Battery abnormity early warning method, device and equipment and computer readable storage medium

Abstract

The embodiment of the invention relates to the technical field of data processing, and discloses a battery abnormity early warning method, device and equipment and a computer readable storage medium. According to the technical scheme, the physical core parameters are determined according to the collected charging data, so that the charging data in the charging process can be converted into the physically interpretable time sequence core parameters on the premise of not increasing sensors; then the electrochemical differential voltage kernel, the thermal inertia kernel and the current-voltage response kernel are input into a preset battery anomaly detection model, the target anomaly probability corresponding to the vehicle battery can be accurately obtained through the trained preset battery anomaly detection model, and then early warning of battery anomaly is performed according to the target anomaly probability. The thermal runaway risk is reduced.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, specifically to a battery anomaly early warning method, apparatus, device, and computer-readable storage medium. Background Technology

[0002] Currently, battery management systems generally use a fixed threshold method to determine battery anomalies, judging signals such as voltage, temperature, and current. However, this fixed threshold method has the following problems: (1) Faults such as lithium plating, micro-internal short circuit, and electrolyte drying that occur in the early stages of charging have extremely weak characteristic signals, and the threshold method will miss them; (2) The battery management system needs to wait for the signal to accumulate to 2-5 amp-hours of charging power before triggering an alarm. At this time, the fault has already developed to an irreversible stage, and it is impossible to provide early warning, which significantly increases the risk of thermal runaway. Summary of the Invention

[0003] In view of the above problems, embodiments of the present invention provide a battery anomaly early warning method, apparatus, device and computer-readable storage medium to solve the problems of the prior art being unable to accurately and effectively provide early warning of battery anomalies and having a high risk of thermal runaway.

[0004] According to one aspect of the present invention, a battery anomaly warning method is provided, the method comprising: During the vehicle battery charging process, physical core parameters are determined based on the collected charging data. The physical core parameters include at least: an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. The electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel are input into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. Battery anomaly warning is issued based on the target anomaly probability.

[0005] In one alternative approach, determining the physical core parameters based on collected charging data during vehicle battery charging includes: During the vehicle battery charging process, charging data is collected at preset power intervals. The charging data includes at least: voltage data, temperature data, and current data. The electrochemical differential physics core was determined based on the voltage data; The thermal inertial core is determined based on the temperature data; The current-voltage response kernel is determined based on the current data.

[0006] In one alternative approach, determining the electrochemical differential physical core based on the voltage data includes: The next voltage for the next charge and the current voltage for the current charge are determined based on the voltage data, and the voltage change is calculated based on the next voltage and the current voltage. Calculate the change in power based on the next power level and the current power level; The electrochemical differential physical nucleus is calculated based on the voltage change and the charge change.

[0007] In one alternative approach, determining the thermal inertial core based on the temperature data includes: Determine the actual temperature under the current power level based on the temperature data; The estimated temperature under the current power level is determined by a preset linear model; The thermal inertial core is calculated based on the difference between the actual temperature and the estimated temperature.

[0008] In one alternative approach, determining the current-voltage response kernel based on the current data includes: Determine the actual voltage under the current power level based on the current data; The estimated voltage under the current power level is determined by a preset nonlinear model; The current-voltage response kernel is calculated based on the difference between the actual voltage and the estimated voltage.

[0009] In one optional approach, inputting the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery includes: The electrochemical differential voltage kernel under each battery, the thermal inertia kernel under each thermometer, and the current-voltage response kernel under each time interval are respectively input into the encoding layer of the preset battery anomaly detection model for encoding, so as to obtain the electrochemical differential voltage kernel hidden vector, thermal inertia kernel hidden vector, and current-voltage response kernel hidden vector of the same dimension. Using the electrochemical differential voltage kernel latent vector as the query, and the thermal inertia kernel latent vector and the current-voltage response kernel latent vector as the key and value, the electrochemical differential voltage kernel latent vector, the thermal inertia kernel latent vector and the current-voltage response kernel latent vector are fused through the cross-attention mechanism in the coding layer to obtain fused features; The target score is obtained by linearly transforming the fused features through the linear layer in the preset battery anomaly detection model. Substituting the target score into a preset activation function yields the target anomaly probability corresponding to the vehicle battery.

