CN116087782B

CN116087782B - Automobile battery fault early warning method, system, device and storage medium

Info

Publication number: CN116087782B
Application number: CN202211395734.5A
Authority: CN
Inventors: 胡正波; 胡正伟
Original assignee: Suzhou Shoufan Electronic Technology Co ltd
Current assignee: Suzhou Shoufan Electronic Technology Co ltd
Priority date: 2022-11-09
Filing date: 2022-11-09
Publication date: 2024-02-02
Anticipated expiration: 2042-11-09
Also published as: CN116087782A

Abstract

The embodiment of the specification provides an automobile battery fault early warning method, system, device and storage medium, wherein the method comprises the following steps: collecting battery data in one or more periods in a preset observation period, wherein the battery data comprises terminal voltage and current of one or more battery units in a battery pack and temperature of one or more preset measurement points; determining battery performance over one or more cycles based on the battery data over the one or more cycles; determining a battery failure probability based on battery performance over one or more cycles; and generating fault early warning information in response to the probability of the battery fault meeting a preset condition.

Description

Automobile battery fault early warning method, system, device and storage medium

Technical Field

The present disclosure relates to the field of battery technologies, and in particular, to a method, a system, an apparatus, and a storage medium for early warning of battery failure of an automobile.

Background

The exacerbation of energy crisis and the rise in carbon emission reduction pressure have driven the technological innovation of power cells. The development and application of new energy automobiles closely related to the technical innovation of power batteries are also hot topics at present.

However, with the global popularization of new energy automobiles, the occurrence of fire events of numerous new energy automobiles in recent years has also raised public and enterprise widespread attention and thinking about the safety and reliability of power batteries. The safety of power cell systems is greatly affected by the numerous dangerous operations and abusive behaviors of the systems during actual use (such as thermal abuse, electrical abuse, mechanical abuse, etc.), and in some extreme cases, thermal runaway, cell rupture and explosion may even be induced.

In recent years, frequent ignition and explosion accidents of the vehicle-mounted power battery system indicate that the loss caused by the failure of the battery is very huge and heavy, and the use safety and reliability of the power battery system are not neglected.

Therefore, it is desirable to provide a method and system for early warning of battery failure in an automobile.

Disclosure of Invention

One or more embodiments of the present disclosure provide a method for early warning of a battery failure of an automobile. The method comprises the following steps: collecting battery data in one or more periods in a preset observation period, wherein the battery data comprises terminal voltage and current of one or more battery units in a battery pack and temperature of one or more preset measurement points; determining battery performance over the one or more periods based on the battery data over the one or more periods; determining a battery failure probability based on battery performance over the one or more cycles; and generating fault early warning information in response to the battery fault probability meeting a preset condition.

One or more embodiments of the present specification provide an automotive battery fault warning system. The system comprises an acquisition module, a first determination module, a second determination module and a generation module; the acquisition module is used for acquiring battery data in one or more periods in a preset observation period, wherein the battery data comprises terminal voltage and current of one or more battery units in the battery pack and temperature of one or more preset measurement points; the first determining module is configured to determine battery performance during the one or more periods based on the battery data during the one or more periods; the second determining module is configured to determine a battery failure probability based on battery performance in the one or more periods; the generation module is used for responding to the battery fault probability to meet a preset condition and generating fault early warning information.

One or more embodiments of the present disclosure provide an automotive battery fault warning device, including a processor for executing an automotive battery fault warning method.

One or more embodiments of the present specification provide a computer-readable storage medium storing computer instructions that, when read by a computer in the storage medium, perform a method for warning of a failure of a battery of an automobile.

Drawings

The present specification will be further elucidated by way of example embodiments, which will be described in detail by means of the accompanying drawings. The embodiments are not limiting, in which like numerals represent like structures, wherein:

FIG. 1 is a block diagram of an automotive battery fault warning system according to some embodiments of the present disclosure;

FIG. 2 is an exemplary flow chart of an automotive battery fault warning method according to some embodiments of the present disclosure;

FIG. 3 is a schematic illustration of determining battery performance and battery failure probability according to some embodiments of the present disclosure;

FIG. 4 is a block diagram of a model of attenuation level determination according to some embodiments of this disclosure;

fig. 5 is a block diagram of a failure probability determination model shown in accordance with some embodiments of the present description.

Detailed Description

In order to more clearly illustrate the technical solutions of the embodiments of the present specification, the drawings that are required to be used in the description of the embodiments will be briefly described below. It is apparent that the drawings in the following description are only some examples or embodiments of the present specification, and it is possible for those of ordinary skill in the art to apply the present specification to other similar situations according to the drawings without inventive effort. Unless otherwise apparent from the context of the language or otherwise specified, like reference numerals in the figures refer to like structures or operations.

It will be appreciated that "system," "apparatus," "unit" and/or "module" as used herein is one method for distinguishing between different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

As used in this specification and the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.

A flowchart is used in this specification to describe the operations performed by the system according to embodiments of the present specification. It should be appreciated that the preceding or following operations are not necessarily performed in order precisely. Rather, the steps may be processed in reverse order or simultaneously. Also, other operations may be added to or removed from these processes.

Fig. 1 is a block diagram of an automotive battery fault warning system according to some embodiments of the present disclosure.

In some embodiments, the automotive battery fault early warning system 100 may include an acquisition module 110, a first determination module 120, a second determination module 130, and a generation module 140.

The acquisition module 110 is configured to acquire battery data in one or more periods of a preset observation period, where the battery data includes terminal voltages and currents of one or more battery cells in the battery pack, and temperatures of one or more preset measurement points. For relevant content of collecting battery data, please refer to the corresponding description of fig. 2.

