Battery failure early warning method, device, equipment, medium and product

Abstract

The application discloses a battery fault early warning method and device, equipment, medium and product, and relates to the technical field of energy storage. The application constructs a two-dimensional entropy domain plane suitable for actual working conditions in advance based on a battery historical voltage sequence, uses information entropy and spectral entropy to represent the time domain disorder degree and frequency domain dispersion degree of the eigenmode function component, filters effective components by distinguishing the differences between double-area noises, replaces the traditional single noise reduction mode, and accurately distinguishes effective fault characteristics and interference noises. On this basis, the voltage signal is reconstructed by superimposing the filtered effective components, invalid noise interference is filtered out, and the early weak fault detail characteristics of the battery are completely retained, which is beneficial to improving the signal-to-noise ratio of the signal. The battery fault early warning is performed based on the optimized high-quality reconstructed signal, the recognition deviation problem caused by noise interference is solved from the signal source, and the recognition accuracy and reliability of the early weak fault of the battery are improved.

Description

Technical Field

[0001] This application belongs to the field of energy storage technology, and in particular relates to a battery fault early warning method, device, equipment, medium and product. Background Technology

[0002] With the rapid development of new energy vehicles and energy storage systems, lithium batteries, as efficient and clean energy storage devices, are widely used in various fields. However, lithium batteries face complex and variable operating environments in practical applications, and are susceptible to factors such as overcharging, internal short circuits, or external mechanical damage, which can lead to safety hazards such as thermal runaway. To ensure the safe operation of lithium battery systems, accurate identification and location of early, subtle faults are particularly important.

[0003] Traditional lithium battery fault diagnosis methods still have limitations in dealing with the problem of fault feature obscuring caused by noise pollution in complex environments and the problem of false alarms and missed alarms caused by the confusion between operating condition disturbances and real fault signals. Summary of the Invention

[0004] The battery fault early warning method, device, equipment, medium, and product provided in this application can accurately distinguish between effective fault characteristics and interference noise, solve the identification deviation problem caused by noise interference from the signal source, and help improve the accuracy and reliability of early weak battery fault identification.

[0005] In a first aspect, embodiments of this application provide a battery fault early warning method, including: Obtain the battery voltage sequence within a preset time window; For each raw voltage in the voltage sequence, perform the following steps: Empirical mode decomposition of the original voltage yields multiple intrinsic mode function components; The information entropy and spectral entropy of each intrinsic mode function component are mapped onto a two-dimensional entropy domain plane to obtain a data point cloud. The two-dimensional entropy domain plane is determined based on multiple historical voltage sequences of the battery sample. The two-dimensional entropy domain plane includes a first region and a second region, and the noise in the first region is less than the noise in the second region. Filter out the target data points located in the first region from the data point cloud; The reconstructed voltage is obtained by superimposing the intrinsic mode function components corresponding to the target data points; Based on the reconfigured voltage, a fault warning is issued for the battery.

[0006] According to any of the foregoing embodiments of this application, the first region includes a first sub-region and a second sub-region, wherein the noise of the second sub-region is greater than or equal to the noise of the first sub-region, and the noise of the second sub-region is less than the noise of the second region. Target data points located in the first region are selected from the point cloud, including: Filter out the first target data point located in the first sub-region from the data point cloud; The information entropy and spectral entropy of each data point in the second sub-region are weighted and summed to obtain the comprehensive score of the effective features of each data point in the second sub-region. Data points whose comprehensive score of effective features is greater than or equal to the first threshold are identified as the second target data points; The target data point is determined based on the first target data point and the second target data point.

[0007] According to any of the foregoing embodiments of this application, the reconstructed voltage is obtained by superimposing the intrinsic mode function components corresponding to the target data point, including: The amplitude of the intrinsic mode function components corresponding to the second target data point is corrected by using dynamic weighting coefficients to obtain optimized components; wherein, the dynamic weighting coefficients are determined based on the comprehensive score of effective features corresponding to the intrinsic mode function components, and the dynamic weighting coefficients are positively correlated with the comprehensive score of effective features; The reconstructed voltage is obtained by superimposing the intrinsic mode function components and the optimized components corresponding to the first target data point.

[0008] According to any of the foregoing embodiments of this application, before obtaining the voltage sequence of the battery within a preset time window, the method further includes: Empirical mode decomposition was performed on multiple historical voltage sequences of the battery samples to obtain multiple historical components; A two-dimensional entropy domain coordinate system is constructed with information entropy as the horizontal axis and spectral entropy as the vertical axis; Map the information entropy and spectral entropy corresponding to each historical component to a two-dimensional entropy domain coordinate system; Based on the distribution of information entropy and spectral entropy corresponding to each historical component in the two-dimensional entropy domain coordinate system, the regional range of the two-dimensional entropy domain plane is determined. The two-dimensional entropy domain plane is divided into regions based on the maximum information entropy and the maximum spectral entropy corresponding to the regional range of the two-dimensional entropy domain plane.

[0009] According to any of the foregoing embodiments of this application, a fault warning for the battery is provided based on the reconfiguration voltage, including: Feature extraction is performed on the reconstructed voltage to obtain a feature vector; Input the feature vector into the trained fault warning model and output the fault warning result.

[0010] According to any of the foregoing embodiments of this application, feature extraction of the reconstructed voltage includes: Multi-scale features are extracted from the reconstructed voltage to obtain multi-scale features; By performing feature enhancement on the multi-scale features, voltage enhancement features are obtained.

[0011] According to any of the foregoing embodiments of this application, the method further includes: Obtain the target parameter sequence of the battery within a preset time window. The target parameter sequence includes at least one of the current sequence, temperature sequence, and insulation resistance sequence. Based on the reconfiguration voltage, the battery provides fault warnings, including: The target parameter sequence and the reconstructed voltage are fused to obtain a fused feature vector; The fused feature vectors are input into the fault warning model, and the fault warning results are output.

[0012] Secondly, embodiments of this application also provide a battery fault warning device, comprising: The acquisition module is used to acquire the battery voltage sequence within a preset time window; The execution module performs the following steps for each raw voltage in the voltage sequence: Empirical mode decomposition of the original voltage yields multiple intrinsic mode function components; The information entropy and spectral entropy of each intrinsic mode function component are mapped onto a two-dimensional entropy domain plane to obtain a data point cloud. The two-dimensional entropy domain plane is determined based on multiple historical voltage sequences of the battery sample. The two-dimensional entropy domain plane includes a first region and a second region, and the noise in the first region is less than the noise in the second region. Filter out the target data points located in the first region from the data point cloud; The reconstructed voltage is obtained by superimposing the intrinsic mode function components corresponding to the target data points; The early warning module is used to provide early warning of battery faults based on the reconfigured voltage.