[0010] In one optional approach, before inputting the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery, the method further includes: The initial sample corresponding to the vehicle battery is input into the initial battery anomaly detection model to obtain the initial anomaly probability corresponding to the vehicle battery; The loss value is calculated based on the initial anomaly probability and the label corresponding to the initial sample; The initial battery anomaly detection model is optimized based on the loss value to obtain a preset battery anomaly detection model.

[0011] According to another aspect of the present invention, a battery abnormality warning device is provided, comprising: The data acquisition module is used to determine physical core parameters based on the collected charging data during the vehicle battery charging process. The physical core parameters include at least: an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. The probability determination module is used to input the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. The battery anomaly warning module is used to issue a battery anomaly warning based on the target anomaly probability.

[0012] According to another aspect of the present invention, a battery abnormality early warning device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus; The memory is used to store at least one executable instruction that causes the processor to perform the operation of the battery anomaly warning method described above.

[0013] According to another aspect of the present invention, a computer-readable storage medium is provided, the storage medium storing at least one executable instruction that causes a battery abnormality warning device / apparatus to perform the operation of the battery abnormality warning method as described above.

[0014] This invention, through its embodiments, determines physical core parameters based on collected charging data during vehicle battery charging. These physical core parameters include at least an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. These parameters are then input into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. Based on this target anomaly probability, a battery anomaly warning is issued. This invention, by determining physical core parameters based on collected charging data, can convert charging data into physically interpretable time-series core parameters without adding sensors. The input of the electrochemical differential voltage core, thermal inertia core, and current-voltage response core into the preset battery anomaly detection model allows for accurate determination of the target anomaly probability corresponding to the vehicle battery through a trained preset battery anomaly detection model. This early warning of battery anomalies based on the target anomaly probability reduces the risk of thermal runaway.

[0015] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0016] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a first embodiment of the battery anomaly warning method provided by the present invention is shown. Figure 2 A flowchart illustrating a second embodiment of the battery anomaly warning method provided by the present invention is shown. Figure 3 A flowchart illustrating a third embodiment of the battery anomaly warning method provided by the present invention is shown. Figure 4 A flowchart illustrating the determination of the target anomaly probability in an embodiment of the battery anomaly early warning method provided by the present invention is shown; Figure 5 This invention provides an overall flowchart of an embodiment of the battery anomaly early warning method. Figure 6 A schematic diagram of the structure of a first embodiment of the battery abnormality warning device provided by the present invention is shown; Figure 7 A schematic diagram of an embodiment of the battery anomaly warning device provided by the present invention is shown. Detailed Implementation

[0017] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0018] Figure 1 A flowchart of a first embodiment of the battery anomaly warning method of the present invention is shown, which is executed by a battery anomaly warning device. Figure 1 As shown, the method includes the following steps: Step 10: During the vehicle battery charging process, determine the physical core parameters based on the collected charging data. The physical core parameters include at least: electrochemical differential voltage core, thermal inertia core, and current-voltage response core.

[0019] Understandably, during the charging process of the vehicle battery, charging data can be collected. The charging data may include data such as the current, voltage, and temperature of the battery pack. The above data can be collected by current sensors, voltage sensors, thermometers, etc. around the battery pack, or by other means. This embodiment does not make specific limitations on this.

[0020] It should be understood that physical core parameters can be determined based on charging data, and the current, voltage, and temperature data during the charging process can be transformed into three types of physically interpretable heterogeneous core features. The physical core parameters may include at least: (1) the electrochemical differential voltage kernel (DVS) used to characterize electrochemical polarization abrupt changes, which is the most sensitive indicator of early faults such as lithium plating and loss of active materials; (2) the thermal inertia kernel (TIK, current-temperature hysteresis residual) used to capture deviations in thermal behavior such as internal short circuits and abnormal heat dissipation; (3) the current-voltage response kernel (IVR, second-order Volterra coefficient sequence) characterizes the dynamic evolution of ohmic + charge transfer impedance, and provides early warning for impedance-related faults such as micro-short circuits and electrolyte drying. This embodiment breaks away from the traditional "amplitude domain" approach and turns to physically interpretable kernels.

