Battery fault root cause tracing method based on incremental capacity curve and equivalent circuit

Abstract

The application discloses a battery fault root cause tracing method based on an incremental capacity curve and an equivalent circuit, and relates to the fields of energy saving, environmental protection and greenness. The method comprises the following steps: obtaining a plurality of effective complete charging segments; determining the incremental capacity sequence characteristics, the incremental capacity topographic characteristics, the equivalent circuit parameters and the equivalent circuit parameter relative deviation characteristics of each single battery of each effective complete charging segment, and inputting the characteristics into a hybrid attention network to obtain the abnormal score of each single battery and the fault category probability output by the hybrid attention network, and the abnormal score of each effective complete charging segment; determining the fault single battery, the abnormal charging segment and the primary fault root cause according to the abnormal score of each single battery and the fault category probability, and the abnormal score of each effective complete charging segment. The application can determine the fault single battery, the abnormal charging segment and the primary fault root cause.

Description

Technical Field

[0001] This invention relates to the fields of energy conservation, environmental protection, and green power technology. Specifically, it relates to a method for tracing the root causes of battery failures based on incremental capacity curves and equivalent circuits. Background Technology

[0002] Electrochemical energy storage power stations, with lithium-ion batteries at their core, are a typical example of methods or devices for directly converting chemical energy into electrical energy. They have been widely applied in power grid frequency regulation, peak shaving, and renewable energy consumption scenarios, and are a key component in building an energy-saving, environmentally friendly, and green power system. However, lithium-ion batteries experience problems such as capacity decay, increased ohmic resistance, intensified polarization, loss of active lithium, loss of active materials, deterioration of cell consistency, soft short circuits, and abnormalities in data acquisition sensors during long-term charging and discharging, temperature fluctuations, high-rate operation, and aging. These issues pose serious challenges to the full life-cycle management of batteries in fields such as electric vehicle energy storage device manufacturing and new energy vehicle manufacturing.

[0003] Existing battery management systems (BMS) typically issue alarms based on individual cell voltage, total current, temperature, state of charge (SOC), and state of health (SOH) thresholds. They can determine whether something is abnormal, but they cannot answer which type of root cause caused the abnormality, which individual cell it occurred in, or which voltage range of the charging curve it appeared in. Summary of the Invention

[0004] To address the aforementioned problems in the existing technology, the present invention aims to provide a battery fault root cause tracing method and related apparatus based on incremental capacity curves and equivalent circuits, which can identify faulty individual cells, abnormal charging segments, and primary fault root causes.

[0005] To solve the above problems, the technical solution adopted by the present invention is as follows: The first aspect of this invention provides a method for tracing the root causes of battery failures based on incremental capacity curves and equivalent circuits, the method comprising: Acquire multiple valid and complete charging segments; Determine the incremental capacity sequence characteristics and incremental capacity morphology characteristics of each individual cell in each effective complete charging segment, and determine the equivalent circuit parameters and relative deviation characteristics of the equivalent circuit parameters of each individual cell in each effective complete charging segment; The incremental capacity sequence features, the incremental capacity morphology features, the equivalent circuit parameters, and the relative deviation features of the equivalent circuit parameters are input into a hybrid attention network to obtain the anomaly score and fault category probability of each individual cell and the anomaly score of each effective complete charging segment output by the hybrid attention network. The faulty individual cell, the abnormal charging segment, and the primary root cause of failure are determined based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment.

[0006] Optionally, determining the faulty individual cell, the abnormal charging segment, and the primary root cause of the fault based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment includes: Faulty individual cells are identified based on the anomaly score of each individual cell, and abnormal charging segments are identified based on the anomaly score of each valid complete charging segment. The strength of evidence for each fault category is calculated based on the incremental capacity morphology characteristics of the faulty individual cells and the weight corresponding to each fault category. The root cause score for each fault category is calculated based on the anomaly score of the faulty individual battery, the anomaly score of the abnormal charging segment, the strength of evidence corresponding to each fault category, and the probability of each fault category. The primary root cause of the failure is determined based on the root cause score.

[0007] Optionally, the hybrid attention network includes an incremental capacity curve encoder, an equivalent circuit parameter encoder, a cross-modal attention module, a temporal attention module, a single-unit attention module, and an output interpretation module, wherein: The incremental capacity curve encoder is used to encode the incremental capacity sequence features to obtain an incremental capacity feature vector; the equivalent circuit parameter encoder is used to encode the equivalent circuit parameters and the relative deviation features of the equivalent circuit parameters to obtain an equivalent circuit feature vector; the cross-modal attention module is used to calculate cross-modal attention based on the incremental capacity feature vector and the equivalent circuit feature vector to obtain a cross-domain feature vector. The time attention module is used to calculate the anomaly score of each valid complete charging segment based on the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; the individual cell attention module is used to calculate the anomaly score of each individual cell based on the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; the output interpretation module is used to calculate a global fusion vector based on the anomaly score of each individual cell, the anomaly score of each valid complete charging segment, the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; and to determine the fault category probability of each individual cell based on the global fusion vector.

[0008] Optionally, obtaining multiple valid complete charging segments includes: Select multiple consecutive charging segments; For each of the continuous charging segments, the state of charge increment of the continuous charging segment is determined based on the difference between the start value and the end value of the state of charge of the continuous charging segment. If the increment of the state of charge is greater than a preset state threshold, then the continuous charging segment is determined to be a valid and complete charging segment.