[0013] Thirdly, embodiments of this application also provide an electronic device, which includes: a processor and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement any of the above-described battery fault warning methods.

[0014] Fourthly, embodiments of this application also provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement any of the above-described battery fault warning methods.

[0015] Fourthly, embodiments of this application also provide a computer program product, wherein the instructions in the computer program product, when executed by the processor of an electronic device, enable the electronic device to execute any of the above-described battery fault warning methods.

[0016] The battery fault early warning method, device, equipment, medium, and product provided in this application pre-construct a two-dimensional entropy domain plane adapted to actual operating conditions based on the battery's historical voltage sequence. Information entropy and spectral entropy characterize the temporal disorder and frequency dispersion of the intrinsic mode function (IMF) components. Effective components are screened by distinguishing noise differences between two regions, replacing the traditional single noise reduction method, and accurately distinguishing effective fault features from interference noise. Based on this, the voltage signal is reconstructed by superimposing the screened effective IMF components. While filtering out invalid noise interference, the detailed features of early-stage weak battery faults are fully preserved, which helps improve the signal-to-noise ratio. Battery fault early warning is performed based on the optimized, high-quality reconstructed signal, solving the identification deviation problem caused by noise interference from the signal source, which helps improve the accuracy and reliability of identifying early-stage weak battery faults. Attached Figure Description

[0017] 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.

[0018] Figure 1 This is a schematic flowchart of a battery fault early warning method provided in one embodiment of this application; Figure 2 This is one of the schematic diagrams of the region division of a two-dimensional entropy domain plane provided in one embodiment of this application; Figure 3 This is the second schematic diagram of the region division of a two-dimensional entropy domain plane provided in one embodiment of this application; Figure 4 This is the third schematic diagram of the region division of a two-dimensional entropy domain plane provided in one embodiment of this application; Figure 5 This is the fourth schematic diagram of the region division of a two-dimensional entropy domain plane provided in one embodiment of this application; Figure 6 This is the fifth schematic diagram of the region division of a two-dimensional entropy domain plane provided in one embodiment of this application; Figure 7 This is a detailed flowchart of step S23 provided in one embodiment of this application; Figure 8 This is a detailed flowchart of S24 provided in one embodiment of this application; Figure 9 This is a detailed flowchart of S3 provided in one embodiment of this application; Figure 10 This is a detailed flowchart illustrating "feature extraction of reconstructed voltage" provided in one embodiment of this application; Figure 11 This is a flowchart illustrating a battery fault warning method provided in another embodiment of this application; Figure 12 This is a flowchart illustrating a battery fault warning method provided in another embodiment of this application; Figure 13 This is a schematic diagram of the structure of a battery fault warning device provided in one embodiment of this application. Figure 14 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0019] 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.

[0020] 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 said element.

[0021] To address the technical problems raised in the background art, embodiments of this application provide a battery fault early warning method, apparatus, device, medium, and product. The battery fault early warning method includes: acquiring a voltage sequence of a battery within a preset time window; for each original voltage in the voltage sequence, performing the following steps: performing empirical mode decomposition on the original voltage to obtain multiple intrinsic mode function components; mapping the information entropy and spectral entropy of each intrinsic mode function component onto a two-dimensional entropy domain plane to obtain a data point cloud; the two-dimensional entropy domain plane is determined based on multiple historical voltage sequences of battery samples, and includes a first region and a second region, where the noise in the first region is less than the noise in the second region; selecting target data points located in the first region from the data point cloud; superimposing the intrinsic mode function components corresponding to the target data points to obtain a reconstructed voltage; and providing a fault early warning for the battery based on the reconstructed voltage.

[0022] Based on this, this application pre-constructs a two-dimensional entropy domain plane adapted to actual operating conditions based on the battery's historical voltage sequence. Information entropy and spectral entropy characterize the temporal disorder and frequency dispersion of the intrinsic mode function components. Effective components are screened by distinguishing noise differences between two regions, replacing the traditional single noise reduction method, and accurately distinguishing effective fault features from interference noise. On this basis, the voltage signal is reconstructed by superimposing the screened effective intrinsic mode function components. While filtering out invalid noise interference, it fully preserves the detailed features of early-stage weak battery faults, which helps improve the signal-to-noise ratio. Battery fault early warning is then performed based on the optimized, high-quality reconstructed signal, solving the identification deviation problem caused by noise interference from the signal source, which helps improve the accuracy and reliability of identifying early-stage weak battery faults.

[0023] The battery fault early warning method provided in the embodiments of this application will be introduced first below.

[0024] Figure 1 A schematic flowchart of a battery fault warning method according to an embodiment of this application is shown. Figure 1 As shown, the battery fault warning method may include the following steps: S1~S3.

[0025] S1. Obtain the battery voltage sequence within a preset time window.

[0026] The battery may include cells, or a battery pack composed of multiple cells.

[0027] The preset time window length can be flexibly set according to actual needs. As an example, the preset time window length is 1 minute to 30 minutes.

[0028] In one example, the length of the preset time window is related to the sampling frequency. Specifically, a shorter time window can be used when the sampling frequency per unit time is higher, and a longer time window can be used when the sampling frequency per unit time is lower.

[0029] This application does not limit the voltage sequence acquisition device, and may include all types of voltage sampling devices known to those skilled in the art, such as voltage sensors or voltage divider circuits.

[0030] In this step, the battery in operation is monitored online. The battery's operating voltage is collected in real time using a voltage sampling device. After collecting the voltage for a preset time window, a voltage sequence is obtained. After collecting one voltage sequence, the next voltage sequence can be collected immediately, or the collection can be started after a set interval.

[0031] S2. Perform the following steps for each original voltage in the voltage sequence: S21~S24.

[0032] In this step, Empirical Mode Decomposition (EMD) is performed on each original voltage in the voltage sequence to obtain multiple intrinsic mode function (EMF) components. Then, the information entropy and spectral entropy of each EMF component are calculated; one EMF component corresponds to a set of information entropy and spectral entropy. The information entropy and spectral entropy corresponding to each EMF component are mapped onto a two-dimensional entropy domain plane. A set of information entropy and spectral entropy corresponds to one data point, resulting in a data point cloud. Based on the distribution of data points on the two-dimensional entropy domain plane, data points located in the first region are retained and used as target data points. The EMF components corresponding to the target data points are superimposed to obtain the reconstructed voltage. The reconstructed voltage is obtained by superimposing the EMF components (the EMF components corresponding to the target data points) obtained from the decomposition of the same original voltage.