[0021] Step 20: Input the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into the preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery.

[0022] Understandably, the preset anomaly detection model refers to a pre-set model used to detect whether the vehicle battery is abnormal. By inputting the electrochemical differential voltage kernel, thermal inertia kernel, and current-voltage response kernel into the preset battery anomaly detection model, the target anomaly probability corresponding to the vehicle battery can be obtained. The target anomaly probability refers to the probability that the vehicle battery is abnormal.

[0023] Step 30: Issue a battery anomaly warning based on the target anomaly probability.

[0024] In a specific implementation, battery anomaly warnings can be issued based on the target anomaly probability. In one feasible embodiment, if the target anomaly probability is greater than a preset probability, the corresponding vehicle battery is determined to be in an abnormal state. The preset probability can be set according to the actual situation, and this embodiment does not impose specific restrictions on it.

[0025] This embodiment determines physical kernel parameters based on collected charging data during vehicle battery charging. These parameters include at least an electrochemical differential voltage kernel, a thermal inertia kernel, and a current-voltage response kernel. These kernels are then input into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. Based on this probability, a battery anomaly warning is issued. This embodiment determines the physical kernel parameters based on collected charging data, converting charging data into physically interpretable time-series kernel parameters without adding sensors. The input of the electrochemical differential voltage kernel, thermal inertia kernel, and current-voltage response kernel into the preset battery anomaly detection model allows for accurate determination of the target anomaly probability of the vehicle battery through a trained model. This early warning of battery anomalies based on the target anomaly probability reduces the risk of thermal runaway.

[0026] Figure 2 A flowchart of a second embodiment of the battery anomaly warning method of the present invention is shown, which is executed by a battery anomaly warning device. Figure 2 As shown, step 10 includes the following steps: Step 101: During the vehicle battery charging process, charging data is collected at preset power intervals. The charging data includes at least voltage data, temperature data, and current data.

[0027] It is understood that in this embodiment, during the vehicle battery charging process, charging data can be collected according to a preset power interval, which can be 0.05 amp-hours, 0.06 amp-hours, etc. This embodiment does not impose any specific restrictions on this.

[0028] In a specific implementation, charging data can be collected according to a preset power interval. The charging data includes at least voltage data, temperature data, and current data. In a feasible embodiment, abnormal values ​​such as null values ​​in the above charging data can be cleaned to make the obtained charging data normal data.

[0029] Step 102: Determine the electrochemical differential physical core based on the voltage data.

[0030] It should be understood that in this embodiment, the electrochemical differential physical core can be calculated based on the voltage data collected under a preset power interval, and the dimension of the electrochemical differential physical core is related to the number of vehicle batteries.

[0031] In an optional embodiment, step 102 includes: determining the next voltage of the next energy level and the current voltage of the current energy level based on the voltage data, and calculating the voltage change based on the next voltage and the current voltage; calculating the energy change based on the next energy level and the current energy level; and calculating the electrochemical differential physical core based on the voltage change and the energy change.

[0032] Understandably, "next energy" refers to the energy collected at the next voltage, "current energy" refers to the energy collected at the current voltage, "voltage change" refers to the difference between the next voltage and the current voltage, and "energy change" refers to the difference between the next energy and the current energy. The electrochemical differential physical nucleus can be calculated using a first preset formula, which is:

[0033] In the formula, This indicates the sampling point number. For example, if the preset battery interval is 0.05 amp-hours, then 0.1 amp-hours represents the first sampling point, 0.15 amp-hours represents the second sampling point, and 0.2 amp-hours represents the third sampling point. Here, 1, 2, and 3 represent... . This represents the change in voltage, based on the next voltage. and current voltage The difference between them is obtained. This indicates the change in electricity consumption, based on the next electricity consumption. and current battery level The difference between them is obtained. For each vehicle battery, the corresponding electrochemical differential physical core can be calculated using the first preset formula mentioned above.