[0009] Optionally, determining the incremental capacity sequence characteristics and incremental capacity morphology characteristics of each individual cell in each of the effective complete charging segments includes: Determine the incremental capacity of each individual cell in each of the effective full charging segments at each sampling point; The incremental capacity curve for each individual cell in each effective full charge segment is determined based on the incremental capacity of each cell at all sampling points. The incremental capacity curve of each cell in each of the effective full charge segments is determined as the incremental capacity sequence feature of each cell in each of the effective full charge segments; and the incremental capacity curve of each cell in each of the effective full charge segments is interpolated to a uniform cell voltage grid. A search is performed on the unified single-cell voltage grid to obtain the main peak characteristics of each single cell; The characteristic variation of the main peak feature of each individual cell is determined based on the median value of the main peak feature; The characteristic variation of the main peak feature of each individual cell is determined as the incremental capacity morphology feature of each individual cell in each effective full charge segment.

[0010] Optionally, determining the incremental capacity of each individual cell in each of the effective full charging segments at each sampling point includes: The score of each cell in each effective complete charging segment at each sampling point is determined based on the voltage, mean voltage, and standard deviation of voltage of each cell at each sampling point. If the absolute value of the score is greater than a preset score threshold, then the sampling point is determined to be an abnormal voltage point. The voltage of the abnormal voltage point is repaired based on the voltage of the adjacent sampling points before and after the abnormal voltage point. The voltages at the repaired abnormal voltage points are smoothed to obtain a smoothed voltage sequence. The capacity change is determined based on the state of charge of the two sampling points before and after each sampling point, and the voltage change is determined based on the voltage of the two sampling points before and after each sampling point. The ratio of the capacity change to the voltage change is determined as the incremental capacity of each individual cell in each effective full charge segment at each sampling point.

[0011] Optionally, determining the equivalent circuit parameters and relative deviation characteristics of the equivalent circuit parameters for each individual cell in each of the effective complete charging segments includes: Within the same effective complete charging segment, the current and terminal voltage of each individual battery cell are collected to establish a second-order resistive-capacitive equivalent circuit model. The recursive least squares method with a forgetting factor is used for parameter identification to obtain the model parameters of the second-order resistive-capacitive equivalent circuit model corresponding to each individual cell. The open-circuit voltage, ohmic internal resistance, short-time polarization resistance, long-time diffusion polarization resistance, short-time polarization capacitance, and long-time diffusion polarization capacitance of each individual cell are calculated based on the model parameters; and the open-circuit voltage, ohmic internal resistance, short-time polarization resistance, long-time diffusion polarization resistance, short-time polarization capacitance, and long-time diffusion polarization capacitance of each individual cell are determined as the equivalent circuit parameters of each individual cell in each effective complete charging segment. Determine the median value of the ohmic internal resistance, the median value of the short-time polarization resistance, and the median value of the long-time diffusion polarization resistance in all individual cells. The relative deviation of the ohmic internal resistance of each individual cell is calculated based on the ohmic internal resistance of each individual cell and the median value of the ohmic internal resistance. The relative deviation of the short-time polarization resistance of each individual cell is calculated based on the short-time polarization resistance of each individual cell and the median value of the short-time polarization resistance. The relative deviation of the long-time diffusion polarization resistance of each individual cell is calculated based on the long-time diffusion polarization resistance of each individual cell and the median value of the long-time diffusion polarization resistance. The relative deviations of the ohmic internal resistance, short-time polarization resistance, and long-time diffusion polarization resistance of each individual cell are determined as the relative deviation characteristics of the equivalent circuit parameters of each individual cell in each effective complete charging segment.

[0012] A second aspect of this invention provides a battery fault root cause tracing device based on incremental capacity curves and equivalent circuits, the device comprising: The information acquisition unit is used to acquire multiple valid and complete charging segments; The feature determination unit is used to determine the incremental capacity sequence features and incremental capacity morphology features of each individual cell in each effective complete charging segment, and to determine the equivalent circuit parameters and the relative deviation features of the equivalent circuit parameters of each individual cell in each effective complete charging segment. The model running unit is used to input the incremental capacity sequence features, the incremental capacity morphology features, the equivalent circuit parameters, and the relative deviation features of the equivalent circuit parameters into the hybrid attention network to obtain the anomaly score and fault category probability of each individual battery and the anomaly score of each effective complete charging segment output by the hybrid attention network. The root cause tracing unit is used to determine the faulty individual cell, the abnormal charging segment, and the primary root cause of the fault based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment.

[0013] A third aspect of the present invention provides an electronic device, including: a processor and a memory; The processor is connected to a memory, wherein the memory is used to store computer programs and the processor is used to invoke the computer programs to execute the method as described in the first aspect of the embodiments of the present invention.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, perform the method as described in the first aspect of the present invention.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: First, this invention uses incremental capacity sequence features, incremental capacity morphology features, equivalent circuit parameters, and relative deviation features of equivalent circuit parameters as input data. The incremental capacity sequence features and incremental capacity morphology features are extracted from the incremental capacity curve. The incremental capacity curve can amplify minute changes near the charging voltage plateau, enabling the detection of capacity decay, polarization anomalies, and individual cell inconsistencies before the traditional voltage threshold is triggered, demonstrating strong early fault identification capabilities. The equivalent circuit parameters and relative deviation features have clear physical meanings, allowing the output of the hybrid attention network to be beyond the limitations of black-box classification results.

