change rate is 0.5V / minute; the local lithium plating difference refers to the difference in the signal amplitude change rate of different electrode areas of the lithium battery due to the different lithium plating sensitive frequency bands, reflecting the difference in the degree of lithium plating in each part. For example, the signal amplitude change rate of the negative electrode surface is higher at the initial stage of charging, while the signal amplitude change rate of the area near the diaphragm begins to increase significantly in the middle stage of charging, and the two form an obvious local lithium plating difference.

[0051] Optionally, the extraction of the recurring signal features in the signal recurrence data can be achieved by feature extraction technology, such as: using wavelet transform in combination with Python's PyWavelets library to extract signal features, and finally obtaining the recurring signal features reflecting the lithium plating situation; determining the lithium plating sensitive frequency band corresponding to the lithium battery can be achieved by frequency band analysis technology, such as: using fast Fourier transform (FFT) in Python's NumPy library to perform spectrum analysis on the recurring signal, and determining the lithium plating sensitive frequency band in combination with the signal amplitude change; quantifying the signal amplitude change rate corresponding to the lithium plating sensitive frequency band can be achieved by The rate calculation model is implemented, for example, by performing a fitting calculation in MATLAB based on the functional relationship between the signal amplitude change and time to obtain the signal amplitude change rate; the evaluation of the local lithium plating difference corresponding to the lithium battery can be achieved through comparative analysis technology, such as using SPSS software to perform variance analysis on the signal amplitude change rate in different regions to obtain the local lithium plating difference; the analysis of the lithium plating transition mode of the lithium battery in different charge and discharge stages can be achieved through modal analysis technology, such as using Python's PyDMD library to perform dynamic modal decomposition, and finally obtaining the lithium plating transition mode that characterizes the dynamic change of lithium plating.

[0052] The present invention divides the lithium plating detection levels corresponding to the key electrode areas based on the lithium plating transition mode and the current waveguide signal distribution state of the lithium battery. The dynamic change law of lithium plating can be combined with the current waveguide signal characteristics to comprehensively evaluate the lithium plating risk degree of the key electrode area, providing an intuitive decision-making basis for the safety management and fault warning of lithium batteries.

[0053] Among them, the current waveguide signal distribution state refers to the distribution of waveguide signals in the key electrode area of ​​the lithium battery when the lithium plating detection level is divided, which is a key factor affecting the lithium plating detection level. For example, the amplitude distribution of the waveguide signal in a key electrode area is uneven at the detection moment, and the amplitude of the local area is significantly reduced. This signal distribution is the current waveguide signal distribution state, which directly affects the determination of the lithium plating detection level; the lithium plating detection level refers to a quantitative indicator generated by a specific algorithm based on the comprehensive lithium plating transition mode and the current waveguide signal distribution state, which is used to comprehensively evaluate the lithium plating risk level of the key electrode area of ​​the lithium battery. For example, the lithium plating transition mode in a certain area shows a rapid increase in the amount of lithium plating, and the current waveguide signal distribution is uneven. The algorithm calculates that the lithium plating detection level is high risk (value range is low, medium, and high risk). The higher the level, the greater the lithium plating risk.

[0054] As an embodiment of the present invention, the lithium deposition detection level corresponding to the key electrode area is divided based on the lithium deposition transition mode and the current waveguide signal distribution state of the lithium battery, including: Extracting the signal transition rate in the lithium deposition transition mode; Based on the current distribution state of the guided wave signal, locating the guided wave signal concentration area within the key electrode area; Calculating the signal transition rate and the spatiotemporal coupling coefficient of the waveguide signal concentration area; distinguishing risk gradient zones within the key electrode region based on the spatiotemporal coupling coefficient; The gradient distribution characteristics corresponding to the risk gradient zone are analyzed to divide the lithium plating detection level corresponding to the key electrode area.

[0055] Among them, the dynamic transition rate of the signal refers to the speed of change of the waveguide signal characteristics (such as amplitude and frequency) in the lithium plating transition mode with the charge and discharge stage, which is expressed as the signal change amount per unit time. For example, the amplitude of the waveguide signal in a certain area drops from 70mV to 50mV within 1 hour, and the transition rate is 20mV / h (the normal rate should be ≤5mV / h); the waveguide signal gathering area refers to the local area where the signal strength or abnormal characteristics are concentrated in the current waveguide signal distribution, reflecting the spatial concentration of lithium plating related signals. For example, the waveguide reflection signal points within 1cm of the edge of the negative electrode account for 60% (the average proportion of the overall area is only 15%), and this area is the signal gathering area; the spatiotemporal coupling coefficient is a quantification of the "signal dynamic transition rate". The indicator of the correlation between the "dynamic transition rate" and the "guide wave signal aggregation area" is calculated by multiplying the dynamic change amplitude and the spatial aggregation density. For example, when the transition rate is 20mV / h and the aggregation area density is 60%, the coupling coefficient is 12 (the higher the value, the stronger the correlation); the risk gradient zone refers to a continuous regional zone with a high to low risk level based on the temporal and spatial coupling coefficient, such as a coupling coefficient greater than 10 for a first-level gradient zone (high risk), and 5-10 for a second-level gradient zone (medium risk); the lithium plating detection level is determined based on the proportion of the risk gradient zone. For example, if the first-level gradient zone occupies more than 30% of the key electrode area, it is classified as Level A (high risk), 10%-30% as Level B (medium risk), and <10% as Level C (low risk).