[0034] Step 103: Determine the thermal inertial core based on the temperature data.

[0035] It should be understood that in this embodiment, the thermal inertial core can be calculated based on the voltage data collected under a preset power interval, and the dimension of the thermal inertial core is related to the number of thermometers.

[0036] In one alternative approach, step 103 includes: determining the actual temperature under the current power level based on the temperature data; determining the estimated temperature under the current power level using a preset linear model; and calculating the thermal inertial core based on the difference between the actual temperature and the estimated temperature.

[0037] It is understood that in this embodiment, the thermal inertial core can be calculated according to the second preset formula, which is: In the formula, This represents the thermal inertial core at the k-th sampling point. This represents the actual battery temperature at the k-th sampling point. This represents the estimated battery temperature at the k-th sampling point.

[0038] In one feasible embodiment, it can be calculated using a third preset formula. The third preset formula is:

[0039] In the formula, the estimated temperature at the k-th sampling point It is the weighted sum of the current input I(kj) over the past M time points. is the system's impulse response coefficient, representing the degree of influence of historical input on current output. M is the model order, which determines the length of the system's memory. Temperature can be estimated using a preset linear model, which can be a pre-defined model that defines the aforementioned linear regression vector.

[0040] In one feasible embodiment, a linear regression vector can be defined: , The third preset formula mentioned above can be expressed as: The second preset formula mentioned above can be expressed as: , It can also be temperature error, representing the difference between the actual temperature and the estimated temperature. The Kalman gain is: This determines the weight of the impact of new data on parameter updates. The parameter updates are as follows: The impulse response coefficient is estimated using error correction. The covariance is updated as follows: [ - Uncertainty in updating parameter estimates This is the forgetting factor, which can range from 0.98 to 0.995. For each thermometer, the corresponding thermal inertia core can be calculated using the second preset formula mentioned above.

[0041] Step 104: Determine the current-voltage response kernel based on the current data.

[0042] In one alternative approach, step 104 includes: determining the actual voltage under the current power based on the current data; determining the estimated voltage under the current power using a preset nonlinear model; and calculating a current-voltage response kernel based on the difference between the actual voltage and the estimated voltage.

[0043] In one feasible embodiment, a nonlinear regression vector can be defined: The regression vector consists of two parts: a linear part and a linear part. (First-order terms) and nonlinear components: (Second-order term). The voltage can be estimated using a preset nonlinear model, which can be a pre-defined model that defines the aforementioned nonlinear regression vector.

[0044] In one feasible embodiment, the steps for calculating the thermal inertial core described above can be referred to. , It can also be voltage error, representing the difference between the actual voltage and the estimated voltage. The Kalman gain is: This determines the weight of the impact of new data on parameter updates. The parameter updates are as follows: The impulse response coefficient is estimated using error correction. The covariance is updated as follows: [ - Uncertainty in updating parameter estimates This is the forgetting factor, ranging from 0.98 to 0.995. Here, M is the IVR core memory length, related to the time of charge transfer and diffusion processes (generally less than 5 seconds), converted to an equal-ampere time interval, using the number of capacity cells as the sampling unit. M = physical time constant τ / capacity per cell ΔQ. In this embodiment, M can be directly set to 10. The IVR(k) matrix is ​​(m,K), where K = M + M(M+1) / 2. For each K, the corresponding current-voltage response core can be calculated using the second preset formula mentioned above.

[0045] This embodiment collects charging data at preset power intervals during vehicle battery charging. The charging data includes at least voltage, temperature, and current data. Then, an electrochemical differential physical core is determined based on the voltage data, a thermal inertia core is determined based on the temperature data, and a current-voltage response core is determined based on the current data. This embodiment converts the voltage, temperature, and current data during charging into three types of physical cores with equidistant capacity, which may include an electrochemical differential physical core, a thermal inertia core, and a current-voltage response core.