[0016] Secondly, the hybrid attention network can output the anomaly score and fault category probability of each individual battery cell, as well as the anomaly score of each valid complete charging segment. Thus, the faulty individual battery cell can be identified based on the anomaly score of each individual battery cell, the abnormal charging segment can be identified based on the anomaly score of each valid complete charging segment, and the primary root cause of the fault can be identified based on the fault category probability of each individual battery cell.

[0017] This battery failure root cause tracing mechanism can effectively solve the problems of difficulty in locating faulty individual cells, determining abnormal charging curve ranges, and analyzing root causes in the battery management systems of new energy vehicles, thereby improving battery safety in electric vehicles, motors, and electric drives under complex operating conditions. In addition to battery safety in the new energy vehicle field, this battery failure root cause tracing mechanism can also be applied to the system safety of methods or devices that directly convert chemical energy into electrical energy, comprehensively serving electric vehicles and energy-saving, environmentally friendly, and green power applications. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A schematic flowchart of a battery fault root cause tracing method based on incremental capacity curves and equivalent circuits provided in an embodiment of the present invention is shown. Figure 2 A schematic diagram of the structure of a second-order RC equivalent circuit model provided in an embodiment of the present invention is shown; Figure 3 This diagram illustrates the structure of a battery fault root cause tracing device based on incremental capacity curves and equivalent circuits according to an embodiment of the present invention. Figure 4 A schematic diagram of the structure of a computer device provided in one embodiment of the present invention is shown. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please refer to Figure 1 This document illustrates a flowchart of a battery fault root cause tracing method based on incremental capacity curves and equivalent circuits provided by an embodiment of the present invention. The multi-dimensional comprehensive safety assessment method for this energy storage power station includes the following steps: Step 101: Obtain multiple valid complete charging segments.

[0022] Specifically, acquiring multiple valid complete charging segments includes: Select multiple consecutive charging segments; For each of the continuous charging segments, the state of charge increment of the continuous charging segment is determined based on the difference between the start value and the end value of the state of charge of the continuous charging segment. If the increment of the state of charge is greater than a preset state threshold, then the continuous charging segment is determined to be a valid and complete charging segment.

[0023] For example, battery pack operating data can be collected through a BMS or cloud-based data platform. Each data entry includes at least the following fields: server_time, sys_status, soc, current, voltage, and temperature, where the voltage field is a sequence composed of multiple individual cell voltages.

[0024] Convert server_time to a time format and sort in ascending order. When sys_status=1, the battery is considered to be charging. Record the start point when transitioning from a non-charging state to a charging state and the end point when exiting a charging state. This generates multiple consecutive charging segments based on server_time and sys_status.

[0025] Furthermore, the number of sampling points for each continuous charging segment can be set to be greater than or equal to the sampling point threshold to avoid transient polarization interference in the initial stage of charging, suppress random noise, and ensure the statistical significance of state estimation.

[0026] For example, the increment of the state of charge is calculated as follows: , in, and These are the initial and final values ​​of the state of charge, respectively. The threshold for determining a complete charging segment is preferably set at 70%. This screening can reduce the distortion of the incremental capacity curve caused by short segments and unstable charging segments.

[0027] Step 102: Determine the incremental capacity sequence characteristics and incremental capacity morphology characteristics of each individual cell in each effective complete charging segment, and determine the equivalent circuit parameters and relative deviation characteristics of the equivalent circuit parameters of each individual cell in each effective complete charging segment.

[0028] The incremental capacity curve (IC curve) amplifies minute changes near the voltage plateau by calculating the derivative of capacity with respect to voltage. Its peak position, peak height, peak width, and peak area are closely related to the electrode reaction plateau, reversible capacity, polarization state, and aging mechanism. Compared to directly observing the voltage curve, the IC curve is more sensitive to early capacity decay, monomer inconsistency, and localized polarization anomalies. The incremental capacity sequence captures the dynamic evolution of IC curve characteristics over time, while the incremental capacity morphology describes the geometric and statistical properties of the IC curve.

[0029] Specifically, determining the incremental capacity sequence characteristics and incremental capacity morphology characteristics of each individual cell in each effective complete charging segment includes: Determine the incremental capacity of each individual cell in each of the effective full charging segments at each sampling point; The incremental capacity curve for each individual cell in each effective full charge segment is determined based on the incremental capacity of each cell at all sampling points. The incremental capacity curve of each cell in each of the effective full charge segments is determined as the incremental capacity sequence feature of each cell in each of the effective full charge segments; and the incremental capacity curve of each cell in each of the effective full charge segments is interpolated to a uniform cell voltage grid. A search is performed on the unified single-cell voltage grid to obtain the main peak characteristics of each single cell; The characteristic variation of the main peak feature of each individual cell is determined based on the median value of the main peak feature; The characteristic variation of the main peak feature of each individual cell is determined as the incremental capacity morphology feature of each individual cell in each effective full charge segment.

[0030] Among them, the incremental capacity morphology features may include, for example, the feature changes of the main peak features. The main peak features may include, for example, the main peak position, main peak height, main peak width, and main peak area. The feature changes of the main peak features may include peak position shift, peak height decrease ratio, peak width increase ratio, and peak area decrease ratio.