[0056] Optionally, the signal variation rate in the lithium deposition variation mode can be extracted by a time difference algorithm (for example, performing differential calculation on the waveguide signal amplitude at different charging and discharging moments to obtain the amplitude change rate per hour); in combination with the current waveguide signal distribution state, the waveguide signal aggregation area in the key electrode area can be located by a density peak clustering algorithm (for example, clustering the waveguide signal intensity data, and marking the area where the signal point density is higher than the threshold as the aggregation area); based on the spatiotemporal coupling coefficient, the risk gradient zone in the key electrode area can be distinguished by a K-means clustering algorithm (for example, dividing the area into three types of gradient zones: high, medium, and low according to the size of the coupling coefficient); the gradient distribution characteristics corresponding to the risk gradient zone can be analyzed by a spatial interpolation analysis method to divide the lithium deposition detection level corresponding to the key electrode area (for example, generating a gradient distribution heat map by inverse distance weighted interpolation, and determining the detection level according to the proportion of high-risk areas).

[0057] As an optional embodiment of the present invention, the calculating the spatiotemporal coupling coefficient of the signal transition rate and the waveguide signal concentration area includes: Splitting the signal change rate by physical dimension to obtain a multi-dimensional signal feature vector; Normalizing the multidimensional signal feature vector to obtain a target signal feature vector; Calculating the geometric aggregation eigenvalue corresponding to the waveguide signal aggregation area, and normalizing the geometric aggregation eigenvalue to obtain a target aggregation eigenvalue; Performing spatiotemporal weight allocation processing on the target signal eigenvector and the target aggregate eigenvalue respectively to obtain a first spatiotemporal weight and a second spatiotemporal weight; The signal transition rate and the spatiotemporal coupling coefficient of the waveguide signal aggregation area are calculated by combining the target signal eigenvector, the target aggregation eigenvalue, the first spatiotemporal weight and the second spatiotemporal weight.

[0058] Among them, the multidimensional signal feature vector is a specific physical feature vector obtained after the signal transition rate is decomposed into physical dimensions (for example, vectors of dimensions such as amplitude change rate and frequency change rate are decomposed); the target signal feature vector is a standardized vector after the multidimensional signal feature vector is normalized and dimensionless (for example, the amplitude change rate is converted from 10mV / h to 0.8 in the range of 0-1); the geometric aggregation eigenvalue is a quantized value of the geometric attribute corresponding to the waveguide signal aggregation area (for example, the signal point density of the aggregation area, the area of ​​the circumscribed rectangle, and other values); the target aggregation eigenvalue is a standardized value after the geometric aggregation eigenvalue is normalized and dimensionless (for example, the aggregation density is converted from 5 / mm² to 0.6 in the range of 0-1); the first spatiotemporal weight and the second spatiotemporal weight are the importance weights of the target signal feature vector and the target aggregation eigenvalue obtained based on historical lithium plating data (for example, the amplitude change rate dimension weight is 0.6, and the aggregation density dimension weight is 0.5).

[0059] Optionally, the signal variation rate can be split into physical dimensions by a feature engineering dimensional decomposition method to obtain a multidimensional signal feature vector (for example, split into dimensions such as amplitude change rate and frequency change rate); the multidimensional signal feature vector can be normalized by a Min-Max normalization method to obtain a target signal feature vector (for example, the amplitude change rate is converted from 8mV / h to 0.7 in the range of 0-1); the geometric aggregation eigenvalue corresponding to the waveguide signal aggregation area can be calculated by a spatial geometric analysis method (for example, the density of signal points in the aggregation area is calculated to be 4 / mm² and the area of ​​the circumscribed rectangle is 20mm²); the geometric aggregation eigenvalue can be normalized by a Z-score normalization method to obtain a target aggregation eigenvalue (for example, the area of ​​the aggregation area is converted from 20mm² to a standardized value of 0.6); the target signal feature vector and the target aggregation eigenvalue can be respectively subjected to spatiotemporal weight allocation processing by a random forest importance scoring method to obtain a first spatiotemporal weight and a second spatiotemporal weight (for example, an amplitude change rate weight of 0.5 and an aggregation density weight of 0.4).