[0033] S21. Perform empirical mode decomposition on the original voltage to obtain multiple intrinsic mode function components.

[0034] This step employs all methods known to those skilled in the art to perform empirical mode decomposition on the original voltage, and is not limited thereto.

[0035] As an example, an adaptive multi-scale decomposition of the original voltage is performed using empirical mode decomposition, and the specific process may include...

[0036] (1) Identify voltage sequences All local maxima and local minima.

[0037] (2) Use cubic spline difference to fit the upper envelope and the lower envelope respectively.

[0038] (3) Calculate the mean of the upper envelope and the lower envelope. The margin signal is obtained by subtracting the mean value from the original voltage. .

[0039] (4) Judgment Whether the intrinsic mode function components are satisfied: the number of extreme points and the number of zero crossings do not differ by more than 1, and the average value of the upper and lower envelopes at any point is 0.

[0040] (5) If the above conditions are met, then It is separated as an eigenmode function component.

[0041] (6) If the above conditions are not met, then Repeat steps (1) to (4) above as the signal to be decomposed until the remaining signal is a trend of point loss or the number of extreme points is insufficient, and then terminate the decomposition.

[0042] (7) The final original voltage can be expressed as: (1) Wherein, IMF1~IMFn are the intrinsic mode function components of each order, and r(t) is the final participating trend term.

[0043] S22. Map the information entropy and spectral entropy of each intrinsic mode function component onto the two-dimensional entropy domain plane to obtain the data point cloud.

[0044] Before performing this step, all methods known to those skilled in the art can be used to calculate the information entropy and spectral entropy corresponding to the intrinsic mode function components, and there are no limitations here.

[0045] Information entropy is used to characterize the degree of disorder of intrinsic mode function components in the time domain, reflecting the randomness of signal temporal fluctuations. This application uses voltage sequences as the analysis object to quantify the uncertainty of signal changes. Pure effective signals have strong temporal variation patterns, low disorder, and relatively small information entropy; after noise is added, the random fluctuations of the signal intensify, the disorder increases, and the information entropy increases accordingly.

[0046] Spectral entropy is used to characterize the dispersion of the energy distribution of intrinsic mode function components in the frequency domain, reflecting the clustering state of frequency domain features. The energy corresponding to effective features will be concentrated in a specific frequency band, resulting in a smaller spectral entropy; noise energy is evenly distributed throughout the entire frequency domain, which will increase the value of spectral entropy.

[0047] As an example, let's take the i-th intrinsic mode component... For example, it can be written as The amplitude range of the intrinsic mode component is divided into N equally spaced intervals. The number of sampling points in each interval is counted, and the probability of that interval is obtained by dividing the number of points in that interval by the total number of points. The information entropy of the intrinsic mode function component can be calculated using formula (2). .

[0048] (2) As an example, let's take the i-th intrinsic mode component... For example, performing a Fast Fourier Transform on this component yields the spectrum. Calculate the power spectrum and normalize it to obtain the energy probability distribution at each frequency point. The spectral entropy of the intrinsic mode function component can be calculated using formula (3). .

[0049] (3) The two-dimensional entropy domain plane is determined based on multiple historical voltage sequences of battery samples. The two-dimensional entropy domain plane includes a first region and a second region, with the noise in the first region being less than that in the second region.

[0050] As an example, the horizontal axis of the two-dimensional entropy domain is information entropy, and the vertical axis is spectral entropy.

[0051] As an example, such as Figure 2 As shown, the two-dimensional entropy domain plane is rectangular, with the vertical axis containing (0, 1) and (2, 3) points. The maximum spectral entropy H is also shown. f,max ) and the maximum information entropy H on the horizontal axis x,max A straight line connecting the first and second regions (0, 0) forms the boundary between them. When the information entropy is a fixed value, the spectral entropy of data points in the first region is less than that in the second region; similarly, when the spectral entropy is a fixed value, the information entropy of data points in the first region is less than that in the second region; therefore, the noise in the first region is less than that in the second region. The first region is closer to the origin O of the two-dimensional entropy domain plane, and therefore has lower noise; the second region is farther from the origin O of the two-dimensional entropy domain plane, and therefore has higher noise.

[0052] In this step, the information entropy and spectral entropy of an intrinsic mode function component are grouped together. The information entropy and spectral entropy of each intrinsic mode function component are mapped onto a two-dimensional entropy domain plane to obtain data points equal in number to the intrinsic mode function components, thus forming a data point cloud.

[0053] S23. Filter out the target data points located in the first region from the data point cloud.

[0054] In this step, the data points are filtered according to their positions in the two-dimensional entropy domain. Data points located in the first region are retained, while data points located in the second region are removed. That is, the intrinsic mode function components containing effective fault characteristics are retained, while the intrinsic mode function components with high noise are removed, thus accurately distinguishing effective fault characteristics from interference noise.

[0055] S24. The intrinsic mode function components corresponding to the target data point are superimposed to obtain the reconstructed voltage.

[0056] In this step, the effective intrinsic mode function components (i.e., the intrinsic function components corresponding to the target data point) obtained by decomposing the same original voltage are superimposed to obtain the reconstructed voltage. Because the above steps filter out interference noise in the original voltage and fully preserve the subtle early fault details of the battery, the reconstructed voltage has a higher signal-to-noise ratio compared to the original voltage.

[0057] This application obtains a reconstructed voltage sequence consisting of multiple reconstructed voltages by performing empirical mode decomposition, mapping screening, and reconstruction on each original voltage in the voltage sequence.

[0058] S3. Provide a fault warning for the battery based on the reconfiguration voltage.

[0059] In this step, the reconstructed voltage has a high signal-to-noise ratio, which is used as the basis for battery fault early warning. This addresses the identification deviation caused by noise interference at the signal source, and helps improve the accuracy and reliability of identifying early and minor battery faults.

[0060] This application does not limit the fault warning method, and all methods known to those skilled in the art can be used, such as feature threshold determination method, trend analysis method, similarity matching or learning model, etc., without limitation.