[0031] For example, taking the 7th monomer as an example, at a certain sampling point Nearby, the SOC and cell voltage recorded by the BMS are as follows: , , The relationship between capacity and SOC: , We can obtain: , , Therefore, the change in capacity is: , The voltage change is: , The incremental capacity value of the 7th monomer at sampling point k is: , If the small voltage difference protection term is considered, it is written as: , in, For the smallest positive number, the preferred choice is... This is used to prevent the calculation results from diverging due to an excessively small denominator.

[0032] If the rated capacity is not introduced, and only the SOC is used to calculate the normalized IC curve, then it is: , This normalization form and standard The curves differ only in the scaling factor. This does not affect the peak position, peak width, or relative differences between monomers.

[0033] For L sampling points, the above calculation is repeated to obtain the IC curve of the 7th monomer: .

[0034] If there are C individual cells, repeating the above calculation for each of the C individual cells will yield the set of IC curves for all individual cells: .

[0035] It can be used to represent the incremental capacity sequence characteristics of each individual cell in any complete charging segment.

[0036] The individual IC curves are interpolated to a unified voltage grid, where the voltage grid range is defined as follows: , After interpolation, the main peak position and peak height are obtained by searching for local maxima on a unified voltage grid. The peak width is calculated using the half-width at half-maximum (WHM) method, and the peak area is calculated using the trapezoidal integral within the voltage range where the peak is located. The main peak feature of the 7th individual cell is extracted on this voltage grid. Calculations are obtained from actual data: , , , , in, The location of the main peak. The height of the main peak The width of the main peak. The area of ​​the main peak.

[0037] Let the median value of the main peak characteristic of the remaining cells in the battery pack be: , , , , The peak shift of the 7th monomer is then: , The percentage decrease in peak height is: , The percentage increase in peak width is: , The percentage decrease in peak area is: .

[0038] Similarly, repeating the above calculations can yield the main peak position, main peak height, main peak width, main peak area, as well as peak position shift, peak height decrease ratio, peak width increase ratio, and peak area decrease ratio for other individual cells.

[0039] Specifically, determining the incremental capacity of each individual cell in each effective complete charging segment at each sampling point includes: The score of each cell in each effective complete charging segment at each sampling point is determined based on the voltage, mean voltage, and standard deviation of voltage of each cell at each sampling point. If the absolute value of the score is greater than a preset score threshold, then the sampling point is determined to be an abnormal voltage point. The voltage of the abnormal voltage point is repaired based on the voltage of the adjacent sampling points before and after the abnormal voltage point. The voltages at the repaired abnormal voltage points are smoothed to obtain a smoothed voltage sequence. The capacity change is determined based on the state of charge of the two sampling points before and after each sampling point, and the voltage change is determined based on the voltage of the two sampling points before and after each sampling point. The ratio of the capacity change to the voltage change is determined as the incremental capacity of each individual cell in each effective full charge segment at each sampling point.

[0040] Where the rated capacity of the battery is known Then the capacity and The correspondence is as follows: , Therefore, it can be determined by the battery's rated capacity and Determine the capacity.

[0041] For example, taking the 7th cell in a battery pack as an example, the average voltage of this cell in the current charging segment is... The standard deviation of voltage is At a certain sampling point The collected unit voltage is Then the score of that point for: , If the preset score threshold is set to 3.0, then because Therefore, this point is identified as an abnormal voltage point. The voltage at this abnormal voltage point is... It cannot be used directly and needs to be repaired. Interpolation can be used for repair. If the effective voltages adjacent to the abnormal voltage point are respectively... , If linear interpolation is used to repair this point, the repaired voltage is: .

[0042] Smoothing methods can use row Savitzky techniques. Golay smoothing, taking a window length of 5 and a polynomial order of 2 as an example, has the following general filtering coefficients: , The voltage value of the 7th cell after repair within the window is: , The smoothed center point voltage is: , Therefore, after outlier removal, interpolation repair, and smoothing, the smoothed voltage value obtained for subsequent IC curve calculation is... .

[0043] Among them, the equivalent circuit model is one of the most core and widely used battery models in BMS. It does not concern itself with the complex electrochemical reactions inside the battery, but uses basic circuit components such as resistors, capacitors, and voltage sources to simulate the voltage and current response characteristics at the battery terminals.

[0044] Specifically, determining the equivalent circuit parameters and relative deviation characteristics of the equivalent circuit parameters for each individual cell in each effective complete charging segment includes: Within the same effective complete charging segment, the current and terminal voltage of each individual battery cell are collected to establish a second-order resistive-capacitive equivalent circuit model. The recursive least squares method with a forgetting factor is used for parameter identification to obtain the model parameters of the second-order resistive-capacitive equivalent circuit model corresponding to each individual cell. The open-circuit voltage, ohmic internal resistance, short-time polarization resistance, long-time diffusion polarization resistance, short-time polarization capacitance, and long-time diffusion polarization capacitance of each individual cell are calculated based on the model parameters; and the open-circuit voltage, ohmic internal resistance, short-time polarization resistance, long-time diffusion polarization resistance, short-time polarization capacitance, and long-time diffusion polarization capacitance of each individual cell are determined as the equivalent circuit parameters of each individual cell in each effective complete charging segment. Determine the median value of the ohmic internal resistance, the median value of the short-time polarization resistance, and the median value of the long-time diffusion polarization resistance in all individual cells. The relative deviation of the ohmic internal resistance of each individual cell is calculated based on the ohmic internal resistance of each individual cell and the median value of the ohmic internal resistance. The relative deviation of the short-time polarization resistance of each individual cell is calculated based on the short-time polarization resistance of each individual cell and the median value of the short-time polarization resistance. The relative deviation of the long-time diffusion polarization resistance of each individual cell is calculated based on the long-time diffusion polarization resistance of each individual cell and the median value of the long-time diffusion polarization resistance. The relative deviations of the ohmic internal resistance, short-time polarization resistance, and long-time diffusion polarization resistance of each individual cell are determined as the relative deviation characteristics of the equivalent circuit parameters of each individual cell in each effective complete charging segment.

