A fault identification method, system, device, equipment, and storage medium

Abstract

This application discloses a fault identification method, system, device, equipment, and storage medium, belonging to the field of electrical engineering technology. The method obtains wavelet coefficients by performing continuous wavelet transform on the voltage signal of the battery cell, and determines the instantaneous frequency estimate based on the wavelet coefficients. The instantaneous frequency estimate is then corrected using a set of harmonic frequencies to obtain a rearranged frequency. Combined with a dynamic weight matrix that can adjust the compression intensity of different time-frequency regions, the wavelet coefficients are compressed and transformed to obtain time-frequency information. Finally, fault identification is carried out based on this time-frequency information. This method can effectively converge harmonic energy, improve the resolution and feature clarity of time-frequency information, and adaptively adapt to the compression requirements of different time-frequency regions. This effectively captures the weak fault signal features under complex operating conditions, significantly improves the accuracy of battery cell fault identification, and reduces the probability of false alarms and missed alarms.

Description

Technical Field

[0001] This application belongs to the field of electrical engineering technology, and in particular relates to a fault identification method, system, device, equipment and storage medium. Background Technology

[0002] With the rapid development of new energy vehicles and energy storage systems, lithium batteries are widely used in various power consumption scenarios due to their high energy density and good cycle performance. However, lithium batteries are susceptible to complex operating conditions in actual operation, and may cause safety accidents such as thermal runaway due to overcharging, internal short circuits, mechanical damage, etc. Therefore, accurate identification and early warning of lithium battery faults are crucial to ensuring the safe operation of the system. Traditional lithium battery fault diagnosis methods mostly rely on single feature value fixed threshold alarms or data-driven modeling, which are difficult to effectively capture fault signal characteristics under complex operating conditions, easily leading to false alarms or missed alarms, and failing to meet the requirements of high-precision fault identification.

[0003] Therefore, how to effectively capture the characteristics of fault signals under complex working conditions, thereby improving the accuracy of fault identification, is a technical problem that urgently needs to be solved. Summary of the Invention

[0004] This application provides a fault identification method, system, device, equipment, and storage medium that can effectively capture fault signal characteristics under complex operating conditions and significantly improve the accuracy of fault identification.

[0005] In a first aspect, embodiments of this application provide a fault identification method, including: Continuous wavelet transform is performed on the voltage signal of the battery cell to obtain the wavelet coefficients corresponding to each scale at each sampling time. Based on the wavelet coefficients corresponding to each scale at each sampling time, the instantaneous frequency estimate of the battery cell at each scale at each sampling time is determined; The instantaneous frequency estimates corresponding to each scale at each sampling time are corrected based on the harmonic frequency set of the voltage signal to obtain the rearranged frequency corresponding to each scale at each sampling time. Based on the dynamic weight matrix and the rearranged frequencies corresponding to each scale at each sampling time, the wavelet coefficients corresponding to each scale at each sampling time are compressed and transformed to obtain the time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity of different time-frequency regions. Based on the time-frequency information corresponding to each frequency point at each sampling time, the cell is fault identified to obtain the fault identification result.

[0006] Secondly, embodiments of this application provide a fault identification system, including: The controller performs continuous wavelet transform on the voltage signal of the battery cell to obtain wavelet coefficients corresponding to each scale at each sampling time; based on the wavelet coefficients corresponding to each scale at each sampling time, it determines the instantaneous frequency estimate of the battery cell at each scale at each sampling time; it corrects the instantaneous frequency estimate corresponding to each scale at each sampling time according to the harmonic frequency set of the voltage signal to obtain the rearranged frequency corresponding to each scale at each sampling time; based on the dynamic weight matrix and the rearranged frequency corresponding to each scale at each sampling time, it performs compression transform on the wavelet coefficients corresponding to each scale at each sampling time to obtain the time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity in different time-frequency regions; based on the time-frequency information corresponding to each frequency point at each sampling time, it performs fault identification on the battery cell to obtain the fault identification result.

[0007] Thirdly, embodiments of this application provide a battery device including at least one battery cell and a fault identification system as described in the second aspect.

[0008] Fourthly, embodiments of this application provide an electronic device, the device including: a processor and a memory storing computer program instructions; The processor implements the fault identification method as described in the first aspect when executing computer program instructions.

[0009] Fifthly, embodiments of this application provide a computer storage medium on which computer program instructions are stored, and when the computer program instructions are executed by a processor, they implement the fault identification method as described in the first aspect.

[0010] Sixthly, embodiments of this application provide a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to perform the fault identification method as described in the first aspect.

[0011] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects: This application provides a fault identification method, comprising: performing continuous wavelet transform on the voltage signal of a battery cell to obtain wavelet coefficients corresponding to each scale at each sampling time; determining the instantaneous frequency estimate of the battery cell at each scale at each sampling time based on the wavelet coefficients corresponding to each scale at each sampling time; correcting the instantaneous frequency estimate at each scale at each sampling time according to the harmonic frequency set of the voltage signal to obtain the rearranged frequency at each scale at each sampling time; performing compression transform on the wavelet coefficients at each scale at each sampling time based on a dynamic weight matrix and the rearranged frequency at each scale at each sampling time to obtain time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity in different time-frequency regions; and performing fault identification on the battery cell based on the time-frequency information corresponding to each frequency point at each sampling time to obtain a fault identification result.

[0012] The technical solution provided in this application obtains wavelet coefficients by performing continuous wavelet transform on the voltage signal of the battery cell, and determines the instantaneous frequency estimate based on the wavelet coefficients; the instantaneous frequency estimate is corrected by using a set of harmonic frequencies to obtain a rearranged frequency, and then the wavelet coefficients are compressed and transformed by a dynamic weight matrix that can adjust the compression intensity of different time and frequency regions to obtain time and frequency information. Finally, fault identification is carried out based on this time and frequency information, which can effectively converge harmonic energy, improve the resolution and feature clarity of time and frequency information, and adaptively adapt to the compression requirements of different time and frequency regions, thereby effectively capturing the weak fault signal features under complex operating conditions, significantly improving the accuracy of battery cell fault identification, and reducing the probability of false alarms and missed alarms.