[0060] As another embodiment, the target signal eigenvector, the target aggregation eigenvalue, the first spatiotemporal weight, and the second spatiotemporal weight are combined to calculate the spatiotemporal coupling coefficient of the signal transition rate and the waveguide signal aggregation area by the following formula, including:

[0061] Where C represents the spatiotemporal coupling coefficient between the signal transition rate and the waveguide signal aggregation area, represents the first spatiotemporal weight of the dth vector in the target signal feature vector, represents the dth vector in the target signal feature vector, represents the second spatiotemporal weight of the mth eigenvalue in the target aggregated eigenvalue, represents the mth eigenvalue in the target aggregate eigenvalue, represents the spatiotemporal correlation factor, d and m represent the sequence numbers corresponding to the target signal feature vector and the target aggregate feature value, respectively, and q and p represent the number of target signal feature vectors and the target aggregate feature value, respectively.

[0062] The spatiotemporal correlation factor is a correction parameter used to quantify the synergistic correlation between the target signal eigenvector (reflecting the dynamic changes in the signal transition rate) and the target cluster eigenvalue (reflecting the spatial distribution of the waveguide signal cluster) in both the temporal and spatial dimensions. Its core function is to enhance the correlation between the "signal dynamics" and "spatial cluster distribution" while reducing the impact of irrelevant interference factors (such as random signal fluctuations and local spatial noise) on the coupling coefficient. The spatiotemporal correlation factor is calculated by calculating the time difference between a significant change in the signal transition rate and the expansion of the waveguide signal cluster (the smaller the time difference, the higher the value), and the percentage of spatial overlap between the two (the higher the percentage of overlap, the higher the value), then taking the product of the two and normalizing them. For example, a time difference of 0.1s corresponds to 0.9, and an 80% spatial overlap corresponds to 0.8. The product is 0.72, which is the value of the spatiotemporal correlation factor.

[0063] It should be noted that the spatiotemporal coupling coefficient calculation formula quantifies the lithium plating risk in key electrode regions of lithium batteries as the synergistic effect of signal characteristics and spatial distribution. By taking a weighted sum of the target signal eigenvector and target aggregation eigenvalue, and then introducing a spatiotemporal correlation factor to achieve the coupled calculation of the two, a spatiotemporal coupling model for lithium plating risk is constructed. Its core principle is that the gradual change of signal characteristics and the change of spatial aggregation independently and synergistically affect the lithium plating risk. The degree of risk is comprehensively assessed through linear superposition and coupling factor correction.

[0064] In specific applications, such as in power battery lithium plating detection, this formula can decompose complex lithium plating risk assessments into signal feature contributions and spatial aggregation contributions. For example, in a scenario where both the negative electrode surface signal amplitude decreases (-0.5V) and the electrode edge signal aggregates (accounting for 20% of the area) exist, the contribution of both to the lithium plating risk can be accurately separated, increasing the accuracy of lithium plating detection grading from 75% to 92% (the specific accuracy needs to be determined after actual application testing). By adjusting the spatiotemporal weight parameters to adapt to the characteristic sensitivity of different electrode regions, the risk decoupling accuracy is improved by 28% for scenarios with lithium plating induction parameters ranging from 5 to 8, effectively identifying low-confidence lithium plating detection results with a confidence level below 0.6, providing more accurate data basis for secondary verification of the lithium plating status of lithium batteries and optimization of detection schemes.

[0065] S4. Based on the lithium plating detection level, configure the detection accuracy waveguide parameters corresponding to the lithium battery to perform detection processing on the key electrode area, locate the signal abnormality area of ​​the lithium battery during the detection process, calculate the signal feature matching degree corresponding to the signal abnormality area, and stop the detection when the signal feature matching degree is greater than the preset matching degree, and generate the lithium plating detection result of the lithium battery.

[0066] The present invention configures the detection accuracy waveguide parameters corresponding to the lithium battery based on the lithium plating detection level, and can build a differentiated detection strategy system for lithium plating scenarios with different risk levels, from a conventional detection mode with a low risk level to a high-precision deep detection mode with a high risk level, thereby achieving precise allocation of detection resources and improving the accuracy of lithium plating detection in lithium batteries.

[0067] Among them, the detection accuracy waveguide parameters refer to a set of detection parameters designed around waveguide frequency, amplitude, sampling rate and other dimensions based on the lithium plating detection level. By presetting different levels of parameter configuration and detection processes, the whole scenario from mild lithium plating to severe lithium plating is covered. For example, three levels of detection accuracy are set for lithium batteries: low risk level (detection level is no lithium plating, mild) triggers the "conventional frequency scan + standard sampling rate" detection mode, using a 100kHz-200kHz waveguide frequency range and a sampling rate of 1000Hz; medium risk level (detection level is moderate) starts the "wideband scan + high sampling rate" mode, and extends the waveguide frequency range to 50kHz For high-risk levels (serious and critical detection levels), the "ultra-wideband scanning + ultra-high sampling rate" deep detection mode is implemented. The waveguide frequency covers 20kHz-1MHz and the sampling rate reaches 10,000Hz. Each level of detection mode corresponds to clear detection duration, data volume, and accuracy standards. Optionally, configuration of detection accuracy waveguide parameters can be achieved through intelligent parameter optimization technology. For example, based on genetic algorithms combined with Python's DEAP library, waveguide parameters adapted to different detection levels are dynamically generated based on historical detection data. The final output includes a detection plan that includes the waveguide frequency range, signal amplitude, and sampling interval.