[0061] As an example, time-domain or frequency-domain features, such as amplitude, fluctuation, variance, peak factor, and frequency features, are extracted from the reconstructed voltage. The extracted features are compared with a set feature threshold, and the battery status is determined based on the comparison results to achieve fault early warning.

[0062] As an example, time-series trend analysis is performed on the reconstructed voltage to monitor voltage fluctuation trends, gradual offsets, and jump inflection points, identify abnormal voltage trends, and capture slow degradation of electromagnetic performance and early gradual faults.

[0063] As an example, the voltage waveform of the battery under normal operating conditions or typical fault conditions is used as a template. The battery state is determined based on the similarity between the reconstructed voltage and the template.

[0064] As an example, the fault warning model trained on the voltage input will be reconstructed, and the warning result will be output. The fault warning model can adopt any learning model known to those skilled in the art, such as the Extreme Gradient Boosting (XGBoost) classification model, and is not limited here.

[0065] The battery fault early warning method provided in this application pre-constructs a two-dimensional entropy domain plane adapted to actual operating conditions based on the battery's historical voltage sequence. Information entropy and spectral entropy characterize the temporal disorder and frequency dispersion of the intrinsic mode function components. Effective components are screened by distinguishing noise differences between two regions, replacing the traditional single noise reduction method, and accurately distinguishing effective fault features from interference noise. Based on this, the voltage signal is reconstructed by superimposing the screened effective intrinsic mode function components. While filtering out invalid noise interference, the detailed features of early-stage weak battery faults are fully preserved, which helps improve the signal-to-noise ratio. Battery fault early warning is performed based on the optimized, high-quality reconstructed signal, solving the identification deviation problem caused by noise interference from the signal source, which helps improve the accuracy and reliability of identifying early-stage weak battery faults.

[0066] In one embodiment, the first region includes a first sub-region and a second sub-region, wherein the noise of the second sub-region is greater than or equal to the noise of the first sub-region, and the noise of the second sub-region is less than the noise of the second region.

[0067] In this embodiment, information entropy is positively correlated with noise; that is, the higher the information entropy, the greater the corresponding noise. Spectral entropy is also positively correlated with noise; that is, the higher the spectral entropy, the greater the corresponding noise.

[0068] Based on this, the first region is divided into two sub-regions: the regions corresponding to the larger information entropy intervals and the regions corresponding to the larger spectral entropy intervals within the first region are divided into the second sub-regions, and the remaining regions within the first region are divided into the first sub-regions. Since the noise in the first sub-region is less than that in the second sub-region, all data points located in the first sub-region are retained. Data points located in the second sub-region are then selected based on preset criteria.

[0069] As an example, such as Figure 3 As shown, the information entropy is equal to the first set information entropy H. x1 The vertical line and the spectral entropy are equal to the first set spectral entropy H. f1 The horizontal line serves as a dividing line, dividing the first region into a first sub-region and a second sub-region. Among these, the sub-region is defined as one whose information entropy is less than a first predetermined information entropy H. x1 And the spectral entropy is less than the first set spectral entropy H f1The region that meets the above conditions is designated as the first sub-region, and the first sub-region is rectangular. The first region that does not meet the above conditions is designated as the second sub-region, i.e., the information entropy in the first region is greater than a first set information entropy H. x1 The region, or the region with a spectral entropy greater than the first set spectral entropy H f1 The region is designated as the second sub-region.

[0070] As an example, such as Figure 4 As shown, the maximum spectral entropy H connected to the vertical axis is... f,max The maximum information entropy H on the horizontal axis x,max The concave arc between the two regions serves as a dividing line, dividing the first region into a first sub-region and a second sub-region. The first sub-region is closer to the origin O of the two-dimensional entropy domain plane, while the second sub-region is farther from the origin O of the two-dimensional entropy domain plane, and the second sub-region lies between the first and second sub-regions.

[0071] As an example, such as Figure 5 As shown, with the origin O of the two-dimensional entropy domain plane as the center, and the second set information entropy H... x2 Or the second set spectral entropy H f2 Draw a sector with a radius of 1, define the sector area as the first sub-region, and define the first sub-region with the sector as the radius of 1 as the second sub-region. Secondly, set the information entropy H. x2 The maximum information entropy H is smaller than that of the two-dimensional entropy field plane. x,max The second setting is the spectral entropy H. f2 The maximum spectral entropy H smaller than that of the two-dimensional entropy domain plane f,max .

[0072] The second setting is information entropy H. x1 The maximum information entropy H is equal to K times. x,max The range of values ​​for K is: .

[0073] As an example, such as Figure 6 As shown, the first region is divided into a first sub-region and a second sub-region by using a diagonal line parallel to the diagonal of the two-dimensional entropy domain plane as the dividing line. The intersection of this diagonal line with the horizontal axis is (H... x3 The intersection point with the vertical axis is (H, 0), 0). f3 ,0). Among them, the area below the diagonal line is the first sub-region, and the area above the diagonal line is the second sub-region.

[0074] This embodiment is not limited to the method of dividing the first sub-region and the second sub-region. Figure 3 , Figure 4 , Figure 5 and Figure 6 The methods shown may include all methods known to those skilled in the art, and are not limited herein.

[0075] Figure 1 S23 may include the following steps: S231~S234, such as Figure 7 As shown.

[0076] S231. Select the first target data point located in the first sub-region from the data point cloud.

[0077] In this step, among the multiple regions in the two-dimensional entropy domain plane, the first sub-region has the least noise. All data points located in the first sub-region are retained and determined as the first target data points.

[0078] S232. The information entropy and spectral entropy of each data point in the second sub-region are weighted and summed to obtain the comprehensive score of the effective features of each data point in the second sub-region.

[0079] S233. Data points whose comprehensive score of effective features is greater than or equal to the first threshold are identified as the second target data points.

[0080] In multiple regions within the two-dimensional entropy domain plane, the second sub-region has relatively low noise, but some of its intervals (such as regions with high information entropy and regions with high spectral entropy) have relatively high noise. If all data points located in the second sub-region are retained, some large interference noise will be retained, thereby reducing the signal-to-noise ratio of the reconstructed voltage.

[0081] In this step, the comprehensive score of effective features is obtained by calculating the weighted sum of the information entropy and spectral entropy of each data point in the second sub-region. A higher comprehensive score indicates a higher proportion of effective features such as battery voltage fluctuations and weak distortions from early faults in the intrinsic mode function component, and a lower proportion of noise. Based on the comparison between the comprehensive score of effective features and the first threshold, data points located in the second sub-region are filtered, retaining those with a comprehensive score of effective features greater than or equal to the first threshold, and removing those with a comprehensive score of effective features less than the first threshold.