[0013] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A flowchart illustrating a fault identification method provided in one embodiment of this application; Figure 2 A voltage time-frequency diagram of a fault identification method provided in one embodiment of this application; Figure 3 A schematic diagram of the structure of a fault identification system provided in another embodiment of this application; Figure 4This is a schematic diagram of the structure of an electronic device provided in yet another embodiment of this application. Detailed Implementation

[0016] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0017] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0018] Traditional lithium battery fault diagnosis methods mainly rely on fixed threshold alarms based on single feature values ​​or data-driven modeling, while some methods are based on time-frequency analysis for fault early warning and identification. These methods lack adaptability when dealing with complex operating conditions, struggle to effectively capture fault signal characteristics under complex conditions, and are prone to false alarms or missed alarms, failing to meet the requirements for high-precision fault identification.

[0019] Based on the aforementioned technical problems, embodiments of this application provide a fault identification method, system, device, equipment, and storage medium. The method includes: acquiring the current state of charge (SOC) of a battery; determining a target confidence factor corresponding to the current SOC based on the correlation between the SOC and a confidence factor, wherein the target confidence factor characterizes the reliability of the current model parameters of a battery state management model, and the battery state management model simulates the voltage response characteristics of the battery; adjusting the current model parameters according to the target confidence factor to obtain target model parameters of the battery state management model; and updating the parameters of the battery state management model according to the target model parameters to obtain a parameter-updated battery state management model.

[0020] The technical solution provided in this application obtains wavelet coefficients by performing continuous wavelet transform on the voltage signal of the battery cell, and determines the instantaneous frequency estimate based on the wavelet coefficients; the instantaneous frequency estimate is corrected by using a set of harmonic frequencies to obtain a rearranged frequency, and then the wavelet coefficients are compressed and transformed by a dynamic weight matrix that can adjust the compression intensity of different time and frequency regions to obtain time and frequency information. Finally, fault identification is carried out based on this time and frequency information, which can effectively converge harmonic energy, improve the resolution and feature clarity of time and frequency information, and adaptively adapt to the compression requirements of different time and frequency regions, thereby effectively capturing the weak fault signal features under complex operating conditions, significantly improving the accuracy of battery cell fault identification, and reducing the probability of false alarms and missed alarms.

[0021] Regarding the execution entity used in the embodiments of this application, it can specifically be a fault identification system capable of monitoring the battery's terminal voltage and current, or other electronic devices capable of controlling the fault identification system, such as desktop computers, laptops, etc., or servers, etc. In addition, the execution entity in the embodiments of this application can also be a software entity, such as a client or software program installed in the fault identification system. The specific type of execution entity corresponding to the fault identification method, system, device, equipment, and storage medium provided in the embodiments of this application is not strictly limited here; it can be flexibly selected and set according to the application scenario and actual needs.

[0022] It should be noted that the specific application scenarios of the fault identification method, system, device, equipment and storage medium provided in the embodiments of this application are not limited. The technical solutions provided in the embodiments of this application can be flexibly applied to various actual scenarios that require fault identification according to actual needs.

[0023] It should be noted that the application scenarios described in the above embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will understand that with the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0024] Figure 1 This is a schematic diagram of a fault identification process provided in one embodiment of this application.

[0025] like Figure 1 As shown, the fault identification method provided in this application includes S101 to S105.

[0026] S101: Perform continuous wavelet transform on the voltage signal of the battery cell to obtain the wavelet coefficients corresponding to each scale at each sampling time.

[0027] S102: Based on the wavelet coefficients corresponding to each scale at each sampling time, determine the instantaneous frequency estimate of the battery cell at each sampling time at each scale.

[0028] S103: Based on the set of harmonic frequencies of the voltage signal, the instantaneous frequency estimates corresponding to each scale at each sampling time are corrected to obtain the rearranged frequencies corresponding to each scale at each sampling time.

[0029] S104: Based on the dynamic weight matrix and the rearranged frequencies corresponding to each scale at each sampling time, the wavelet coefficients corresponding to each scale at each sampling time are compressed and transformed to obtain the time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity of different time-frequency regions.

[0030] S105: Based on the time-frequency information corresponding to each frequency point at each sampling time, perform fault identification on the battery cell to obtain the fault identification result.

[0031] In the aforementioned fault identification method, this application obtains wavelet coefficients by performing continuous wavelet transform on the voltage signal of the battery cell, and determines the instantaneous frequency estimate based on the wavelet coefficients; the instantaneous frequency estimate is corrected by using a set of harmonic frequencies to obtain a rearranged frequency, and then the wavelet coefficients are compressed by combining a dynamic weight matrix with adjustable compression intensity in different time-frequency regions to obtain time-frequency information. Finally, fault identification is carried out based on this time-frequency information, which can effectively converge harmonic energy, improve the resolution and feature clarity of time-frequency information, and adaptively adapt to the compression requirements of different time-frequency regions, thereby effectively capturing the weak fault signal features under complex operating conditions, significantly improving the accuracy of fault identification, and reducing the probability of false alarms and missed alarms.

[0032] In S101 above, the cell refers to the smallest electrochemical unit that constitutes the battery pack, and the battery type includes, but is not limited to, lithium batteries or other types of batteries; the voltage signal refers to the discrete sampling sequence of the terminal voltage at both ends of the cell changing over time.

[0033] The aforementioned voltage signal is a discrete sampling sequence of the terminal voltages at both ends of the battery cell changing over time. In this embodiment, the battery management system or voltage acquisition device acquires the original voltage signal of the battery cell at a fixed sampling frequency and time length; the voltage signal is filtered, and the processed original voltage signal is preprocessed, including but not limited to missing value imputation, outlier detection, and normalization, thereby obtaining the voltage signal of the battery cell. .

[0034] In some embodiments, a time series is determined based on the voltage signal of the battery cell. ,in The sampling time in the voltage signal is used; and the range of values ​​for the scale parameter is adaptively determined based on the sampling frequency of the voltage signal. , The scale, in time-frequency analysis, is the scaling parameter corresponding to the frequency dimension. The scale value is inversely proportional to the signal frequency; a smaller scale corresponds to high-frequency signal components, and a larger scale corresponds to low-frequency signal components. Besides the above method, a set of discrete scales can also be pre-defined to form a scale parameter set.