[0046] Figure 3 A flowchart of a third embodiment of the battery anomaly warning method of the present invention is shown, which is executed by a battery anomaly warning device. Figure 3 As shown, step 20 includes the following steps: Step 201: Input the electrochemical differential voltage kernel under each battery, the thermal inertia kernel under each thermometer, and the current-voltage response kernel under each time interval into the encoding layer of the preset battery anomaly detection model for encoding, and obtain the electrochemical differential voltage kernel latent vector, thermal inertia kernel latent vector, and current-voltage response kernel latent vector of the same dimension.

[0047] Understandably, Figure 4A flowchart illustrating the determination of the target anomaly probability in an embodiment of the battery anomaly early warning method provided by the present invention is shown, as follows: Figure 4 As shown, the electrochemical differential voltage kernel for each battery, the thermal inertia kernel for each thermometer, and the current-voltage response kernel for each time interval are input into the encoding layer of a pre-defined battery anomaly detection model for encoding. The encoding layer may include 1D-CNN and Transformer to obtain the same-dimensional electrochemical differential voltage kernel hidden vector, thermal inertia kernel hidden vector, and current-voltage response kernel hidden vector. Figure 4 In the diagram, DVS(m,cellnums)→1D-CNN→(m,d)→Transformer→h_DVS(B,1,d), where m is the total number of data entries, cellnums is the number of batteries, and h_DVS(B,1,d) is the electrochemical differential voltage kernel hidden vector; TIK(m,Tnums)→1D-CNN→(m,d)→Transformer→h_TIK(B,m,d), where Tnums is the number of thermometers, and h_TIK(B,m,d) is the thermal inertia kernel hidden vector; IVR(m,K)→1D-CNN→(m,d)→Transformer→h_IVR(B,m,d), where h_IVR(B,m,d) is the current-voltage response kernel hidden vector, and B is the batch size.

[0048] Step 202: Using the electrochemical differential voltage kernel latent vector as the query, the thermal inertia kernel latent vector and the current-voltage response kernel latent vector as the key and value, the electrochemical differential voltage kernel latent vector, the thermal inertia kernel latent vector and the current-voltage response kernel latent vector are fused through the cross-attention mechanism in the coding layer to obtain fused features.

[0049] It should be understood that the DVS latent vector is used as the query, and the thermal inertia latent vector and current-voltage response latent vector are used as the key and value, respectively. That is, the DVS latent vector is used as the query, and the TIK and IVR latent vectors are used as the key / value pairs. The DVS latent vector, TIK latent vector, and IVR latent vector are fused through a cross-attention mechanism in the encoding layer to obtain fused features. For example... Figure 4 Cross-Attention(DVS-Q,TIK+IVR-KV) is used to obtain the fused feature fused(B,1,d).

[0050] Step 203: Perform a linear transformation on the fused features through the linear layer in the preset battery anomaly detection model to obtain the target score.

[0051] Understandably, the target score is calculated by performing a linear transformation on the fused features through a linear layer (Linear), then weighting and summing the fused features.

[0052] Step 204: Substitute the target score into the preset activation function to obtain the target anomaly probability corresponding to the vehicle battery.

[0053] It should be understood that substituting the target score into the preset activation function sigmoid maps the target score to the (0,1) interval. The larger the target score, the closer the target anomaly probability prob(B,1) is to 1, and the smaller the target score, the closer the target anomaly probability is to 0.

[0054] In one alternative approach, before step 20, the method further includes: inputting an initial sample corresponding to the vehicle battery into an initial battery anomaly detection model to obtain an initial anomaly probability corresponding to the vehicle battery; calculating a loss value based on the initial anomaly probability and the label corresponding to the initial sample; and optimizing the initial battery anomaly detection model based on the loss value to obtain a preset battery anomaly detection model.

[0055] Understandably, the initial battery anomaly detection model can be trained to obtain a preset battery anomaly detection model. The structure of the initial battery anomaly detection model is the same as that of the preset battery anomaly detection model. The initial sample refers to the sample used to train the initial anomaly detection model. By inputting the initial sample into the initial battery anomaly detection model, the initial anomaly probability corresponding to the vehicle battery can be obtained.

[0056] In one feasible embodiment, the loss function may be:

[0057] In the formula, Indicates the loss value. ∈{0,1} represents the label, that is, the anomaly probability of the initial manually labeled sample. ∈{0,1} represents the initial anomaly probability, which is the anomaly probability output by the initial battery anomaly detection model.