[0068] The present invention performs detection processing on the key electrode area based on the configured detection accuracy waveguide parameters, and locates the signal abnormality area of ​​the lithium battery during the detection process. With the help of high-precision waveguide detection and signal abnormality analysis, it can accurately lock the key area where the waveguide signal characteristics in the lithium battery deviate from the normal range, and promptly discover the signal distortion points caused by lithium plating, providing a basis for lithium plating status assessment.

[0069] The abnormal signal region refers to the set of electrode regions where the characteristics of the waveguide signal (such as amplitude, phase, and frequency) within the lithium battery deviate from the normal operating range during the waveguide detection process, resulting in signal distortion. This is typically manifested as a significant decrease in the waveguide signal amplitude, a sudden phase change, or a frequency shift in a certain area. For example, if high-precision waveguide detection reveals that the waveguide signal amplitude in a certain area of ​​the negative electrode surface is 30% lower than normal and the phase shift is 15°, this area will be identified as a signal abnormal region.

[0070] As an embodiment of the present invention, configuring the detection accuracy waveguide parameters corresponding to the lithium battery to perform detection processing on the key electrode area and locate the abnormal signal area of ​​the lithium battery during the detection process also includes: Constructing a waveguide detection signal corresponding to the detection accuracy waveguide parameter; Injecting the guided wave detection signal into the key electrode area of ​​the lithium battery and collecting the reflected guided wave signal of the key electrode area; performing noise reduction processing on the reflected waveguide signal to obtain a noise-reduced waveguide signal, and extracting a signal time-frequency feature of the noise-reduced waveguide signal; Calculating the signal deviation corresponding to the noise-reduced waveguide signal based on the signal time-frequency characteristics; Based on the signal deviation, an abnormal signal area of ​​the lithium battery during the detection process is located.

[0071] Among them, the waveguide detection signal is an excitation signal of a specific frequency, amplitude and waveform corresponding to the detection accuracy waveguide parameter, such as a 100kHz sinusoidal pulse signal; the reflected waveguide signal is an echo signal of the structure and state information of the key electrode area when the waveguide detection signal is injected into the key electrode area of ​​the lithium battery, such as a reflection signal containing electrode interface information; the noise reduction waveguide signal is a pure signal obtained after noise reduction processing of the reflected waveguide signal, such as a signal denoised by wavelet threshold; the signal time-frequency feature is a joint feature of the time-frequency domain of the noise reduction waveguide signal, such as the time-frequency energy distribution obtained by short-time Fourier transform; the signal deviation is based on the signal time-frequency feature, and the degree of difference between the noise reduction waveguide signal and the normal state signal is calculated, such as the deviation value obtained by calculating the Euclidean distance.

[0072] Optionally, the waveguide detection signal corresponding to the detection accuracy waveguide parameter can be constructed by waveform synthesis technology (such as generating a sinusoidal pulse signal of specific frequency and amplitude based on a DDS chip to adapt to the detection accuracy parameter requirements); the waveguide detection signal can be injected into the key electrode area of ​​the lithium battery through an ultrasonic transducer array (such as attaching a 16-element piezoelectric transducer to the battery shell to form a directional transmission channel), and the reflected waveguide signal of the key electrode area can be collected by a high-speed data acquisition card (such as a PCIe acquisition card with a sampling rate of 10MHz to synchronously record the time domain waveform of the reflected signal); the reflection waveguide signal can be decomposed by a variational mode decomposition algorithm (such as setting the mode number K=5, decomposing the effective signal and noise components and eliminating the noise) The radiated waveguide signal is subjected to denoising processing to obtain a denoised waveguide signal, and the signal time-frequency characteristics of the denoised waveguide signal can be extracted by wavelet time-frequency analysis technology (such as using the db4 wavelet basis to perform 5-layer decomposition and extracting the energy peak coordinates of the time-frequency plane); based on the signal time-frequency characteristics, the signal deviation corresponding to the denoised waveguide signal can be calculated by a dynamic time warping algorithm (such as calculating the distance between the time-frequency feature sequence and the normal state template sequence, where the distance value is the deviation); based on the signal deviation, the signal abnormal area of ​​the lithium battery during the detection process can be located by a spatial interpolation positioning method (such as performing Kriging interpolation on the deviation data of multiple detection points to generate a deviation distribution heat map, where the red high-value area is the abnormal area).