[0035] The process involves acquiring the voltage signal of the battery cell, selecting a mother wavelet function (e.g., complex Morlet wavelet) adapted to the characteristics of the low-frequency voltage signal, scaling the mother wavelet function according to each scale value, then translating it point-by-point along the time axis, and sequentially performing inner product operations with the voltage signal. After traversing all sampling times and all scales, the wavelet coefficients corresponding to each scale at each sampling time are obtained. Each wavelet coefficient corresponds to a sampling time and a scale, and its amplitude characterizes the energy intensity of the signal component at that time-frequency position. The sampling time is the discrete sampling moment of the voltage signal, corresponding one-to-one with the sampling frequency of the voltage signal, forming the time axis of the time-frequency representation.

[0036] As an optional implementation method, the specific transformation formula is as follows: .in, For the mother wavelet function; Represents the complex conjugate of the mother wavelet function; t represents the wavelet coefficients corresponding to scale a at sampling time b; t is the time variable used to represent any moment on the continuous time axis; b is the time shift factor, corresponding to each sampling time of the voltage signal. Let be the differential element when integrating over the time variable t.

[0037] In S102 above, instantaneous frequency estimation characterizes the local instantaneous frequency value of the signal at a certain time and at a certain scale. It is calculated by differentiating the phase information of the wavelet coefficients with respect to time, reflecting the local frequency change characteristics of the signal, and is the basis for subsequent frequency rearrangement.

[0038] In some embodiments, the instantaneous frequency estimate can be obtained by calculating the partial derivative of the phase angle of the wavelet coefficients with respect to time. The specific calculation formula is as follows: .in, This is an estimate of the instantaneous frequency corresponding to scale a at sampling time b. This represents taking the partial derivative with respect to the time shift factor; The phase angle of the wavelet coefficients is given. Besides the method described above, other methods can also be used to calculate the instantaneous frequency estimate, such as dividing the phase difference of the wavelet coefficients at adjacent sampling points by the time interval. This embodiment does not impose any restrictions on this method.

[0039] In S103 above, the harmonic frequency set is a set of frequencies consisting of the fundamental frequency of the signal and its integer multiples of harmonic frequencies, corresponding to the frequency components in the voltage signal that have harmonic correlation; rearranged frequency refers to the high-precision frequency value after harmonic correlation correction.

[0040] In some embodiments, by comparing the instantaneous frequency estimates corresponding to each scale at each sampling time with the theoretical frequencies in the harmonic frequency set, the instantaneous frequency estimates that deviate from the theoretical values ​​are attracted to the vicinity of the nearest harmonic frequency by utilizing the pulling effect of the theoretical harmonics, thereby filtering out frequency jitter caused by noise, enhancing the clarity of the harmonic structure, and thus completing the correction of the instantaneous frequency estimates to obtain the rearranged frequencies.

[0041] As an optional implementation, a soft threshold adsorption strategy is adopted. When the difference between the instantaneous frequency estimate and the nearest theoretical harmonic frequency is less than a preset threshold, it is directly corrected to the theoretical harmonic frequency; otherwise, the original value is kept unchanged to balance frequency alignment accuracy and noise resistance robustness. In addition to the above method, other methods can also be used for correction, and this embodiment does not limit them.

[0042] In S104 above, the dynamic weight matrix is ​​a two-dimensional array that changes with sampling time and scale, used to adaptively adjust the compression intensity in different time-frequency regions. The dynamic weight matrix can be constructed based on the local time-varying energy correlation of the voltage signal; or, an iterative reweighting strategy can be used to update the dynamic weight matrix, that is, based on the time-frequency energy distribution obtained from the previous compression transformation, the weight parameters for the next transformation are adjusted in reverse to achieve secondary energy focusing and further improve the clarity of the time-frequency representation. The time-frequency information includes the signal energy amplitude at each frequency point under each sampling time, and the signal energy amplitude is used to characterize the time-frequency energy distribution pattern of the signal.

[0043] In some embodiments, the wavelet coefficients in the scale domain are redistributed to the frequency axis according to the principle of energy conservation, based on the rearranged frequencies corresponding to each scale at each sampling time, and energy is aggregated by combining dynamic weights, thereby obtaining the time-frequency information of each frequency point at each sampling time.

[0044] In step S105 above, by analyzing the energy distribution characteristics in the time-frequency information, feature quantities that reflect the battery's health status are extracted to determine whether the cell has fault risks such as thermal runaway, and to assess the severity and development stage of the fault, thereby outputting the cell's fault identification result. The fault identification result can include graded information such as whether the battery is normal, in an early warning state, or has experienced substantial thermal runaway, providing maintenance personnel with intuitive decision-making basis.

[0045] As an optional implementation, the time-frequency information corresponding to each frequency point at each sampling time is converted into a two-dimensional time-frequency image, and then input into a pre-trained convolutional neural network (CNN) model. The CNN automatically extracts texture features from the two-dimensional time-frequency image, outputs the battery fault type and its corresponding probability, thereby obtaining the cell fault identification result. The CNN is trained using a set of time-frequency image samples labeled with fault types.

[0046] In some embodiments, the above-mentioned S103 can be performed in the following manner: For the instantaneous frequency estimates corresponding to each scale at each sampling time, determine the nearest neighbor theoretical harmonic frequency in the harmonic frequency set of the instantaneous frequency estimates; For each instantaneous frequency estimate corresponding to each scale at each sampling time, calculate the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency; For the instantaneous frequency estimates corresponding to each scale at each sampling time, the instantaneous frequency estimates are corrected based on the difference between the instantaneous frequency estimates and the nearest neighbor theoretical harmonic frequencies to obtain the rearranged frequencies corresponding to each scale at each sampling time.

[0047] The aforementioned nearest neighbor theoretical harmonic frequencies It is a frequency estimation for a specific instant, within the set of harmonic frequencies. The theoretical harmonic frequency that is smallest in value from the estimated instantaneous frequency.

[0048] In some embodiments, for the instantaneous frequency estimate corresponding to each sampling time and each scale, the harmonic frequency set is first traversed and matched to determine the frequency point with the smallest numerical difference from the current instantaneous frequency estimate, which is then taken as the nearest neighbor theoretical harmonic frequency. Subsequently, the difference between the current instantaneous frequency estimate and this nearest neighbor theoretical harmonic frequency is calculated. This difference carries both amplitude and direction attributes. Finally, this difference is used as the basis for correction, and the instantaneous frequency estimate is offset and adjusted so that the corrected rearranged frequency moves closer to the nearest neighbor theoretical harmonic frequency, thus completing the frequency rearrangement optimization and obtaining the rearranged frequency corresponding to each sampling time and each scale.