[0058] It should be understood that the initial battery anomaly detection model can be optimized based on the loss value to obtain a preset battery anomaly detection model. Specifically, when the loss value gradually converges, it can be determined that the initial battery anomaly detection model has been trained and the trained initial battery anomaly detection model can be used as the preset battery anomaly detection model.

[0059] In the specific implementation, Figure 5 A flowchart illustrating an embodiment of the battery anomaly warning method provided by the present invention is shown, as follows: Figure 5As shown, the data input can be voltage, current, temperature, etc. Data processing and resampling are then performed to obtain outlier-handled data. Charging time-series data is then extracted, resulting in three physical kernel parameters: the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel. A 1-D CNN+Transformer encoding method is then used to output a hidden vector of the same dimension. Simultaneously, the DVS hidden vector is used as the query, and the TIK and IVR hidden vectors are used as the key / value pairs. These are fused through a cross-modal attention mechanism to output a fused feature. This fused feature is then input into a feedforward classification head, which directly outputs the target anomaly probability (0~1), thereby performing battery anomaly detection.

[0060] This embodiment encodes the electrochemical differential voltage kernel for each battery, the thermal inertia kernel for each thermometer, and the current-voltage response kernel for each time interval into the encoding layer of a preset battery anomaly detection model. This yields latent vectors of the same dimension for the electrochemical differential voltage kernel, thermal inertia kernel, and current-voltage response kernel. Using the electrochemical differential voltage kernel as the query and the thermal inertia and current-voltage response kernels as the key and value, a cross-attention mechanism in the encoding layer fuses these latent vectors to obtain a fused feature. This fused feature is then linearly transformed by a linear layer in the preset battery anomaly detection model to obtain a target score. The target score is then substituted into a preset activation function to obtain the target anomaly probability corresponding to the vehicle battery. This embodiment uses a cross-attention mechanism in the encoding layer to fuse the electrochemical differential voltage kernel, thermal inertia kernel, and current-voltage response kernel to obtain a fused feature, and then accurately and effectively calculates the target anomaly probability corresponding to the vehicle battery based on this fused feature.

[0061] Figure 6 A schematic diagram of an embodiment of the battery malfunction warning device of the present invention is shown. Figure 6 As shown, the device 600 includes: a data acquisition module 610, a probability determination module 620, and a battery anomaly warning module 630; The data acquisition module 610 is used to determine physical core parameters based on the collected charging data during the vehicle battery charging process. The physical core parameters include at least: an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. The probability determination module 620 is used to input the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. The battery anomaly warning module 630 is used to provide a battery anomaly warning based on the target anomaly probability.

[0062] In an optional embodiment, the data acquisition module 610 is further configured to acquire charging data at preset power intervals during vehicle battery charging, the charging data including at least: voltage data, temperature data, and current data; determine an electrochemical differential physical core based on the voltage data; determine a thermal inertia core based on the temperature data; and determine a current-voltage response core based on the current data.

[0063] In an optional embodiment, the data acquisition module 610 is further configured to determine the next voltage of the next energy level and the current voltage of the current energy level based on the voltage data, and calculate the voltage change based on the next voltage and the current voltage; calculate the energy change based on the next energy level and the current energy level; and calculate the electrochemical differential physical core based on the voltage change and the energy change.

[0064] In an optional embodiment, the data acquisition module 610 is further configured to determine the actual temperature under the current power level based on the temperature data; determine the estimated temperature under the current power level using a preset linear model; and calculate the thermal inertia core based on the difference between the actual temperature and the estimated temperature.

[0065] In an optional embodiment, the data acquisition module 610 is further configured to determine the actual voltage under the current power consumption based on the current data; determine the estimated voltage under the current power consumption through a preset nonlinear model; and calculate the current-voltage response kernel based on the difference between the actual voltage and the estimated voltage.