[0082] As an example, the comprehensive score of effective features is calculated using formula (4).

[0083] (4) in, For effective feature comprehensive scoring, For information entropy, For spectral entropy, and As the weights, their sum equals 1. Preferably, , It balances the characteristics of timing fault distortion and frequency domain energy characteristics.

[0084] As an example, before performing a weighted summation of the information entropy and spectral entropy of each data point in the second sub-region, the information entropy and spectral entropy of all intrinsic mode function components are normalized so that the magnitudes of the information entropy and spectral entropy are distributed in [0,1].

[0085] S234. Determine the target data point based on the first target data point and the second target data point.

[0086] In this step, the first target data point retained in the first sub-region and the second target data point retained in the second sub-region are merged to form the target data point.

[0087] The battery fault early warning method provided in this application divides the first region into two sub-regions. All data points in the low-noise first sub-region are retained, achieving rapid retention of effective components. For the second sub-region with relatively high noise levels and mixed features, a comprehensive score of effective features is obtained by weighted summation of information entropy and spectral entropy. Combined with a second threshold judgment, components that still have value are selected. This hierarchical screening mode takes into account both screening efficiency and recognition accuracy. It not only fully retains the fault features of the low-noise region, but also fully explores the effective information in the mixed region and further filters out interference components. This can further improve the signal-to-noise ratio, reduce the bias caused by single-partition screening, and help improve the accuracy and reliability of fault early warning results.

[0088] In one embodiment, Figure 1 S24 may include the following steps: S241~S242, such as Figure 8 As shown.

[0089] S241. The amplitude of the intrinsic mode function components corresponding to the second target data point is corrected by using dynamic weighting coefficients to obtain the optimized components.

[0090] The dynamic weight coefficient is determined based on the comprehensive score of effective features corresponding to the intrinsic mode function components, and the dynamic weight coefficient is positively correlated with the comprehensive score of effective features.

[0091] A higher effective comprehensive score for the intrinsic mode function (IMF) components indicates a higher proportion of effective features and a lower proportion of noise. By assigning them a higher dynamic weighting coefficient, the optimized components have a larger amplitude, thus enhancing the influence of effective features on the reconstructed voltage. Conversely, a lower effective comprehensive score for the IMF components indicates a lower proportion of effective features and a higher proportion of noise. By assigning them a lower dynamic weighting coefficient, the optimized components have a smaller amplitude, thus reducing the influence of noise on the reconstructed voltage.

[0092] S242. The intrinsic mode function components and optimized components corresponding to the first target data point are superimposed to obtain the reconstructed voltage.

[0093] As an example, the reconstructed voltage can be calculated using formula (5).

[0094] (5) in, To reconstruct the voltage, The intrinsic mode function components corresponding to the first target data point. The intrinsic mode function components corresponding to the second target data point. For dynamic weighting coefficients, To optimize the components.

[0095] As an example, dynamic weighting coefficients Comprehensive score with effective features It satisfies the power function mapping relationship. Effective feature comprehensive score. The larger the value, the higher the dynamic weighting coefficient. The larger the value, the higher the retained amplitude; effective feature comprehensive score The smaller the value, the higher the dynamic weighting coefficient. The smaller the value, the lower the retained amplitude.

[0096] As an example, the dynamic weighting coefficients are calculated using formula (6).

[0097] (6) in, To smooth the adjustment coefficient, its value ranges from [0.8, 1.2]. It can be adaptively fine-tuned according to battery conditions to avoid distortion of the reconstructed signal caused by sudden weight changes. When the effective feature comprehensive score... When it approaches 1, the dynamic weighting coefficient Approaching 1, preserving weak fault characteristics to the greatest extent; when the effective characteristic comprehensive score When it approaches 0, the dynamic weighting coefficient Approaching 0, it significantly suppresses random noise interference.

[0098] It should be noted that this example only illustrates the comprehensive score of effective features. With dynamic weight coefficients The relationship is non-linear, but this does not constitute a limitation on the battery fault early warning method provided in the embodiments of this application. In other embodiments, a comprehensive scoring of effective features can also be set. With dynamic weight coefficients The relationship is linear, but this is not a limitation.

[0099] The battery fault early warning method provided in this application introduces a dynamic weighting coefficient to correct the amplitude of the intrinsic mode function component corresponding to the second sub-region. It can adaptively adjust the signal amplitude according to the effective feature ratio of the component, weakening the interference effect of noise. The uncorrected low-noise component and the optimized component are superimposed to complete the signal reconstruction. This not only preserves the fault details in the pure component, but also effectively suppresses the noise components in the mixed component, further improving the signal-to-noise ratio of the reconstructed voltage. This makes the signal characteristics more consistent with the actual operating state of the battery, reducing the impact of noise interference on the subsequent identification process from the source, and effectively improving the accuracy and reliability of fault early warning.

[0100] In one embodiment, prior to S1, the method may further include the following steps.

[0101] Empirical mode decomposition was performed on multiple historical voltage sequences of the battery samples to obtain multiple historical components; A two-dimensional entropy domain coordinate system is constructed with information entropy as the horizontal axis and spectral entropy as the vertical axis; Map the information entropy and spectral entropy corresponding to each historical component to a two-dimensional entropy domain coordinate system; Based on the distribution of information entropy and spectral entropy corresponding to each historical component in the two-dimensional entropy domain coordinate system, the regional range of the two-dimensional entropy domain plane is determined. The two-dimensional entropy domain plane is divided into regions based on the maximum information entropy and the maximum spectral entropy corresponding to the regional range of the two-dimensional entropy domain plane.

[0102] In this embodiment, before monitoring battery fault warning, historical voltage sequences of battery samples or the battery to be monitored under normal operating conditions and before a fault occurs are obtained. Empirical mode decomposition (EMD) is performed on these historical voltage sequences to obtain multiple historical components, and the information entropy and spectral entropy of each historical component are calculated. The EMD of the historical voltage sequences and the calculation of the dual entropy are the same as in the previous embodiments and will not be repeated here.

[0103] Meanwhile, a two-dimensional entropy domain coordinate system of information entropy-spectral entropy is established to evaluate from two orthogonal dimensions: "time domain uncertainty" and "frequency domain uniformity".