[0049] In this embodiment, by using the nearest neighbor theoretical harmonic frequency as a reference and correcting the instantaneous frequency estimate based on the frequency difference, the dispersed harmonic energy can be guided toward the theoretical harmonic ridge, effectively solving the problem of harmonic energy dispersion and significantly enhancing the time-frequency discernibility of multi-component harmonic signals.

[0050] In some embodiments, the instantaneous frequency estimate is corrected based on the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency to obtain the rearranged frequency corresponding to the scale at the sampling time. The processing flow includes: The correction direction is determined based on the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency; The correction step size is determined by multiplying the magnitude of the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency with the cooperative compression coefficient; the cooperative compression coefficient is used to characterize the intensity of harmonic convergence. The instantaneous frequency estimate is shifted according to the correction direction and the correction step size to obtain the rearranged frequency corresponding to the scale at the sampling time.

[0051] The aforementioned correction direction is the offset direction determined by the sign of the difference between the instantaneous frequency estimate and the nearest-neighbor theoretical harmonic frequency, through the sign function. Determine: If the difference is positive, correct in the direction of increasing frequency; if the difference is negative, correct in the direction of decreasing frequency.

[0052] Cooperative compression coefficient Used to characterize the intensity of harmonic convergence, when When =0, it degenerates into a traditional synchronous compression transform, with no harmonic co-correction effect; when When =1, the energy is forced to be fully aligned with the theoretical harmonic ridge; therefore, in this embodiment, the cooperative compression coefficient... The value range can be set to [0.2, 0.8] to achieve soft adsorption and avoid overfitting.

[0053] The correction step size refers to the magnitude of the shift of the instantaneous frequency estimate towards the nearest-neighbor theoretical harmonic frequency, which is determined by both the magnitude of the frequency difference and the cooperative compression coefficient. In this embodiment, the correction step size is the product of the magnitude of the difference and the cooperative compression coefficient, i.e. , The magnitude of the difference.

[0054] In some embodiments, the instantaneous frequency estimation for each sampling time and each scale is... Based on the instantaneous frequency estimation and the nearest neighbor theoretical harmonic frequency The difference The symbol, through the symbol function The correction direction is determined to ensure that the corrected frequency approaches the nearest-neighbor theoretical harmonic frequency. The product of the difference amplitude and the cooperative compression coefficient is calculated to obtain the correction step size, which is controlled by the cooperative compression coefficient to achieve harmonic convergence effects of different intensities. The instantaneous frequency estimate is shifted according to the determined correction direction and step size; the shifted value is the rearranged frequency corresponding to that scale at that sampling time. .

[0055] Specifically, the formula for calculating the rearrangement frequency can be: ,in, This is a harmonic correlation function that maps the instantaneous frequency to the detected set of dominant or harmonic frequencies. Adsorption is used to address the problem of harmonic energy dispersion, and its mathematical definition is as follows: ,in, Estimation of instantaneous frequency at a given sampling time and scale , Set of harmonic frequencies Any theoretical harmonic frequency within.

[0056] In this embodiment, frequency shift is achieved by determining the correction direction and correction step size respectively. The intensity of harmonic convergence can be flexibly adjusted with the help of the cooperative compression coefficient. While optimizing the frequency focusing effect, soft adsorption of harmonic energy is achieved, avoiding overfitting caused by excessive alignment, and taking into account both time and frequency resolution and signal authenticity.

[0057] In some embodiments, the above-described S104 can be implemented according to the following steps: Based on the rearrangement frequency corresponding to each scale at each sampling time, a mapping relationship between scale parameters and frequency points is established; Based on the dynamic weight matrix, the wavelet coefficients corresponding to each scale at each sampling time are weighted to obtain the weighted wavelet coefficients corresponding to each scale at each sampling time. Based on the mapping relationship, the weighted wavelet coefficients corresponding to each scale at each sampling time are rearranged and aggregated from the scale axis to the frequency axis to obtain the time-frequency information corresponding to each frequency point at each sampling time.

[0058] The above mapping relationship is based on the rearrangement frequency corresponding to each scale at each sampling time. Established scale a and frequency points The correspondence between them, that is, each rearranged frequency value corresponds to a frequency point. .

[0059] In some embodiments, the rearrangement frequency is based on the rearrangement frequency corresponding to each scale a at each sampling time b. Establish scale a and frequency points The mapping relationship between them is clarified, and the frequency point ω corresponding to the rearrangement frequency at each scale a is defined, providing a basis for subsequent energy rearrangement; the dynamic weight moments are mapped to the wavelet coefficients corresponding to each scale a at each sampling time b. Perform a dot product operation to obtain the weighted wavelet coefficients at each sampling time b and at each scale a. This enables adaptive adjustment of compression intensity in different time-frequency regions.

[0060] Based on the established mapping relationship, all weighted wavelet coefficients at scale a are rearranged to the frequency axis according to their corresponding rearrangement frequencies. The frequency points at each sampling time b are obtained by integration and aggregation. The corresponding time-frequency information completes the energy mapping from the scale domain to the frequency domain.

[0061] In this embodiment, by establishing a scale-frequency mapping relationship and combining dynamic weights to weight wavelet coefficients and then rearranging and aggregating them from the scale axis to the frequency axis, it is possible to achieve differentiated adjustment of compression intensity in different time-frequency regions, complete efficient energy aggregation from the scale domain to the frequency domain, and effectively improve the overall focusing performance of the time-frequency representation.

[0062] In some embodiments, according to the mapping relationship, the weighted wavelet coefficients corresponding to each scale at each sampling time are rearranged and aggregated from the scale axis to the frequency axis to obtain the time-frequency information corresponding to each frequency point at each sampling time. The specific processing steps include: For each frequency point at each sampling time, all target scales whose rearranged frequencies match the frequency point are selected from multiple scales at the sampling time based on the mapping relationship; For each frequency point at each sampling time, the weighted wavelet coefficients corresponding to each target scale are integrated and aggregated along the scale dimension to obtain the time-frequency information corresponding to the frequency point at the sampling time.

[0063] In some embodiments, for each frequency point at each sampling time b Based on the mapping relationship established in step S401, the rearrangement frequency is selected from all scales a corresponding to the current sampling time b. With frequency point All matching target scales a. In this embodiment, the matching condition is that the absolute value of the difference between the rearranged frequency and the frequency point is less than a preset frequency tolerance threshold, i.e. .