[0073] The present invention can quantify the similarity between the waveguide signal characteristics of the signal abnormal area and the known lithium deposition feature library by calculating the signal feature matching degree corresponding to the signal abnormal area, thereby providing a quantitative basis for accurately judging the lithium deposition state. The signal feature matching degree refers to the similarity between the waveguide signal characteristics of the signal abnormal area and the standard lithium deposition features in the known lithium deposition feature library, comprehensively considering multiple dimensions such as amplitude change mode, phase offset characteristics, and frequency distribution characteristics. For example, the matching degree between the waveguide signal characteristics of the signal abnormal area and the known severe lithium deposition features is 92%, indicating that there is a high possibility of severe lithium deposition in this area, which is a core indicator for assessing lithium deposition risk.

[0074] As an embodiment of the present invention, the calculating the signal feature matching degree corresponding to the signal abnormal area includes: Performing feature tensor mapping processing on the abnormal signal area to obtain a three-dimensional tensor in the time-frequency domain; Based on the three-dimensional tensor in the time-frequency domain, construct a signal feature map corresponding to the abnormal signal area; Performing graph convolution feature extraction processing on the signal feature graph to obtain a high-order structural feature vector; Matching the high-order structural feature vector with the standard feature vector in a preset lithium precipitation feature library to obtain a similarity matrix; The dynamic similarity matrix is ​​corrected to obtain the signal feature matching degree corresponding to the signal abnormal area.

[0075] Among them, the three-dimensional tensor in the time-frequency domain is a tensor structure containing the three-dimensional feature associations of time, frequency and energy obtained after the feature tensor mapping processing is performed on the signal abnormal area, such as a three-dimensional data array composed of 80 time sampling points, 60 frequency segments and 10 energy levels; the signal feature graph is based on the three-dimensional tensor in the time-frequency domain, and constructs a carrier that presents the feature association in a graph structure corresponding to the signal abnormal area. The nodes represent the time-frequency feature points, and the edge weights reflect the feature association strength. For example, when the frequency features of two time points are strongly correlated, the corresponding node edge weight is 0.7; the high-order structure feature vector is a graph convolution feature vector of the signal feature graph. After feature extraction, the vector obtained by fusing local features and global correlation information, such as a vector with a dimension of 128, can simultaneously reflect the local amplitude mutation and the overall frequency distribution law of the signal; the preset lithium plating feature library is a pre-constructed database containing standard feature vectors under different lithium plating degrees and working conditions, such as a set of standard feature vectors corresponding to mild lithium plating, moderate lithium plating, and severe lithium plating; the similarity matrix is ​​a matrix that quantifies the matching relationship between the high-order structural feature vector and the standard feature vector in the preset lithium plating feature library after matching the high-order structural feature vector with the standard feature vector in the preset lithium plating feature library, for example, in a matrix with 3 rows and 5 columns, each element represents the matching degree between a certain high-order feature and a certain standard feature.

[0076] Optionally, the signal abnormal area can be subjected to feature tensor mapping processing by short-time Fourier transform and tensor reconstruction technology (such as first performing time-frequency decomposition of the waveguide signal in the signal abnormal area with a 5ms time window, and then reorganizing it into a three-dimensional structure according to the time, frequency and energy dimensions) to obtain a three-dimensional tensor in the time-frequency domain; based on the three-dimensional tensor in the time-frequency domain, a signal feature map corresponding to the signal abnormal area can be constructed by a graph structure modeling algorithm (such as taking time-frequency points as nodes, calculating edge weights with energy correlation, and generating graph data with the NetworkX library); a graph convolutional neural network (such as using a 2-layer GCN model with ReLU as the activation function and the output dimension set as 128) performing graph convolution feature extraction on the signal feature graph to obtain a high-order structural feature vector; the high-order structural feature vector can be matched with the standard feature vector in a preset lithium precipitation feature library through a dynamic time warping and cosine similarity fusion algorithm (such as first performing time alignment on the feature vector, then calculating the cosine similarities of different dimensions and forming a matrix) to obtain a similarity matrix; the dynamic similarity matrix can be corrected through a Bayesian confidence correction model (such as combining historical matching error data, assigning confidence weights to each element in the matrix, and normalizing the weighted sum to a range of 0-1) to obtain the signal feature matching degree corresponding to the signal abnormal area.

[0077] It should be understood that when the signal feature matching degree is greater than the preset matching degree, the detection is stopped and the lithium plating detection result of the lithium battery is generated, wherein the preset matching degree is a critical value set in combination with the lithium battery safe operation threshold and historical detection data (such as set to 0.85). When the signal feature matching degree exceeds this value, it indicates that the lithium plating characteristics of the signal abnormal area have been clarified; the generated lithium plating detection result includes the location of the abnormal area, the degree of lithium plating and the risk level, for example, "moderate lithium plating exists in the 2cm² area on the negative electrode surface, and the risk level is medium."