[0045] In a specific embodiment provided by this invention, the equivalent circuit model is a second-order capacitor-resistance RC equivalent circuit model. The second-order RC equivalent circuit model can identify the ohmic internal resistance from the current and voltage responses. Short-time polarization branch and long-term diffusion polarization branch These parameters reflect the conduction, polarization, and diffusion processes inside the battery. While these physical parameters have clear engineering implications, they are easily affected by operating conditions, state of charge (SOC) range, and noise when used alone.

[0046] Please refer to Figure 2 This diagram illustrates a structural schematic of a second-order RC equivalent circuit model provided in an embodiment of the present invention. The second-order RC equivalent circuit model includes open-circuit voltage. Ohmic internal resistance Short-time polarization branch and long-term diffusion polarization branch The terminal voltage equation is: , , , Discretize the second-order RC model and rearrange it into an autoregressive exogenous input model: , Constructing the regression vector: , Identification using recursive least squares method with forgetting factor : , , , , in, is the forgetting factor, preferably 0.995; P is the covariance matrix.

[0047] Calculate physical parameters based on the identification coefficient: , (1), , , , , in, Used to characterize ohmic internal resistance, short-time polarization resistance Long-term diffusion polarization resistance and time constant , Used to characterize short-time polarization and long-time diffusion polarization It is a short-time polarization capacitor. For long-term diffused polarization capacitors, the default sampling interval is... .

[0048] For example, within the same charging segment, the current is sampled. and terminal voltage Establish a second-order RC equivalent circuit model, whose discrete form is: , The parameters were identified using recursive least squares with a forgetting factor, yielding the parameters corresponding to the 7th individual: .

[0049] First, calculate the open-circuit voltage: , The internal resistance of the ohm is: , Based on the second-order RC parameter conversion relationship, we further obtain: , , The two polarization time constants are: , , The median value of the individual battery pack parameters within the same charging segment is: , The relative deviation of the ohmic internal resistance of the 7th individual is: , The relative deviation of short-time polarization resistance is: , The relative deviation of the long-term polarization resistance is: .

[0050] Step 103: Input the incremental capacity sequence features, the incremental capacity morphology features, the equivalent circuit parameters, and the relative deviation features of the equivalent circuit parameters into the hybrid attention network to obtain the anomaly score and fault category probability of each individual cell and the anomaly score of each effective complete charging segment output by the hybrid attention network.

[0051] For example, for the c-th cell of the t-th valid complete charging segment, its incremental capacity sequence characteristics are as follows: The incremental capacity morphology features are The equivalent circuit parameters are The relative deviation of the equivalent circuit parameters is characterized by: ,in: , , , .

[0052] Hybrid attention networks combine various attention variants with different characteristics, such as linear attention, local attention, global attention, sparse attention, channel attention, and spatial attention.

[0053] Fault categories may include, for example, increased polarization, abnormal ohmic internal resistance, capacity decay, and sensor malfunction.

[0054] Specifically, the hybrid attention network includes an incremental capacity curve encoder, an equivalent circuit parameter encoder, a cross-modal attention module, a temporal attention module, a single-unit attention module, and an output interpretation module, wherein: The incremental capacity curve encoder is used to encode the incremental capacity sequence features to obtain an incremental capacity feature vector; the equivalent circuit parameter encoder is used to encode the equivalent circuit parameters and the relative deviation features of the equivalent circuit parameters to obtain an equivalent circuit feature vector; the cross-modal attention module is used to calculate cross-modal attention based on the incremental capacity feature vector and the equivalent circuit feature vector to obtain a cross-domain feature vector. The time attention module is used to calculate the anomaly score of each valid complete charging segment based on the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; the individual cell attention module is used to calculate the anomaly score of each individual cell based on the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; the output interpretation module is used to calculate a global fusion vector based on the anomaly score of each individual cell, the anomaly score of each valid complete charging segment, the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; and to determine the fault category probability of each individual cell based on the global fusion vector.

[0055] Among them, the IC curve encoder can adopt one-dimensional convolution, TCN or Transformer structure, for example: , The input is , indicating the first The monomer in the first The IC sequence characteristics under each charging segment are output as follows: Represents the incremental capacity feature vector. For IC curve encoders, available encoders include 1D-CNN, TCN, Transformer, etc.

[0056] The RC encoder can be a multilayer sensor or a gated loop unit. , The input is and For example, the two can be concatenated to represent the first... The monomer in the first The equivalent circuit parameters and their relative deviations under each charging segment are characterized by... , , , , , and Composition, output is Represents the eigenvectors of the equivalent circuit. For equivalent circuit parameter encoders, the available encoders are MLP, GRU, etc.