[0064] For each frequency point at each sampling time b The weighted wavelet coefficients corresponding to all target scales a obtained through the above screening are... Integrating and aggregating along scale dimension a, and summing the energy of all matching scales, yields the frequency point at that sampling time b. Corresponding time-frequency coefficients This allows us to obtain the time-frequency information, including the time-frequency coefficients, corresponding to the frequency point at that sampling time.

[0065] Specifically, for each frequency point at each sampling time b Time-frequency coefficient The calculation formula can be: , For Diclave function, it represents the rearrangement of energy onto the frequency axis; Let be a differential element of scale a.

[0066] In this embodiment, by filtering the target scales that match each frequency point and performing integral aggregation, the energy aggregation operation corresponding to each frequency point can be accurately completed, ensuring the accuracy of the time-frequency coefficient calculation and further improving the energy concentration and feature clarity of the time-frequency representation.

[0067] In some embodiments, the processing flow in S105 above may be as follows: From the time-frequency information corresponding to each frequency point at each sampling time, high-frequency components with frequencies greater than a preset frequency threshold and low-frequency components with frequencies less than or equal to the preset frequency threshold are extracted respectively. If the energy percentage of the high-frequency component is higher than a preset energy threshold within a preset time window, the cell is determined to be in the first state of thermal runaway. When the energy proportion of the low-frequency component shows a continuous increasing trend and meets the preset growth conditions, the cell is determined to be in a second state of thermal runaway; wherein the degree of thermal runaway in the second state is greater than the degree of thermal runaway in the first state.

[0068] The aforementioned preset frequency threshold is a critical value used to distinguish between high-frequency and low-frequency components. In this embodiment, it can be set to 0.05Hz, or it can be dynamically adjusted according to actual conditions. This embodiment does not impose any restrictions on this. The preset time window is a continuous sampling interval used to monitor the energy proportion of high-frequency components, and is used to capture abnormal characteristics in the early stages of thermal runaway. The preset growth condition is a judgment condition used to determine the abnormal growth of low-frequency components, including the growth rate of the energy proportion of low-frequency components exceeding the preset threshold, and the energy proportion of low-frequency components continuously increasing within the continuous time window.

[0069] The first state of thermal runaway mentioned above represents the early stage of thermal runaway, characterized by the appearance of abnormal energy in the high-frequency region and the beginning of changes in the internal state of the battery; the second state of thermal runaway represents the stage of intensified thermal runaway, characterized by the diffusion of abnormal energy from high frequency to mid-low frequency, and the degree of thermal runaway is higher than that of the first state.

[0070] In some embodiments, from each frequency point at each sampling time b In the corresponding time-frequency information, high-frequency components with frequencies greater than a preset frequency threshold and low-frequency components with frequencies less than or equal to the preset frequency threshold are extracted respectively. The energy ratio of the high-frequency component within a preset time window is calculated. If the energy ratio is continuously higher than the preset energy threshold, it indicates that high-frequency abnormal characteristics have appeared in the signal, and the cell is determined to be in the first state of thermal runaway. The energy ratio of the low-frequency component is continuously monitored. If the energy ratio of the low-frequency component shows a continuous increasing trend and meets the preset growth condition, it indicates that the energy has spread from high frequency to mid-low frequency, and the degree of thermal runaway fault has been further aggravated, and the cell is determined to be in the second state of thermal runaway.

[0071] like Figure 2 As shown, Figure 2 This is a time-frequency plot of the voltage of a faulty battery cell. The voltage signal is sampled at a frequency of 0.1 Hz. The horizontal axis represents sampling time in seconds (s), and the coordinates of the horizontal axis are generated based on the number of consecutive sampling points. The vertical axis represents frequency in hertz (Hz). The color bars on the right represent the signal energy at each time-frequency position in dB (dB), where red and orange areas indicate high energy, yellow and green areas indicate medium energy, and blue areas indicate low energy. Figure 2 As can be seen, before the 15000 sampling point on the horizontal axis, the energy is mainly blue and the time-frequency distribution is relatively flat; at the 15000 sampling point on the horizontal axis, a bright high-frequency part (frequency greater than 0.05Hz) appears, indicating that the internal state of the battery has begun to change; thereafter, the energy gradually diffuses from the high-frequency orange area to the mid-low frequency red area, indicating that the single cell is experiencing thermal runaway and the degree is gradually increasing.

[0072] In this embodiment, by dividing high and low frequency components and setting two-level thermal runaway state judgment rules, the early warning stage and fault aggravation stage of battery thermal runaway can be accurately identified. It can capture weak early fault characteristics and clearly assess the degree of fault development, thereby improving the accuracy and practicality of fault identification classification.

[0073] In some embodiments, the set of harmonic frequencies of the voltage signal described above can be determined by the following steps: The global spectral characteristics of the voltage signal are calculated based on the wavelet coefficients corresponding to each scale at each sampling time. Based on the global spectral characteristics, the dominant frequency component of the voltage signal is identified, and the dominant frequency component is determined as the fundamental frequency. Calculate the product of the fundamental frequency and at least one preset harmonic order to obtain the harmonic frequency corresponding to each preset harmonic order; The set of harmonic frequencies of the voltage signal is obtained based on the harmonic frequencies corresponding to each preset harmonic order.

[0074] For example, based on the wavelet coefficients at each sampling time b and at each scale a The global spectral characteristics of the voltage signal are calculated using spectral analysis or autocorrelation methods to obtain the energy distribution across the entire frequency band. Based on the global spectral characteristics, the dominant frequency component with the highest energy in the voltage signal is identified and determined as the fundamental frequency. ; Calculate the fundamental frequency respectively The product of the product with each preset harmonic order k yields the harmonic frequency k× corresponding to each preset harmonic order. By summing up the harmonic frequencies corresponding to all preset harmonic orders, a set of harmonic frequencies for the voltage signal is obtained. , where K is the maximum preset harmonic order.

[0075] The aforementioned global spectral characteristics can reflect the spectral characteristics of the energy distribution across the entire frequency band of the signal; the dominant frequency component is the frequency component with the highest energy in the global spectrum of the voltage signal, reflecting the main oscillation frequency of the signal; the preset harmonic order is a pre-set harmonic number, which is a positive integer, and the maximum preset harmonic order is a preset parameter used to limit the range of the harmonic frequency set.