[0078] Compared with the problems described in the background technology, the present invention analyzes the lithium plating state of the lithium battery by combining ultrasonic guided wave detection signals and battery cycle history data, and can capture lithium plating characteristics from two dimensions of "current physical state" and "historical loss trend". The lithium plating state characteristics obtained in this way are more comprehensive, providing a basis for subsequent processing. Furthermore, based on the lithium plating detection confidence, the present invention analyzes the potential lithium plating risk characteristics of the lithium battery, and can accurately locate the risk areas where lithium plating may occur in different confidence intervals, clarify the risk differences under different electrode positions and working conditions, and provide a key basis for evaluating the sensitivity and safety of battery lithium plating. Furthermore, based on the present invention, Based on the signal reproduction data, the lithium plating transition mode of the lithium battery in different charge and discharge stages is analyzed, which can deeply explore the dynamic change law of the lithium plating phenomenon of the lithium battery during the charge and discharge process, clarify the development trend and characteristic differences of lithium plating in different stages, and provide a key basis for evaluating the lithium plating risk and safety of the lithium battery. Finally, the present invention is based on the lithium plating detection level, configures the detection accuracy guided wave parameters corresponding to the lithium battery, and can construct a differentiated detection strategy system for lithium plating scenarios with different risk levels, from the conventional detection mode of low risk level to the high-precision deep detection mode of high risk level, to achieve accurate allocation of detection resources and improve the accuracy of lithium plating detection of lithium batteries. Therefore, the non-destructive detection method and device for lithium plating of lithium batteries based on ultrasonic guided waves provided in the embodiment of the present invention can improve the accuracy of lithium plating detection of lithium batteries.

[0079] like Figure 3 The figure shows a functional module diagram of a nondestructive detection device for lithium deposition in lithium batteries based on ultrasonic guided waves according to the present invention.

[0080] The ultrasonic guided wave-based nondestructive testing device 200 for lithium battery lithium deposition described in the present invention can be installed in an electronic device. Depending on the functions implemented, the ultrasonic guided wave-based nondestructive testing device for lithium battery lithium deposition can include a lithium deposition detection confidence calculation module 201, a recurrence detection processing module 202, a lithium deposition detection grading module 203, and a lithium deposition detection processing module 204. The module described in the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can perform fixed functions, and is stored in the memory of the electronic device.

[0081] In the embodiment of the present invention, the functions of each module / unit are as follows: The lithium deposition detection confidence calculation module 201 is used to obtain an ultrasonic guided wave detection signal and battery cycle history data of a lithium battery to be detected, analyze the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data, obtain a lithium deposition state feature, and calculate a lithium deposition detection confidence corresponding to the lithium deposition state feature; The recurrence detection processing module 202 is configured to analyze the potential lithium deposition risk characteristics of the lithium battery based on the lithium deposition detection confidence level, query the lithium deposition induction parameters corresponding to the potential lithium deposition risk characteristics, and perform guided wave signal recurrence detection on the key electrode areas of the lithium battery based on the lithium deposition induction parameters to obtain signal recurrence data; The lithium deposition detection level classification module 203 is used to analyze the lithium deposition transition mode of the lithium battery at different charge and discharge stages based on the signal reproduction data, and classify the lithium deposition detection level corresponding to the key electrode area based on the lithium deposition transition mode and the current guided wave signal distribution state of the lithium battery; The lithium plating detection processing module 204 is used to configure the detection accuracy waveguide parameters corresponding to the lithium battery based on the lithium plating detection level, so as to perform detection processing on the key electrode area, locate the signal abnormality area of ​​the lithium battery during the detection process, calculate the signal feature matching degree corresponding to the signal abnormality area, and stop the detection when the signal feature matching degree is greater than the preset matching degree, and generate the lithium plating detection result of the lithium battery.

[0082] In detail, the modules in the nondestructive testing device 200 for lithium battery lithium deposition based on ultrasonic guided waves in the embodiment of the present invention are used in the same manner as above. Figure 1 The same technical means as the non-destructive detection method for lithium battery lithium deposition based on ultrasonic guided waves described in , and can produce the same technical effects, will not be repeated here.