[0066] In an optional manner, the probability determination module 620 is further configured to input the electrochemical differential voltage kernel under each battery, the thermal inertia kernel under each thermometer, and the current-voltage response kernel under each time interval into the encoding layer of a preset battery anomaly detection model for encoding, thereby obtaining electrochemical differential voltage kernel latent vectors, thermal inertia kernel latent vectors, and current-voltage response kernel latent vectors of the same dimension; using the electrochemical differential voltage kernel latent vector as the query, and the thermal inertia kernel latent vector and the current-voltage response kernel latent vector as the key and value, the electrochemical differential voltage kernel latent vector, the thermal inertia kernel latent vector, and the current-voltage response kernel latent vector are fused through the cross-attention mechanism in the encoding layer to obtain fused features; the fused features are linearly transformed through the linear layer in the preset battery anomaly detection model to obtain a target score; and the target score is substituted into a preset activation function to obtain the target anomaly probability corresponding to the vehicle battery.

[0067] In an optional embodiment, the probability determination module 620 is further configured to input the initial sample corresponding to the vehicle battery into the initial battery anomaly detection model to obtain the initial anomaly probability corresponding to the vehicle battery; calculate a loss value based on the initial anomaly probability and the label corresponding to the initial sample; and optimize the initial battery anomaly detection model based on the loss value to obtain a preset battery anomaly detection model.

[0068] This invention, through its embodiments, determines physical core parameters based on collected charging data during vehicle battery charging. These physical core parameters include at least an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. These parameters are then input into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. Based on this target anomaly probability, a battery anomaly warning is issued. This invention, by determining physical core parameters based on collected charging data, can convert charging data into physically interpretable time-series core parameters without adding sensors. The input of the electrochemical differential voltage core, thermal inertia core, and current-voltage response core into the preset battery anomaly detection model allows for accurate determination of the target anomaly probability corresponding to the vehicle battery through a trained preset battery anomaly detection model. This early warning of battery anomalies based on the target anomaly probability reduces the risk of thermal runaway.

[0069] Figure 7 The diagram shows a structural schematic of an embodiment of the battery anomaly warning device of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the battery anomaly warning device.

[0070] like Figure 7 As shown, the battery abnormality warning device may include: a processor 702, a communication interface 404, a memory 706, and a communication bus 708.

[0071] The processor 702, communication interface 704, and memory 706 communicate with each other via communication bus 708. Communication interface 704 is used to communicate with other network elements such as clients or other servers. The processor 702 executes program 710, specifically performing the relevant steps described above in the embodiment of the battery anomaly warning method.

[0072] Specifically, program 710 may include program code, which includes computer-executable instructions.

[0073] The processor 702 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The battery malfunction warning device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0074] Memory 706 is used to store program 710. Memory 706 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0075] Specifically, program 710 can be called by processor 402 to cause the battery abnormality warning device to perform the following operations: During the vehicle battery charging process, physical core parameters are determined based on the collected charging data. The physical core parameters include at least: an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. The electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel are input into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. Battery anomaly warning is issued based on the target anomaly probability.

[0076] This invention, through its embodiments, determines physical core parameters based on collected charging data during vehicle battery charging. These physical core parameters include at least an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. These parameters are then input into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. Based on this target anomaly probability, a battery anomaly warning is issued. This invention, by determining physical core parameters based on collected charging data, can convert charging data into physically interpretable time-series core parameters without adding sensors. The input of the electrochemical differential voltage core, thermal inertia core, and current-voltage response core into the preset battery anomaly detection model allows for accurate determination of the target anomaly probability corresponding to the vehicle battery through a trained preset battery anomaly detection model. This early warning of battery anomalies based on the target anomaly probability reduces the risk of thermal runaway.

[0077] This invention provides a computer-readable storage medium storing at least one executable instruction that, when executed on a battery anomaly warning device / app, causes the battery anomaly warning device / app to perform the battery anomaly warning method in any of the above method embodiments.

[0078] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Furthermore, the embodiments of this invention are not directed to any particular programming language.

[0079] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. Similarly, for the sake of brevity and to aid in understanding one or more aspects of the invention, in the description of exemplary embodiments of the invention above, various features of the embodiments are sometimes grouped together in a single embodiment, figure, or description thereof. The claims, which follow the detailed description, are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0080] Those skilled in the art will understand that the modules in the device of the embodiment can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiment can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components, except that at least some of such features and / or processes or units are mutually exclusive.