[0076] In this embodiment, global spectral features are extracted based on wavelet coefficients and the fundamental frequency is determined to generate a set of harmonic frequencies. This fully leverages the advantages of time-frequency analysis based on wavelet transform to enhance the anti-interference capability of fundamental frequency detection, ensures the accuracy of the harmonic frequency reference, and provides a reliable reference for subsequent frequency rearrangement correction.

[0077] In some embodiments, the above dynamic weight matrix can be determined as follows: Based on the wavelet coefficients corresponding to each scale at each sampling time, calculate the local time-varying energy corresponding to each sampling time. An energy adaptive factor is constructed based on the local time-varying energy corresponding to each sampling time; the energy adaptive factor increases monotonically with the increase of local time-varying energy. The energy adaptive factor corresponding to each sampling time is multiplied by the scale correction factor corresponding to each scale to obtain a dynamic weight matrix; the scale correction factor is used to suppress interference in a preset scale range.

[0078] The above-mentioned local time-varying energy It is a parameter reflecting the overall energy level of the signal at a certain sampling time b, obtained by integrating the modulus square of the wavelet coefficients along scale a. The specific calculation formula is as follows: ; Let be a differential element of scale a.

[0079] Energy adaptive factor It is a monotonically increasing function of local time-varying energy, with values ​​ranging from [0, 1]; when the local time-varying energy... When the energy is relatively large (corresponding to high-energy components such as impacts and transients), the energy adaptive factor Approaching 1, strong compression is implemented; when the local time-varying energy... When the value is relatively small (corresponding to the stable region and the noisy region), the energy adaptive factor Approaching 0 preserves the original features and achieves adaptive adjustment.

[0080] Scale correction factor It is a pre-defined weighting function for scale a, used to suppress interference from invalid scale intervals (such as scales far from the effective frequency band of the signal) and improve the purity of the time-frequency representation. Each scale corresponds to a scale correction factor, the value of which can be a pre-defined fixed value or a value dynamically generated according to the Sigmoid attenuation function. This embodiment does not limit this.

[0081] In some embodiments, an energy adaptive factor is constructed based on the local time-varying energy corresponding to each sampling time b, ensuring that it increases monotonically with the increase of local time-varying energy and satisfies the value constraint of [0,1], thereby achieving an adaptive effect of strong compression in high-energy regions and weak compression in low-energy regions. The energy adaptive factor corresponding to each sampling time b is multiplied by the scale correction factor corresponding to each scale a, i.e., corresponding element-wise multiplication, to obtain the dynamic weight matrix. , .

[0082] In this embodiment, a monotonically increasing energy adaptive factor is constructed by using local time-varying energy, and a dynamic weight matrix is ​​generated by combining it with a scale correction factor. This enables adaptive adjustment of strong compression in high-energy impact regions and weak compression in low-energy stable regions, while suppressing interference from invalid scale intervals, effectively improving the robustness and noise resistance of the algorithm.

[0083] For example, the overall process of the fault identification method provided in this embodiment is as follows: First, the voltage signal of the battery cell in the lithium battery is preprocessed; then, wavelet transform is performed on the preprocessed voltage signal to obtain a time-frequency matrix, and based on the time-frequency matrix, instantaneous frequency estimation, local time-varying energy calculation and harmonic detection are respectively extracted; a dynamic weighting function is constructed based on the local time-varying energy, and frequency-scale collaborative rearrangement is performed on the instantaneous frequency estimation in combination with the harmonic detection results; finally, based on the dynamic weighting function, the collaborative rearrangement result and the time-frequency matrix, dynamic scale-frequency collaborative compression transformation of time-frequency information is performed to obtain a high-resolution time-frequency representation, and then the potential fault risk of the battery cell is accurately assessed and warned by drawing a time-frequency diagram.

[0084] Based on the same inventive concept as the above-mentioned fault identification method, this application also provides a fault identification system, which can be referred to in detail. Figure 3 As shown.

[0085] Figure 3 This is a schematic diagram of the structure of a fault identification system provided in another embodiment of this application.

[0086] like Figure 3 As shown in the figure, this application embodiment also provides a fault identification system 300, including: The controller 301 is used to perform continuous wavelet transform on the voltage signal of the battery cell to obtain wavelet coefficients corresponding to each scale at each sampling time; based on the wavelet coefficients corresponding to each scale at each sampling time, determine the instantaneous frequency estimate of the battery cell at each scale at each sampling time; correct the instantaneous frequency estimate corresponding to each scale at each sampling time according to the harmonic frequency set of the voltage signal to obtain the rearranged frequency corresponding to each scale at each sampling time; perform compression transform on the wavelet coefficients corresponding to each scale at each sampling time based on the dynamic weight matrix and the rearranged frequency corresponding to each scale at each sampling time to obtain the time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity of different time-frequency regions; and perform fault identification on the battery cell based on the time-frequency information corresponding to each frequency point at each sampling time to obtain the fault identification result.

[0087] In some embodiments, the controller 301 may include a transformation module, a processing module, a correction module, and an identification module.

[0088] The aforementioned transformation module is used to perform continuous wavelet transform on the voltage signal of the battery cell to obtain the wavelet coefficients corresponding to each scale at each sampling time. The above processing module is used to determine the instantaneous frequency estimate of the battery cell at each sampling time and at each scale based on the wavelet coefficients corresponding to each scale at each sampling time. The aforementioned correction module is used to correct the instantaneous frequency estimate corresponding to each scale at each sampling time based on the harmonic frequency set of the voltage signal, so as to obtain the rearranged frequency corresponding to each scale at each sampling time. The aforementioned transformation module is further configured to perform compression transformation on the wavelet coefficients corresponding to each scale at each sampling time based on the dynamic weight matrix and the rearranged frequencies corresponding to each scale at each sampling time, thereby obtaining the time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity of different time-frequency regions. The aforementioned identification module is used to identify faults in the battery cell based on the time-frequency information corresponding to each frequency point at each sampling time, and to obtain fault identification results.