[0057] The cross-modal attention module uses the incremental capacity feature vector as the query vector and the equivalent circuit feature vector as the key vector and value vector to learn the cross-domain correspondence between IC morphology changes and RC parameter drift.

[0058] For example, with a single head: , During a bullish phase: , , The input is and The output is a cross-domain feature vector. The relationship between other parameters and inputs is as follows: , , .

[0059] Among them, the time attention module is Attention is calculated between each valid complete charging segment to locate abnormal modules: , in For input, representing the first... Aggregation features of fragments, In actual implementation, given the first The splicing characteristics of all individual units within a fragment The result is obtained through mean pooling. ,Right now: , To represent the first... The abnormal score is determined by the number of valid complete charging segments, and the remaining parameters are network hyperparameters that are adjusted during network training.

[0060] Among them, the individual attention module is Calculate attention between individual cells to locate abnormal individual cells: , in For input, by , , composition, To represent the first... The abnormal score of each individual cell is used, and the remaining parameters are network hyperparameters that are adjusted during network training.

[0061] The output interpretation module includes a feature weighted fusion module and a fault type classification module. The feature weighted fusion module is as follows: , in, is the global fusion vector, and is the output.

[0062] The fault type classification module is as follows: , in The global fusion vector is the input. for The probability distribution of the root cause is the output, and the remaining parameters are hyperparameters that need to be adjusted during network training.

[0063] Step 104: Determine the faulty individual cell, the abnormal charging segment, and the primary root cause of the fault based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment.

[0064] Specifically, determining the faulty individual cell, the abnormal charging segment, and the primary root cause of the fault based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment includes: Faulty individual cells are identified based on the anomaly score of each individual cell, and abnormal charging segments are identified based on the anomaly score of each valid complete charging segment. The strength of evidence for each fault category is calculated based on the incremental capacity morphology characteristics of the faulty individual cells and the weight corresponding to each fault category. The root cause score for each fault category is calculated based on the anomaly score of the faulty individual battery, the anomaly score of the abnormal charging segment, the strength of evidence corresponding to each fault category, and the probability of each fault category. The primary root cause of the failure is determined based on the root cause score.

[0065] For example, by inputting the incremental capacity sequence features, incremental capacity morphology features, equivalent circuit parameters, and relative deviation features of equivalent circuit parameters of the 7th single cell into a hybrid attention network, the probabilities of various fault categories are obtained as follows: P(increased polarization) = 0.82, P(Ohmic internal resistance anomaly) = 0.61, P(capacity decay) = 0.46, P(sensor malfunction) = 0.08.

[0066] The 12th valid complete charging segment has the highest anomaly score, and it is: The anomaly score for a single cell is: The average attention weight for other individuals is: =0.01.

[0067] By inputting the incremental capacity sequence features, incremental capacity morphology features, equivalent circuit parameters, and relative deviation features of equivalent circuit parameters of each individual cell into a hybrid attention network, the aforementioned values ​​can be determined. Then, the abnormal individual cell is identified based on the abnormal score of the largest individual cell. Finally, the abnormal charging segment is identified based on the maximum abnormal score of the effective complete charging segment of the abnormal individual cell.

[0068] The formula for calculating the strength of evidence is as follows: ; Where k is the sequence number corresponding to the fault category. For example, when k=1, the fault category is increased polarization; when k=2, the fault category is abnormal ohmic internal resistance; when k=3, the fault category is capacity decay; and when k=4, the fault category is sensor malfunction. c is the sequence number corresponding to the individual battery cell, and t is the sequence number corresponding to the effective complete charging segment. , , , Hyperparameters that need to be adjusted during training.

[0069] For example, when k=1, the fault category is polarization intensification, and the strength of evidence is constructed as follows: , Pick: , Substitute: , Calculation yields: , , The root cause composite score is: , Regarding the root cause of "intensified polarization": .

[0070] Similarly, other fault categories can be calculated using the same method, and then the fault category corresponding to the highest root cause score is determined as the primary root cause.

[0071] In summary, compared with the prior art, the beneficial effects of the present invention are as follows: First, this invention uses incremental capacity sequence features, incremental capacity morphology features, equivalent circuit parameters, and relative deviation features of equivalent circuit parameters as input data. The incremental capacity sequence features and incremental capacity morphology features are extracted from the incremental capacity curve. The incremental capacity curve can amplify minute changes near the charging voltage plateau, enabling the detection of capacity decay, polarization anomalies, and individual cell inconsistencies before the traditional voltage threshold is triggered, demonstrating strong early fault identification capabilities. The equivalent circuit parameters and relative deviation features have clear physical meanings, allowing the output of the hybrid attention network to be beyond the limitations of black-box classification results.

[0072] Secondly, the hybrid attention network can output the anomaly score and fault category probability of each individual battery cell, as well as the anomaly score of each valid complete charging segment. Thus, the faulty individual battery cell can be identified based on the anomaly score of each individual battery cell, the abnormal charging segment can be identified based on the anomaly score of each valid complete charging segment, and the primary root cause of the fault can be identified based on the fault category probability of each individual battery cell.

[0073] This battery failure root cause tracing mechanism can effectively solve the problems of difficulty in locating faulty individual cells, determining abnormal charging curve ranges, and analyzing root causes in the battery management systems of new energy vehicles, thereby improving battery safety in electric vehicles, motors, and electric drives under complex operating conditions. In addition to battery safety in the new energy vehicle field, this battery failure root cause tracing mechanism can also be applied to the system safety of methods or devices that directly convert chemical energy into electrical energy, comprehensively serving electric vehicles and energy-saving, environmentally friendly, and green power applications.