[0081] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A battery anomaly early warning method, characterized in that, The method includes: During the vehicle battery charging process, physical core parameters are determined based on the collected charging data. The physical core parameters include at least: an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. The electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel are input into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. Battery anomaly warning is issued based on the target anomaly probability.

2. The method according to claim 1, characterized in that, During the vehicle battery charging process, the physical core parameters are determined based on the collected charging data, including: During the vehicle battery charging process, charging data is collected at preset power intervals. The charging data includes at least: voltage data, temperature data, and current data. The electrochemical differential physics core was determined based on the voltage data; The thermal inertial core is determined based on the temperature data; The current-voltage response kernel is determined based on the current data.

3. The method according to claim 2, characterized in that, The step of determining the electrochemical differential physics core based on the voltage data includes: The next voltage for the next charge and the current voltage for the current charge are determined based on the voltage data, and the voltage change is calculated based on the next voltage and the current voltage. Calculate the change in power based on the next power level and the current power level; The electrochemical differential physical nucleus is calculated based on the voltage change and the charge change.

4. The method according to claim 2, characterized in that, The step of determining the thermal inertial core based on the temperature data includes: Determine the actual temperature under the current power level based on the temperature data; The estimated temperature under the current power level is determined by a preset linear model; The thermal inertial core is calculated based on the difference between the actual temperature and the estimated temperature.

5. The method according to claim 2, characterized in that, The step of determining the current-voltage response kernel based on the current data includes: Determine the actual voltage under the current power level based on the current data; The estimated voltage under the current power level is determined by a preset nonlinear model; The current-voltage response kernel is calculated based on the difference between the actual voltage and the estimated voltage.

6. The method according to any one of claims 1-5, characterized in that, The step of inputting the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery includes: The electrochemical differential voltage kernel under each battery, the thermal inertia kernel under each thermometer, and the current-voltage response kernel under each time interval are respectively input into the encoding layer of the preset battery anomaly detection model for encoding, so as to obtain the electrochemical differential voltage kernel hidden vector, thermal inertia kernel hidden vector, and current-voltage response kernel hidden vector of the same dimension. Using the electrochemical differential voltage kernel latent vector as the query, and the thermal inertia kernel latent vector and the current-voltage response kernel latent vector as the key and value, the electrochemical differential voltage kernel latent vector, the thermal inertia kernel latent vector and the current-voltage response kernel latent vector are fused through the cross-attention mechanism in the coding layer to obtain fused features; The target score is obtained by linearly transforming the fused features through the linear layer in the preset battery anomaly detection model. Substituting the target score into a preset activation function yields the target anomaly probability corresponding to the vehicle battery.

7. The method as described in claim 6, characterized in that, Before inputting the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery, the method further includes: The initial sample corresponding to the vehicle battery is input into the initial battery anomaly detection model to obtain the initial anomaly probability corresponding to the vehicle battery; The loss value is calculated based on the initial anomaly probability and the label corresponding to the initial sample; The initial battery anomaly detection model is optimized based on the loss value to obtain a preset battery anomaly detection model.

8. A battery abnormality early warning device, characterized in that, The device includes: The data acquisition module is used to determine physical core parameters based on the collected charging data during the vehicle battery charging process. The physical core parameters include at least: an electrochemical differential voltage core, a thermal inertia core, and a current-voltage response core. The probability determination module is used to input the electrochemical differential voltage kernel, the thermal inertia kernel, and the current-voltage response kernel into a preset battery anomaly detection model to obtain the target anomaly probability corresponding to the vehicle battery. The battery anomaly warning module is used to issue a battery anomaly warning based on the target anomaly probability.

9. A battery anomaly early warning device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation of the battery anomaly warning method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores at least one executable instruction, which, when executed on the battery abnormality warning device / apparatus, causes the battery abnormality warning device / apparatus to perform the operation of the battery abnormality warning method as described in any one of claims 1-7.

Images