[0089] In some embodiments, the controller 301 described above is specifically used for: For the instantaneous frequency estimates corresponding to each scale at each sampling time, determine the nearest neighbor theoretical harmonic frequency in the harmonic frequency set of the instantaneous frequency estimates; For each instantaneous frequency estimate corresponding to each scale at each sampling time, calculate the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency; For the instantaneous frequency estimates corresponding to each scale at each sampling time, the instantaneous frequency estimates are corrected based on the difference between the instantaneous frequency estimates and the nearest neighbor theoretical harmonic frequencies to obtain the rearranged frequencies corresponding to each scale at each sampling time.

[0090] In some embodiments, the controller 301 described above is specifically used for: The correction direction is determined based on the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency; The correction step size is determined by multiplying the magnitude of the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency with the cooperative compression coefficient; the cooperative compression coefficient is used to characterize the intensity of harmonic convergence. The instantaneous frequency estimate is shifted according to the correction direction and the correction step size to obtain the rearranged frequency corresponding to the scale at the sampling time.

[0091] In some embodiments, the controller 301 described above is specifically used for: Based on the rearrangement frequency corresponding to each scale at each sampling time, a mapping relationship between scale parameters and frequency points is established; Based on the dynamic weight matrix, the wavelet coefficients corresponding to each scale at each sampling time are weighted to obtain the weighted wavelet coefficients corresponding to each scale at each sampling time. Based on the mapping relationship, the weighted wavelet coefficients corresponding to each scale at each sampling time are rearranged and aggregated from the scale axis to the frequency axis to obtain the time-frequency information corresponding to each frequency point at each sampling time.

[0092] In some embodiments, the controller 301 is further configured to: For each frequency point at each sampling time, all target scales whose rearranged frequencies match the frequency point are selected from multiple scales at the sampling time based on the mapping relationship; For each frequency point at each sampling time, the weighted wavelet coefficients corresponding to each target scale are integrated and aggregated along the scale dimension to obtain the time-frequency information corresponding to the frequency point at the sampling time.

[0093] In some embodiments, the controller 301 described above is specifically used for: From the time-frequency information corresponding to each frequency point at each sampling time, high-frequency components with frequencies greater than a preset frequency threshold and low-frequency components with frequencies less than or equal to the preset frequency threshold are extracted respectively. If the energy percentage of the high-frequency component is higher than a preset energy threshold within a preset time window, the cell is determined to be in the first state of thermal runaway. When the energy proportion of the low-frequency component shows a continuous increasing trend and meets the preset growth conditions, the cell is determined to be in a second state of thermal runaway; wherein the degree of thermal runaway in the second state is greater than the degree of thermal runaway in the first state.

[0094] In some embodiments, the controller 301 is further configured to: The global spectral characteristics of the voltage signal are calculated based on the wavelet coefficients corresponding to each scale at each sampling time. Based on the global spectral characteristics, the dominant frequency component of the voltage signal is identified, and the dominant frequency component is determined as the fundamental frequency. Calculate the product of the fundamental frequency and at least one preset harmonic order to obtain the harmonic frequency corresponding to each preset harmonic order; The set of harmonic frequencies of the voltage signal is obtained based on the harmonic frequencies corresponding to each preset harmonic order.

[0095] In some embodiments, the controller 301 is specifically used to: obtain historical model parameters of the battery state management model when the target confidence factor is greater than or equal to a preset factor threshold; Based on the wavelet coefficients corresponding to each scale at each sampling time, calculate the local time-varying energy corresponding to each sampling time. An energy adaptive factor is constructed based on the local time-varying energy corresponding to each sampling time; the energy adaptive factor increases monotonically with the increase of local time-varying energy. The energy adaptive factor corresponding to each sampling time is multiplied by the scale correction factor corresponding to each scale to obtain the dynamic weight matrix; the scale correction factor is used to suppress interference in the preset scale range.

[0096] The Battery Management System (BMS) in this application is used to perform at least one of the following functions for battery cells: state monitoring, state analysis, charge / discharge control, safety protection, thermal management, high-voltage power distribution, and information management. In addition, the BMS in this application can also perform the functions of a controller in an electrical device, such as a vehicle control unit (VCU) or a motor control unit (MCU), etc., and this application does not impose any limitations on this.

[0097] It should be noted that the fault identification system in this application can be integrated as a controller into the battery device, such as into the battery pack or energy storage box. The fault identification system in this application can also be integrated as a controller into electrical devices, such as in a vehicle or vehicle chassis. The fault identification system in this application can also be integrated into the charging device as a controller, such as into the charging device or the battery swapping device. The fault identification system in this application can also be deployed as control software on a server. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms, such as vehicle networking cloud, APP backend, etc.

[0098] Based on the same inventive concept, embodiments of this application also provide a battery device, including a battery and a fault identification system as described in the above embodiments.

[0099] Figure 4 This is a schematic diagram of the structure of an electronic device provided in yet another embodiment of this application.

[0100] The electronic device may include a processor 401 and a memory 402 storing computer program instructions.

[0101] Specifically, the processor 401 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0102] Memory 402 may include mass storage for data or instructions. For example, and not limitingly, memory 402 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 402 may include removable or non-removable (or fixed) media. Where appropriate, memory 402 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 402 is non-volatile solid-state memory.

[0103] In a particular embodiment, memory 402 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Thus, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform operations described with reference to any of the fault identification methods disclosed in this application.

[0104] The processor 401 implements any of the fault identification methods described in the above embodiments by reading and executing computer program instructions stored in the memory 402.

[0105] In one example, the electronic device may also include a communication interface 403 and a bus 410. For example, Figure 4 As shown, the processor 401, memory 402, and communication interface 403 are connected through bus 410 and complete communication with each other.

[0106] The communication interface 403 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0107] Bus 410 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 410 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0108] Furthermore, in conjunction with the fault identification methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the fault identification methods in the above embodiments.

[0109] This application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the fault identification methods described in the above embodiments.

[0110] The above are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A fault identification method, characterized in that, The method includes: Continuous wavelet transform is performed on the voltage signal of the battery cell to obtain the wavelet coefficients corresponding to each scale at each sampling time. Based on the wavelet coefficients corresponding to each scale at each sampling time, the instantaneous frequency estimate of the battery cell at each scale at each sampling time is determined; The instantaneous frequency estimates corresponding to each scale at each sampling time are corrected based on the harmonic frequency set of the voltage signal to obtain the rearranged frequency corresponding to each scale at each sampling time. Based on the dynamic weight matrix and the rearranged frequencies corresponding to each scale at each sampling time, the wavelet coefficients corresponding to each scale at each sampling time are compressed and transformed to obtain the time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity of different time-frequency regions. Based on the time-frequency information corresponding to each frequency point at each sampling time, the cell is fault identified to obtain the fault identification result.