[0104] The information entropy and spectral entropy of each historical component are mapped to a two-dimensional entropy domain coordinate system to obtain historical data point clouds. Based on the distribution of historical data point clouds in the two-dimensional entropy domain coordinate system, the region enclosed by the vertical line X (maximum information entropy), the horizontal line Y (maximum spectral entropy), and the two coordinate axes of the coordinate system is used to determine the region of the two-dimensional entropy domain plane system.

[0105] After determining the region of the two-dimensional entropy domain plane, it can be done as follows: Figure 2 As shown, the two-dimensional entropy domain plane can be divided into a first region and a second region, or as follows: Figures 3-6As shown in any figure, the two-dimensional entropy domain plane is divided into a first sub-region, a second sub-region, and a second region.

[0106] In one embodiment, Figure 1 S3 may include the following steps: S31~S32, such as Figure 9 As shown.

[0107] S31. Extract features from the reconstructed voltage to obtain the feature vector.

[0108] This step extracts features from the reconstructed voltage with a high signal-to-noise ratio, extracts key information that can characterize the battery's operating state, removes redundant data, and forms a feature vector with strong feature orientation.

[0109] S32. Input the feature vector into the trained fault warning model and output the fault warning result.

[0110] This step involves inputting the feature vectors into the trained fault warning model for analysis and identification, which can accurately capture the early fault characteristics hidden in the voltage signal.

[0111] The battery fault early warning method provided in this application obtains an optimized reconstructed voltage by performing pre-noise reduction and hierarchical screening on the original voltage. Then, feature extraction and model recognition are performed on the reconstructed voltage to effectively reduce the interference of invalid information on the judgment process and improve the sensitivity and accuracy of fault identification.

[0112] In one embodiment, "feature extraction of the reconstructed voltage" may include the following steps: S311~S312, such as Figure 10 As shown.

[0113] S311. Multi-scale feature extraction is performed on the reconstructed voltage to obtain multi-scale features.

[0114] In this step, multi-scale parallel convolution can be used to extract features from the reconstructed voltage, extracting both local and global features.

[0115] As an example, a parallel feature extraction network composed of small-scale, medium-scale, and large-scale convolutional kernels is constructed to simultaneously extract local detail features, mid-frequency fluctuation features, and global trend features of the reconstructed voltage, thereby capturing fault features from multiple dimensions and reducing the loss of weak fault information by single-scale features.

[0116] S312. Perform feature enhancement on the multi-scale features to obtain voltage enhancement features.

[0117] As an example, adding a residual connection module to the feature extraction network allows the network to retain the original effective feature information when performing deep feature learning, avoiding gradient vanishing and feature degradation, and improving the stability and recognizability of deep fault features.

[0118] The battery fault early warning method provided in this application performs multi-scale feature extraction on the reconstructed voltage, mining the state information of the signal at different scales to fully capture the subtle feature differences between normal battery operation and early faults, avoiding feature omissions caused by single-scale extraction. Furthermore, the multi-scale features are enhanced to further highlight effective fault features, suppress residual weak interference, and improve feature discriminative power. The voltage enhancement feature information obtained after the above processing is richer and more recognizable, providing high-quality input for subsequent model recognition and improving the sensitivity and accuracy of fault early warning.

[0119] In one embodiment, such as Figure 11 As shown, the method may further include the following step: S4.

[0120] S4. Obtain the target parameter sequence of the battery within a preset time window.

[0121] The target parameter sequence includes at least one of the following: current sequence, temperature sequence, and insulation resistance sequence.

[0122] This step can be performed simultaneously with S1, that is, simultaneously acquiring the voltage sequence and target parameter sequence of the battery within a preset time window to achieve time alignment of different data sequences.

[0123] S3 may include the following steps: S33~S34.

[0124] S33. Perform feature fusion on the target parameter sequence and the reconstructed voltage to obtain the fused feature vector.

[0125] As an example, before feature fusion of the target parameter sequence and the reconstructed voltage, features of current, temperature, and insulation resistance are extracted separately. Then, the extracted features (or enhanced features) of the reconstructed voltage are concatenated and fused with the extracted features of current, temperature, and insulation resistance to form a fused feature vector that includes electrical, thermal, and insulation characteristics. This comprehensively reflects the battery's operating state from multiple dimensions, compensating for the lack of information from single-type signals.

[0126] S34. Input the fused feature vector into the fault warning model and output the fault warning result.

[0127] This step integrates feature vector input into the fault early warning model, which can fully leverage the synergistic effect of multi-dimensional information, enhance the model's ability to identify different fault types and early, subtle faults, and improve the comprehensiveness, accuracy, and robustness of fault early warning.

[0128] The fault warning model infers and identifies the battery status to determine whether a fault exists. It can locate early and minor faults, prevent the fault from escalating, and help reduce losses.

[0129] It should be noted that before performing mode decomposition on the voltage sequence and before feature fusion on the reconstructed voltage and target parameter sequences, all types of preprocessing known to those skilled in the art can be applied to the acquired raw signal. For example, anti-interference filtering can be performed on the raw signal to remove abnormal waveforms; missing value imputation and outlier detection can be performed on the processed signal to form a complete monitoring data sequence.

[0130] Before feature extraction, amplitude normalization is performed on various signals. For example, the Z-score normalization method is used to uniformly transform the amplitude distribution of various signals, so that the data scales of different physical quantities are consistent.

[0131] For example, such as Figure 12 As shown, the battery fault early warning method may further include the following steps: S1200, Start.

[0132] S1201: Collects the battery's voltage, temperature, current, and insulation resistance.

[0133] S1202. Preprocess the collected data.

[0134] Preprocessing may include data cleaning, filling in missing values, and removing outlier data.

[0135] S1203, voltage EMD decomposition.

[0136] This step performs EMD decomposition on each acquired voltage signal to obtain multiple IMF components.

[0137] S1204. Calculate the information entropy and spectral entropy of each IMF component.

[0138] S1205. Construct a two-dimensional entropy domain.

[0139] In this step, a two-dimensional entropy domain is constructed using information entropy as the X-axis and spectral entropy as the Y-axis. A three-dimensional evaluation is performed from two orthogonal dimensions: "time-domain uncertainty" and "frequency-domain uniformity," forming a multi-dimensional, more robust, and more interpretable evaluation framework. The diagonal is divided from the upper left to the lower right corner; the region closer to the origin is the first region (i.e., the low-noise region), and the region farther from the origin is the second region (i.e., the high-noise region). The two-dimensional entropy domain is determined based on the historical voltage sequence of the battery samples, and operations S1202~S1204 are performed on the historical voltage sequence.