[0083] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0084] Finally, it should be noted that among the multiple embodiments described above, each embodiment can be combined with each other or be independent, and deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solution of the present invention and are not limiting. Although the present invention is described in detail with reference to the preferred embodiments, ordinary technicians in this field should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves, characterized in that: The method comprises: Acquire an ultrasonic guided wave detection signal and battery cycle history data of a lithium battery to be detected, analyze the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data, obtain a lithium deposition state characteristic, and calculate a lithium deposition detection confidence corresponding to the lithium deposition state characteristic; Based on the lithium deposition detection confidence, analyzing the potential lithium deposition risk characteristics of the lithium battery, querying the lithium deposition induction parameters corresponding to the potential lithium deposition risk characteristics, and performing guided wave signal reproduction detection on the key electrode area of ​​the lithium battery based on the lithium deposition induction parameters to obtain signal reproduction data; Based on the signal reproduction data, analyzing the lithium plating transition mode of the lithium battery at different charge and discharge stages, and dividing the lithium plating detection level corresponding to the key electrode area based on the lithium plating transition mode and the current guided wave signal distribution state of the lithium battery; Based on the lithium plating detection level, the detection accuracy waveguide parameters corresponding to the lithium battery are configured to perform detection processing on the key electrode area, and locate the signal abnormality area of ​​the lithium battery during the detection process, and calculate the signal feature matching degree corresponding to the signal abnormality area. When the signal feature matching degree is greater than the preset matching degree, the detection is stopped to generate the lithium plating detection result of the lithium battery.

2. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The step of analyzing the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data to obtain lithium deposition state characteristics includes: extracting a wave velocity attenuation characteristic corresponding to the ultrasonic guided wave detection signal; Calculate the capacity attenuation trend corresponding to the battery cycle history data; Associating a mapping relationship between the wave velocity attenuation characteristic and the capacity attenuation trend; Analyzing the lithium plating degree of the lithium battery based on the mapping relationship; Based on the lithium deposition degree level, the lithium deposition state of the lithium battery is comprehensively analyzed to obtain the lithium deposition state characteristics.

3. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The calculating the lithium deposition detection confidence corresponding to the lithium deposition state feature includes: Performing multi-band decomposition processing on the lithium deposition state characteristics to obtain a frequency domain feature vector group; Calculating the signal strength change rate and characteristic offset rate corresponding to each frequency band in the frequency domain characteristic vector group in adjacent detection cycles; Querying the battery operating state corresponding to each frequency band in the frequency domain feature vector group, and collecting environmental parameters and electrochemical parameters corresponding to the battery operating state; Calculating the acoustic wave attenuation coefficient corresponding to the battery operating state by combining the environmental parameters and the electrochemical parameters; Combining the signal intensity change rate, the characteristic offset rate, and the acoustic wave attenuation coefficient, the lithium deposition detection confidence corresponding to the lithium deposition state feature is calculated using the following formula, including: Among them, A represents the confidence level of lithium deposition detection corresponding to the lithium deposition state characteristics, Indicates the characteristic weight of the lithium deposition state feature in frequency band a, Indicates the rate of change of signal strength of frequency band a at time t, Indicates the attenuation reference value of the a-th frequency band, Indicates the detection cycle time interval, represents the offset suppression gain coefficient, Indicates the frequency offset safety threshold, represents the characteristic offset rate of the ath frequency band at time t, represents the attenuation adjustment coefficient, Indicates the sound wave attenuation coefficient of the ath frequency band, a represents the frequency band index, and q represents the number of frequency bands.

4. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The method of performing guided wave signal reproduction detection on a key electrode region of the lithium battery based on the lithium plating induction parameter to obtain signal reproduction data includes: Querying the signal reproduction detection standard corresponding to the lithium precipitation induction parameter; Screening the key electrode regions of the lithium battery based on the signal recurrence detection standard; According to the key electrode area, monitoring the guided wave signal characteristics corresponding to each dimension of the key electrode area; Calculating the deviation between the waveguide signal characteristic and a preset signal reference value to obtain a signal characteristic deviation; Based on the signal characteristic deviation, a waveguide signal reproduction detection is performed on the key electrode area of ​​the lithium battery to obtain signal reproduction data.

5. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The step of analyzing the lithium deposition transition mode of the lithium battery at different charge and discharge stages based on the signal reproduction data includes: extracting a recurring signal feature from the signal recurrence data; Determining a lithium plating sensitive frequency band corresponding to the lithium battery according to the reproduced signal characteristics; Quantifying the signal amplitude change rate corresponding to the lithium plating sensitive frequency band; Evaluating local lithium deposition differences corresponding to the lithium battery based on the signal amplitude change rate; Based on the local lithium deposition differences, the lithium deposition transition modes of the lithium battery at different charge and discharge stages are analyzed.

6. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The step of dividing the lithium deposition detection level corresponding to the key electrode area based on the lithium deposition transition mode and the current waveguide signal distribution state of the lithium battery includes: Extracting the signal transition rate in the lithium deposition transition mode; Based on the current distribution state of the guided wave signal, locating the guided wave signal concentration area within the key electrode area; Calculating the signal transition rate and the spatiotemporal coupling coefficient of the waveguide signal concentration area; distinguishing risk gradient zones within the key electrode region based on the spatiotemporal coupling coefficient; The gradient distribution characteristics corresponding to the risk gradient zone are analyzed to divide the lithium plating detection level corresponding to the key electrode area.

7. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The calculating of the signal transition rate and the spatiotemporal coupling coefficient of the waveguide signal concentration area includes: Splitting the signal change rate by physical dimension to obtain a multi-dimensional signal feature vector; Normalizing the multidimensional signal feature vector to obtain a target signal feature vector; Calculating the geometric aggregation eigenvalue corresponding to the waveguide signal aggregation area, and normalizing the geometric aggregation eigenvalue to obtain a target aggregation eigenvalue; Performing spatiotemporal weight allocation processing on the target signal eigenvector and the target aggregate eigenvalue respectively to obtain a first spatiotemporal weight and a second spatiotemporal weight; Combining the target signal eigenvector, the target aggregation eigenvalue, the first spatiotemporal weight, and the second spatiotemporal weight, the following formula is used to calculate the spatiotemporal coupling coefficient between the signal transition rate and the waveguide signal aggregation area, including: Where C represents the spatiotemporal coupling coefficient between the signal transition rate and the waveguide signal aggregation area, represents the first spatiotemporal weight of the dth vector in the target signal feature vector, represents the dth vector in the target signal feature vector, represents the second spatiotemporal weight of the mth eigenvalue in the target aggregated eigenvalue, represents the mth eigenvalue in the target aggregate eigenvalue, represents the spatiotemporal correlation factor, d and m represent the sequence numbers corresponding to the target signal feature vector and the target aggregate feature value, respectively, and q and p represent the number of target signal feature vectors and the target aggregate feature value, respectively.

8. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The configuration of the detection accuracy waveguide parameters corresponding to the lithium battery to perform detection processing on the key electrode area and locate the abnormal signal area of ​​the lithium battery during the detection process also includes: Constructing a waveguide detection signal corresponding to the detection accuracy waveguide parameter; Injecting the guided wave detection signal into the key electrode area of ​​the lithium battery and collecting the reflected guided wave signal of the key electrode area; performing noise reduction processing on the reflected waveguide signal to obtain a noise-reduced waveguide signal, and extracting a signal time-frequency feature of the noise-reduced waveguide signal; Calculating the signal deviation corresponding to the noise reduction waveguide signal based on the signal time-frequency characteristics; Based on the signal deviation, an abnormal signal area of ​​the lithium battery during the detection process is located.

9. The nondestructive testing method for lithium battery lithium deposition based on ultrasonic guided waves according to claim 1, characterized in that: The calculating the signal feature matching degree corresponding to the signal abnormal area includes: Performing feature tensor mapping processing on the abnormal signal area to obtain a three-dimensional tensor in the time-frequency domain; Based on the three-dimensional tensor in the time-frequency domain, construct a signal feature map corresponding to the abnormal signal area; Performing graph convolution feature extraction processing on the signal feature graph to obtain a high-order structural feature vector; Matching the high-order structural feature vector with the standard feature vector in a preset lithium precipitation feature library to obtain a similarity matrix; The dynamic similarity matrix is ​​corrected to obtain the signal feature matching degree corresponding to the signal abnormal area.

10. A non-destructive testing device for lithium deposition in lithium batteries based on ultrasonic guided waves, characterized in that: The device comprises: a lithium deposition detection confidence calculation module, configured to obtain an ultrasonic guided wave detection signal and battery cycle history data of a lithium battery to be detected, analyze the lithium deposition state of the lithium battery based on the ultrasonic guided wave detection signal and the battery cycle history data, obtain a lithium deposition state characteristic, and calculate a lithium deposition detection confidence corresponding to the lithium deposition state characteristic; a reproduction detection processing module, configured to analyze the potential lithium deposition risk characteristics of the lithium battery based on the lithium deposition detection confidence level, query the lithium deposition induction parameters corresponding to the potential lithium deposition risk characteristics, and perform guided wave signal reproduction detection on the key electrode area of ​​the lithium battery based on the lithium deposition induction parameters to obtain signal reproduction data; a lithium deposition detection level classification module, configured to analyze the lithium deposition transition mode of the lithium battery at different charge and discharge stages based on the signal reproduction data, and classify the lithium deposition detection level corresponding to the key electrode area based on the lithium deposition transition mode and the current guided wave signal distribution state of the lithium battery; The lithium plating detection processing module is used to configure the detection accuracy waveguide parameters corresponding to the lithium battery based on the lithium plating detection level, so as to perform detection processing on the key electrode area, locate the signal abnormality area of ​​the lithium battery during the detection process, calculate the signal feature matching degree corresponding to the signal abnormality area, and stop the detection when the signal feature matching degree is greater than the preset matching degree, and generate the lithium plating detection result of the lithium battery.

Images