2. The method according to claim 1, characterized in that, The step of correcting the instantaneous frequency estimate corresponding to each scale at each sampling time based on the harmonic frequency set of the voltage signal to obtain the rearranged frequency corresponding to each scale at each sampling time includes: For the instantaneous frequency estimates corresponding to each scale at each sampling time, determine the nearest neighbor theoretical harmonic frequency in the harmonic frequency set of the instantaneous frequency estimates; For each instantaneous frequency estimate corresponding to each scale at each sampling time, calculate the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency; For the instantaneous frequency estimates corresponding to each scale at each sampling time, the instantaneous frequency estimates are corrected based on the difference between the instantaneous frequency estimates and the nearest neighbor theoretical harmonic frequencies to obtain the rearranged frequencies corresponding to each scale at each sampling time.

3. The method according to claim 2, characterized in that, The step of correcting the instantaneous frequency estimate based on the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency to obtain the rearranged frequency corresponding to the scale at the sampling time includes: The correction direction is determined based on the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency; The correction step size is determined by multiplying the magnitude of the difference between the instantaneous frequency estimate and the nearest neighbor theoretical harmonic frequency with the cooperative compression coefficient; the cooperative compression coefficient is used to characterize the intensity of harmonic convergence. The instantaneous frequency estimate is shifted according to the correction direction and the correction step size to obtain the rearranged frequency corresponding to the scale at the sampling time.

4. The method according to claim 1, characterized in that, The method involves performing a collaborative compression transformation on the wavelet coefficients corresponding to each scale at each sampling time, based on a dynamic weight matrix and the rearranged frequencies corresponding to each scale at each sampling time, to obtain the time-frequency information corresponding to each frequency point at each sampling time, including: Based on the rearrangement frequency corresponding to each scale at each sampling time, a mapping relationship between scale parameters and frequency points is established; Based on the dynamic weight matrix, the wavelet coefficients corresponding to each scale at each sampling time are weighted to obtain the weighted wavelet coefficients corresponding to each scale at each sampling time. Based on the mapping relationship, the weighted wavelet coefficients corresponding to each scale at each sampling time are rearranged and aggregated from the scale axis to the frequency axis to obtain the time-frequency information corresponding to each frequency point at each sampling time.

5. The method according to claim 4, characterized in that, The step of rearranging and aggregating the weighted wavelet coefficients corresponding to each scale at each sampling time from the scale axis to the frequency axis according to the mapping relationship, to obtain the time-frequency information corresponding to each frequency point at each sampling time, includes: For each frequency point at each sampling time, all target scales whose rearranged frequencies match the frequency point are selected from multiple scales at the sampling time based on the mapping relationship; For each frequency point at each sampling time, the weighted wavelet coefficients corresponding to each target scale are integrated and aggregated along the scale dimension to obtain the time-frequency information corresponding to the frequency point at the sampling time.

6. The method according to claim 1, characterized in that, The method of identifying faults in the battery cell based on the time-frequency information corresponding to each frequency point at each sampling time, and obtaining fault identification results, includes: From the time-frequency information corresponding to each frequency point at each sampling time, high-frequency components with frequencies greater than a preset frequency threshold and low-frequency components with frequencies less than or equal to the preset frequency threshold are extracted respectively. If the energy percentage of the high-frequency component is higher than a preset energy threshold within a preset time window, the cell is determined to be in the first state of thermal runaway. When the energy proportion of the low-frequency component shows a continuous increasing trend and meets the preset growth conditions, the cell is determined to be in a second state of thermal runaway; wherein the degree of thermal runaway in the second state is greater than the degree of thermal runaway in the first state.

7. The method according to any one of claims 1 to 6, characterized in that, The determination of the set of harmonic frequencies of the voltage signal includes: The global spectral characteristics of the voltage signal are calculated based on the wavelet coefficients corresponding to each scale at each sampling time. Based on the global spectral characteristics, the dominant frequency component of the voltage signal is identified, and the dominant frequency component is determined as the fundamental frequency. Calculate the product of the fundamental frequency and at least one preset harmonic order to obtain the harmonic frequency corresponding to each preset harmonic order; The set of harmonic frequencies of the voltage signal is obtained based on the harmonic frequencies corresponding to each preset harmonic order.

8. The method according to any one of claims 1 to 6, characterized in that, The determination of the dynamic weight matrix includes: Based on the wavelet coefficients corresponding to each scale at each sampling time, calculate the local time-varying energy corresponding to each sampling time. An energy adaptive factor is constructed based on the local time-varying energy corresponding to each sampling time; the energy adaptive factor increases monotonically with the increase of local time-varying energy. The energy adaptive factor corresponding to each sampling time is multiplied by the scale correction factor corresponding to each scale to obtain the dynamic weight matrix; the scale correction factor is used to suppress interference in the preset scale range.

9. A fault identification system, characterized in that, The system includes: The controller performs continuous wavelet transform on the voltage signal of the battery cell to obtain wavelet coefficients corresponding to each scale at each sampling time; based on the wavelet coefficients corresponding to each scale at each sampling time, it determines the instantaneous frequency estimate of the battery cell at each scale at each sampling time; it corrects the instantaneous frequency estimate corresponding to each scale at each sampling time according to the harmonic frequency set of the voltage signal to obtain the rearranged frequency corresponding to each scale at each sampling time; based on the dynamic weight matrix and the rearranged frequency corresponding to each scale at each sampling time, it performs compression transform on the wavelet coefficients corresponding to each scale at each sampling time to obtain the time-frequency information corresponding to each frequency point at each sampling time; the dynamic weight matrix is ​​used to adjust the compression intensity in different time-frequency regions; based on the time-frequency information corresponding to each frequency point at each sampling time, it performs fault identification on the battery cell to obtain the fault identification result.

10. A battery device, characterized in that, The device includes at least one battery cell and the fault identification system as described in claim 9.

11. An electronic device, characterized in that, The electronic device includes: a processor and a memory storing computer program instructions; The processor reads and executes the computer program instructions to implement the fault identification method as described in any one of claims 1-8.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the fault identification method as described in any one of claims 1-8.

Images