[0140] It should be noted that this step is not performed when online fault warnings are issued based on real-time battery operating data.

[0141] S1206, Two-dimensional entropy domain mapping.

[0142] This step maps the information entropy and spectral entropy of the IMF component to a two-dimensional entropy domain. The information entropy and spectral entropy of an IMF component are grouped together, corresponding to a data point in the two-dimensional entropy domain.

[0143] S1207. Determine whether to retain.

[0144] This step filters IMF components based on the location of data points in the two-dimensional entropy domain. If a data point falls into the first region (low noise region), the IMF component is retained; otherwise, it is discarded.

[0145] S1208, Reconstruct voltage signal.

[0146] In this step, the voltage signal is reconstructed using the retained IMF components, thereby obtaining an effective signal with a high signal-to-noise ratio, i.e., the reconstructed voltage.

[0147] S1209, Training the model.

[0148] This step involves reconstructing the voltage signal and extracting and fusing features from temperature, current, and insulation resistance to obtain a fused feature vector. This fused feature vector is then used as the model input to train and optimize the model, resulting in a fault early warning model.

[0149] S1210, Fault Warning.

[0150] In this step, the model input includes the temperature, current, and insulation resistance collected in S1201, as well as the reconstructed voltage obtained after processing in S1202~S1204 and S1206~S1208. Amplitude normalization is performed on the above signals to ensure consistent data scales across different physical quantities. Feature extraction and fusion are performed on the reconstructed voltage signal, along with the temperature, current, and insulation resistance, to obtain a fused feature vector. This fused feature vector is then input into the trained fault warning model. The fault warning model is used to infer and identify the battery state, determine whether a fault exists in the battery, and output a warning result.

[0151] Based on the battery fault warning method provided in the above embodiments, this application also provides specific implementation methods of the battery fault warning device. Please refer to the following embodiments.

[0152] First see Figure 13 The battery fault warning device 1300 provided in this application embodiment includes: an acquisition module 1301, an execution module 1302, and a warning module 1303.

[0153] The acquisition module 1301 is used to acquire the voltage sequence of the battery within a preset time window.

[0154] The execution module 1302 is used to perform the following steps for each original voltage in the voltage sequence: perform empirical mode decomposition on the original voltage to obtain multiple intrinsic mode function components; map the information entropy and spectral entropy of each intrinsic mode function component to a two-dimensional entropy domain plane to obtain a data point cloud; the two-dimensional entropy domain plane is determined based on multiple historical voltage sequences of the battery sample, and the two-dimensional entropy domain plane includes a first region and a second region, the noise of the first region is less than the noise of the second region; select target data points located in the first region from the data point cloud; and superimpose the intrinsic mode function components corresponding to the target data points to obtain the reconstructed voltage.

[0155] The early warning module 1303 is used to provide early warning of battery faults based on the reconfiguration voltage.

[0156] The battery fault early warning device provided in this application pre-constructs a two-dimensional entropy domain plane adapted to actual operating conditions based on the battery's historical voltage sequence. Information entropy and spectral entropy characterize the temporal disorder and frequency dispersion of the intrinsic mode function components. Effective components are screened by distinguishing noise differences between two regions, replacing the traditional single noise reduction method, and accurately distinguishing effective fault features from interference noise. Based on this, the voltage signal is reconstructed by superimposing the screened effective intrinsic mode function components. While filtering out invalid noise interference, it fully preserves the detailed features of early-stage weak battery faults, which helps improve the signal-to-noise ratio. Battery fault early warning is performed based on the optimized, high-quality reconstructed signal, solving the identification deviation problem caused by noise interference from the signal source, which helps improve the accuracy and reliability of identifying early-stage weak battery faults.

[0157] In one embodiment, the first region includes a first sub-region and a second sub-region, wherein the noise of the second sub-region is greater than or equal to the noise of the first sub-region, and the noise of the second sub-region is less than the noise of the second region. The execution module is used to filter target data points located in the first region from the data point cloud, including: filtering the first target data points located in the first sub-region from the data point cloud; weighting and summing the information entropy and spectral entropy of each data point in the second sub-region to obtain the comprehensive score of effective features for each data point in the second sub-region; determining the data points whose comprehensive score of effective features is greater than or equal to a first threshold as the second target data points; and determining the target data points based on the first target data points and the second target data points.

[0158] In one embodiment, the execution module is used to superimpose the intrinsic mode function components corresponding to the target data point to obtain a reconstructed voltage, including: using dynamic weighting coefficients to correct the amplitude of the intrinsic mode function components corresponding to the second target data point to obtain an optimized component; wherein, the dynamic weighting coefficients are determined based on the comprehensive score of the effective features corresponding to the intrinsic mode function components, and the dynamic weighting coefficients are positively correlated with the comprehensive score of the effective features; and superimposing the intrinsic mode function components and the optimized component corresponding to the first target data point to obtain a reconstructed voltage.

[0159] In one embodiment, the device may further include a plane determination module. The plane determination module is used to: perform empirical mode decomposition on multiple historical voltage sequences of the battery sample to obtain multiple historical components; construct a two-dimensional entropy domain coordinate system with information entropy as the abscissa and spectral entropy as the ordinate; map the information entropy and spectral entropy corresponding to each historical component into the two-dimensional entropy domain coordinate system; determine the regional range of the two-dimensional entropy domain plane based on the distribution of the information entropy and spectral entropy corresponding to each historical component in the two-dimensional entropy domain coordinate system; and divide the two-dimensional entropy domain plane into regions based on the maximum information entropy and maximum spectral entropy values ​​corresponding to the regional range of the two-dimensional entropy domain plane.

[0160] In one embodiment, the early warning module is used to provide early warning of battery faults based on the reconstructed voltage, including: extracting features from the reconstructed voltage to obtain a feature vector; inputting the feature vector into a trained fault early warning model and outputting a fault early warning result.

[0161] In one embodiment, the early warning module is used to extract features from the reconstructed voltage, including: extracting multi-scale features from the reconstructed voltage to obtain multi-scale features; and enhancing the multi-scale features to obtain voltage enhancement features.