[0074] Figure 3 A schematic diagram of a battery fault root cause tracing device based on incremental capacity curves and equivalent circuits according to an embodiment of the present invention is shown. The device includes: Information acquisition unit 301 is used to acquire multiple valid and complete charging segments; The feature determination unit 302 is used to determine the incremental capacity sequence features and incremental capacity morphology features of each individual cell in each effective complete charging segment, and to determine the equivalent circuit parameters and the relative deviation features of the equivalent circuit parameters of each individual cell in each effective complete charging segment. The model running unit 303 is used to input the incremental capacity sequence features, the incremental capacity morphology features, the equivalent circuit parameters, and the relative deviation features of the equivalent circuit parameters into the hybrid attention network to obtain the abnormal score and fault category probability of each individual battery and the abnormal score of each effective complete charging segment output by the hybrid attention network. The root cause tracing unit 304 is used to determine the faulty individual cell, the abnormal charging segment, and the primary root cause of the fault based on the abnormal score and fault category probability of each individual cell and the abnormal score of each valid complete charging segment.

[0075] Figure 4 The diagram illustrates the structure of a computer device according to an embodiment of the present invention, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the functions of the computer system of the battery failure root cause tracing method based on incremental capacity curves and equivalent circuits in any of the above embodiments.

[0076] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a computer, causes the computer to perform the functions of the computer system of the battery failure root cause tracing method based on incremental capacity curves and equivalent circuits in any of the above embodiments.

[0077] The present invention also provides a computer program product containing instructions that, when executed by a computer, cause the computer to perform the functions of the computer system of the battery fault root cause tracing method based on incremental capacity curves and equivalent circuits in any of the above embodiments.

[0078] It is understood that the specific examples in this invention are only intended to help those skilled in the art better understand the embodiments of this invention, and are not intended to limit the scope of this invention.

[0079] It is understood that in the various embodiments of the present invention, the order of the process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not limit the implementation process of the embodiments of the present invention in any way.

[0080] It is understood that the various embodiments described in this invention can be implemented individually or in combination, and the embodiments of this invention are not limited in this respect.

[0081] Unless otherwise stated, all technical and scientific terms used in embodiments of the present invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the scope of the invention. The term "and / or" as used in this invention includes any and all combinations of one or more of the associated listed items. The singular forms "a," "the," and "the" as used in embodiments of the invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0082] It is understood that the processor in the embodiments of the present invention can be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method embodiments can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0083] It is understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Specifically, the non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM). It should be noted that the memory in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0084] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0085] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the aforementioned method implementations, and will not be repeated here.

[0086] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0087] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0088] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0089] If the functionality is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0090] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for tracing the root causes of battery failures based on incremental capacity curves and equivalent circuits, characterized in that, The method includes: Acquire multiple valid and complete charging segments; Determine the incremental capacity sequence characteristics and incremental capacity morphology characteristics of each individual cell in each effective complete charging segment, and determine the equivalent circuit parameters and relative deviation characteristics of the equivalent circuit parameters of each individual cell in each effective complete charging segment; The incremental capacity sequence features, the incremental capacity morphology features, the equivalent circuit parameters, and the relative deviation features of the equivalent circuit parameters are input into a hybrid attention network to obtain the anomaly score and fault category probability of each individual cell and the anomaly score of each effective complete charging segment output by the hybrid attention network. The faulty individual cell, the abnormal charging segment, and the primary root cause of failure are determined based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment.

2. The method according to claim 1, characterized in that, The process of determining the faulty individual cell, the abnormal charging segment, and the primary root cause of the fault based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment includes: Faulty individual cells are identified based on the anomaly score of each individual cell, and abnormal charging segments are identified based on the anomaly score of each valid complete charging segment. The strength of evidence for each fault category is calculated based on the incremental capacity morphology characteristics of the faulty individual cells and the weight corresponding to each fault category. The root cause score for each fault category is calculated based on the anomaly score of the faulty individual battery, the anomaly score of the abnormal charging segment, the strength of evidence corresponding to each fault category, and the probability of each fault category. The primary root cause of the failure is determined based on the root cause score.

3. The method according to claim 2, characterized in that, The hybrid attention network includes an incremental capacity curve encoder, an equivalent circuit parameter encoder, a cross-modal attention module, a temporal attention module, a single-unit attention module, and an output interpretation module, wherein: The incremental capacity curve encoder is used to encode the incremental capacity sequence features to obtain an incremental capacity feature vector; the equivalent circuit parameter encoder is used to encode the equivalent circuit parameters and the relative deviation features of the equivalent circuit parameters to obtain an equivalent circuit feature vector; the cross-modal attention module is used to calculate cross-modal attention based on the incremental capacity feature vector and the equivalent circuit feature vector to obtain a cross-domain feature vector. The time attention module is used to calculate the anomaly score of each valid complete charging segment based on the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; the individual cell attention module is used to calculate the anomaly score of each individual cell based on the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; the output interpretation module is used to calculate a global fusion vector based on the anomaly score of each individual cell, the anomaly score of each valid complete charging segment, the incremental capacity feature vector, the equivalent circuit feature vector, and the cross-domain feature vector; and to determine the fault category probability of each individual cell based on the global fusion vector.