[0162] In one embodiment, the acquisition module is further configured to: acquire a target parameter sequence of the battery within a preset time window, wherein the target parameter sequence includes at least one of a current sequence, a temperature sequence, and an insulation resistance sequence. The early warning module is further configured to: perform feature fusion on the target parameter sequence and the reconstructed voltage to obtain a fused feature vector; input the fused feature vector into the fault early warning model, and output a fault early warning result.

[0163] Figure 14 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0164] An electronic device may include a processor 1401 and a memory 1402 storing computer program instructions.

[0165] Specifically, the processor 1401 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.

[0166] Memory 1402 may include mass storage for data or instructions. For example, and not limitingly, memory 1402 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 1402 may include removable or non-removable (or fixed) media. Where appropriate, memory 1402 may be internal or external to an electronic device. In a particular embodiment, memory 1402 is a non-volatile solid-state memory.

[0167] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, 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 the operations described with reference to the methods according to one aspect of this disclosure.

[0168] The processor 1401 reads and executes computer program instructions stored in the memory 1402 to implement any of the battery fault warning methods in the above embodiments.

[0169] In one example, the electronic device may also include a communication interface 1403 and a bus 1404. For example, Figure 14 As shown, the processor 1401, memory 1402, and communication interface 1403 are connected through bus 1404 and complete communication with each other.

[0170] The communication interface 1403 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0171] Bus 1404 includes hardware, software, or both, that couples components of an electronic 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), a Hypertext Transfer (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel 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 Bus (VESA Local Bus, VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 1404 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0172] Furthermore, in conjunction with the battery fault warning method 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 battery fault warning methods in the above embodiments.

[0173] This application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the battery fault warning methods described in the above embodiments.

[0174] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0175] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable-ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0176] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0177] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0178] The above description is merely a specific implementation 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 battery fault early warning method, characterized in that, include: Obtain the battery voltage sequence within a preset time window; For each original voltage in the voltage sequence, perform the following steps: Empirical mode decomposition is performed on the original voltage to obtain multiple intrinsic mode function components; The information entropy and spectral entropy of each intrinsic mode function component are mapped onto a two-dimensional entropy domain plane to obtain a data point cloud. The two-dimensional entropy domain plane is determined based on multiple historical voltage sequences of battery samples. The two-dimensional entropy domain plane includes a first region and a second region, and the noise in the first region is less than the noise in the second region. Target data points located in the first region are selected from the data point cloud; The reconstructed voltage is obtained by superimposing the intrinsic mode function components corresponding to the target data point; Based on the reconstructed voltage, a fault warning is issued for the battery.

2. The method according to claim 1, characterized in that, The first region includes a first sub-region and a second sub-region, wherein the noise of the second sub-region is greater than or equal to the noise of the first sub-region, and the noise of the second sub-region is less than the noise of the second region. The step of filtering target data points located in the first region from the data point cloud includes: Filter out the first target data point located in the first sub-region from the data point cloud; The information entropy and spectral entropy of each data point in the second sub-region are weighted and summed to obtain the comprehensive score of the effective features of each data point in the second sub-region. Data points whose comprehensive score of effective features is greater than or equal to the first threshold are identified as the second target data points; Based on the first target data point and the second target data point, the target data point is determined.

3. The method according to claim 2, characterized in that, The step of superimposing the intrinsic mode function components corresponding to the target data point to obtain the reconstructed voltage includes: The intrinsic mode function components corresponding to the second target data point are amplitude-corrected using dynamic weighting coefficients to obtain optimized components; wherein, the dynamic weighting coefficients are determined based on the comprehensive score of effective features corresponding to the intrinsic mode function components, and the dynamic weighting coefficients are positively correlated with the comprehensive score of effective features; The reconstructed voltage is obtained by superimposing the intrinsic mode function components corresponding to the first target data point and the optimized components.

4. The method according to claim 1, characterized in that, Before obtaining the battery voltage sequence within a preset time window, the method further includes: Empirical mode decomposition is performed on multiple historical voltage sequences of the battery sample to obtain multiple historical components; A two-dimensional entropy domain coordinate system is constructed with information entropy as the horizontal axis and spectral entropy as the vertical axis; Map the information entropy and spectral entropy corresponding to each of the historical components to the two-dimensional entropy domain coordinate system; Based on the distribution of the information entropy and spectral entropy corresponding to each historical component in the two-dimensional entropy domain coordinate system, the regional range of the two-dimensional entropy domain plane is determined. The two-dimensional entropy domain plane is divided into regions based on the maximum information entropy and the maximum spectral entropy corresponding to the regional range of the two-dimensional entropy domain plane.

5. The method according to any one of claims 1-4, characterized in that, The step of providing a fault warning for the battery based on the reconstructed voltage includes: Feature extraction is performed on the reconstructed voltage to obtain a feature vector; The feature vector is input into the trained fault warning model, and the fault warning result is output.

6. The method according to claim 5, characterized in that, Feature extraction of the reconstructed voltage includes: Multi-scale feature extraction is performed on the reconstructed voltage to obtain multi-scale features; The multi-scale features are enhanced to obtain voltage enhancement features.

7. The method according to claim 5, characterized in that, The method further includes: Obtain the target parameter sequence of the battery within a preset time window, the target parameter sequence including at least one of current sequence, temperature sequence and insulation resistance sequence; The step of providing a fault warning for the battery based on the reconstructed voltage includes: The target parameter sequence and the reconstructed voltage are fused to obtain a fused feature vector; The fused feature vector is input into the fault warning model, and the fault warning result is output.

8. A battery fault early warning device, characterized in that, include: The acquisition module is used to acquire the battery voltage sequence within a preset time window; The execution module is configured to perform the following steps for each raw voltage in the voltage sequence: Empirical mode decomposition is performed on the original voltage to obtain multiple intrinsic mode function components; The information entropy and spectral entropy of each intrinsic mode function component are mapped onto a two-dimensional entropy domain plane to obtain a data point cloud. The two-dimensional entropy domain plane is determined based on multiple historical voltage sequences of battery samples. The two-dimensional entropy domain plane includes a first region and a second region, and the noise in the first region is less than the noise in the second region. Target data points located in the first region are selected from the data point cloud; The reconstructed voltage is obtained by superimposing the intrinsic mode function components corresponding to the target data point; The early warning module is used to provide early warning of battery failure based on the reconstructed voltage.

9. 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 battery fault warning method as described in any one of claims 1-7.

10. 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 battery fault warning method as described in any one of claims 1-7.

11. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device is able to perform the battery fault warning method as described in any one of claims 1-7.

Images