4. The method according to claim 1, characterized in that, The acquisition of multiple valid complete charging segments includes: Select multiple consecutive charging segments; For each of the continuous charging segments, the state of charge increment of the continuous charging segment is determined based on the difference between the start value and the end value of the state of charge of the continuous charging segment. If the increment of the state of charge is greater than a preset state threshold, then the continuous charging segment is determined to be a valid and complete charging segment.

5. The method according to claim 1, characterized in that, The determination of the incremental capacity sequence characteristics and incremental capacity morphology characteristics of each individual cell in each effective complete charging segment includes: Determine the incremental capacity of each individual cell in each of the effective full charging segments at each sampling point; The incremental capacity curve for each individual cell in each effective full charge segment is determined based on the incremental capacity of each cell at all sampling points. The incremental capacity curve of each cell in each of the effective full charge segments is determined as the incremental capacity sequence feature of each cell in each of the effective full charge segments; and the incremental capacity curve of each cell in each of the effective full charge segments is interpolated to a uniform cell voltage grid. A search is performed on the unified single-cell voltage grid to obtain the main peak characteristics of each single cell; The characteristic variation of the main peak feature of each individual cell is determined based on the median value of the main peak feature; The characteristic variation of the main peak feature of each individual cell is determined as the incremental capacity morphology feature of each individual cell in each effective full charge segment.

6. A method according to claim 5, characterized in that, Determining the incremental capacity of each individual cell in each of the effective complete charging segments at each sampling point includes: The score of each cell in each effective complete charging segment at each sampling point is determined based on the voltage, mean voltage, and standard deviation of voltage of each cell at each sampling point. If the absolute value of the score is greater than a preset score threshold, then the sampling point is determined to be an abnormal voltage point. The voltage of the abnormal voltage point is repaired based on the voltage of the adjacent sampling points before and after the abnormal voltage point. The voltages at the repaired abnormal voltage points are smoothed to obtain a smoothed voltage sequence. The capacity change is determined based on the state of charge of the two sampling points before and after each sampling point, and the voltage change is determined based on the voltage of the two sampling points before and after each sampling point. The ratio of the capacity change to the voltage change is determined as the incremental capacity of each individual cell in each effective full charge segment at each sampling point.

7. The method according to claim 5, characterized in that, The determination of the equivalent circuit parameters and relative deviation characteristics of the equivalent circuit parameters for each individual cell in each effective complete charging segment includes: Within the same effective complete charging segment, the current and terminal voltage of each individual battery cell are collected to establish a second-order resistive-capacitive equivalent circuit model. The recursive least squares method with a forgetting factor is used for parameter identification to obtain the model parameters of the second-order resistive-capacitive equivalent circuit model corresponding to each individual cell. The open-circuit voltage, ohmic internal resistance, short-time polarization resistance, long-time diffusion polarization resistance, short-time polarization capacitance, and long-time diffusion polarization capacitance of each individual cell are calculated based on the model parameters; and the open-circuit voltage, ohmic internal resistance, short-time polarization resistance, long-time diffusion polarization resistance, short-time polarization capacitance, and long-time diffusion polarization capacitance of each individual cell are determined as the equivalent circuit parameters of each individual cell in each effective complete charging segment. Determine the median value of the ohmic internal resistance, the median value of the short-time polarization resistance, and the median value of the long-time diffusion polarization resistance in all individual cells. The relative deviation of the ohmic internal resistance of each individual cell is calculated based on the ohmic internal resistance of each individual cell and the median value of the ohmic internal resistance. The relative deviation of the short-time polarization resistance of each individual cell is calculated based on the short-time polarization resistance of each individual cell and the median value of the short-time polarization resistance. The relative deviation of the long-time diffusion polarization resistance of each individual cell is calculated based on the long-time diffusion polarization resistance of each individual cell and the median value of the long-time diffusion polarization resistance. The relative deviations of the ohmic internal resistance, short-time polarization resistance, and long-time diffusion polarization resistance of each individual cell are determined as the relative deviation characteristics of the equivalent circuit parameters of each individual cell in each effective complete charging segment.

8. A battery fault root cause tracing device based on incremental capacity curves and equivalent circuits, characterized in that, The device includes: The information acquisition unit is used to acquire multiple valid and complete charging segments; The feature determination unit is used to determine the incremental capacity sequence features and incremental capacity morphology features of each individual cell in each effective complete charging segment, and to determine the equivalent circuit parameters and the relative deviation features of the equivalent circuit parameters of each individual cell in each effective complete charging segment. The model running unit is used to input the incremental capacity sequence features, the incremental capacity morphology features, the equivalent circuit parameters, and the relative deviation features of the equivalent circuit parameters into the hybrid attention network to obtain the anomaly score and fault category probability of each individual battery and the anomaly score of each effective complete charging segment output by the hybrid attention network. The root cause tracing unit is used to determine the faulty individual cell, the abnormal charging segment, and the primary root cause of the fault based on the anomaly score and fault category probability of each individual cell and the anomaly score of each valid complete charging segment.

9. An electronic device, characterized in that, include: Processor and memory; The processor is connected to a memory, wherein the memory is used to store a computer program, and the processor is used to invoke the computer program to perform the 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 a computer program, the computer program including program instructions that, when executed by a processor, perform the method as described in any one of claims 1-7.

Images