The first determination module 120 is configured to determine battery performance over one or more cycles based on battery data over one or more cycles. The battery performance includes a battery degradation predicted by a degradation determination model that is a machine learning model that includes a first embedded layer and a first determination layer. For details regarding determining the battery performance and the degradation determination model, please refer to the corresponding descriptions of fig. 3 and fig. 4.

In some embodiments, the second determination module 130 is configured to determine a battery failure probability based on battery performance over one or more cycles. The second determination model 130 may also process the battery performance over the one or more cycles to determine the battery failure probability through a failure probability determination model. The fault probability determining model is a machine learning model and comprises a second embedding layer and a second determining layer. The battery fault probability is related to a driving state, the input of the fault probability determination model comprises the driving state, and the driving state comprises one or more of speed, steering, vehicle-mounted weight and power consumption in a vehicle when the vehicle runs. For the description of determining the probability of battery failure and the failure probability determination model, reference is made to fig. 3 and 5.

The generating module 140 is configured to generate fault early warning information in response to the probability of the battery fault meeting a preset condition. For relevant content in generating fault warning information, please refer to the relevant description of fig. 2.

It should be understood that the system shown in fig. 1 and its modules may be implemented in a variety of ways. It should be noted that the foregoing description of the vehicle battery fault warning system and the modules thereof is merely for convenience of description and is not intended to limit the present disclosure to the scope of the illustrated embodiments. It will be appreciated by those skilled in the art that, given the principles of the system, various modules may be combined arbitrarily or a subsystem may be constructed in connection with other modules without departing from such principles. In some embodiments, the acquisition module 110, the first determination module 120, the second determination module 130, and the generation module 140 disclosed in fig. 1 may be different modules in one system, or may be one module to implement the functions of two or more modules described above. For example, each module may share one memory module, or each module may have a respective memory module. Such variations are within the scope of the present description.

Fig. 2 is an exemplary flow chart of a method of warning of a vehicle battery fault according to some embodiments of the present disclosure. The flow 200 includes steps 210, 220, 230, and 240, which are performed by the acquisition module 110, the first determination module 120, the second determination module 130, and the generation module 140, respectively.

Step 210, collecting battery data during one or more of a preset observation period.

The preset observation period refers to a time period set according to actual conditions, and the time period is used as a time reference for data acquisition, data processing or other operations. For example, data is collected every 10 minutes with 10 minutes as a preset observation period, and the battery performance of the current period is calculated once. In some embodiments, one or more of the preset observation periods may be the consecutive one or more periods closest to the current time.

Battery data refers to data related to the performance of a battery. In some embodiments, the battery data includes terminal voltages and currents of one or more battery cells within the battery pack, temperatures of one or more preset measurement points, or other related data.

The battery data acquisition method comprises the steps of acquiring through various sensors or intelligent battery monitoring chips. Wherein the type of sensor includes a temperature sensor, a hall current sensor, or other related sensor.

Step 220, determining battery performance over one or more cycles based on the battery data over the one or more cycles.

Battery performance is data that characterizes the current operating condition of a battery and may be used to account for comparisons of the rated operating state of the battery over one or more cycles and under the same conditions. Rated operating conditions include, but are not limited to, rated capacity, rated voltage, charge-discharge rate, impedance, lifetime, and self-discharge rate.

In some embodiments, the battery performance includes a proximity of the collected voltage, current, etc. parameter to a nominal parameter (voltage, current of a standard reference) under the same conditions (temperature, vehicle state, etc.), based on which the battery performance is described.

In some embodiments, one or more battery data vectors may be determined based on battery data over one or more cycles, while a nominal parameter vector is determined based on battery nominal parameters under the same environmental conditions (temperature, vehicle state, battery usage), and battery performance over one or more cycles is determined based on the vector distance of the one or more battery data vectors from the nominal parameter vector, respectively. In some embodiments, battery performance may be inversely related to the vector distance of the battery data. For example, the greater the vector distance value of the battery data vector, the greater the difference between the battery data in the one or more cycles and the battery rating parameter for the same environmental condition, the lower the corresponding battery performance.

In some embodiments, battery performance may be determined by mathematical calculations (e.g., calculating their similarity to nominal parameters under equivalent conditions) or machine learning models, for more relevant description see the corresponding description of fig. 3.

At step 230, a probability of battery failure is determined based on the battery performance over one or more cycles.

The battery failure probability refers to the probability of battery failure in one or more cycles. For example, the probability of occurrence of a failure of the battery in one cycle is 80%.

In some embodiments, the performance vector may be constructed based on battery performance over one or more cycles. The elements of each dimension in the performance vector correspond to one cycle of battery performance. The historical performance vector may be determined based on battery performance during one or more historical periods prior to the historical fault. The battery failure probability may be determined based on a vector distance of the performance vector from the historical performance vector. The battery failure probability may be inversely related to the vector distance of the performance vector. For example, under the same conditions, the smaller the vector distance of the performance vector, the greater the battery failure probability.

The probability of battery failure may be determined based on battery performance over one or more cycles, by manual rules, mathematical calculations, machine learning model predictions, or any other feasible manner, and further description of the determination may be found in relation to fig. 3 or fig. 5.

And step 240, generating fault early warning information in response to the probability of the battery fault meeting the preset condition.

The preset condition may be that the battery failure probability is greater than a preset threshold. For example, whether the probability of occurrence of the fault of the battery in one or more periods is greater than 60% may be set as a criterion of the preset condition, and if the probability of occurrence of the fault is greater than 60%, the preset condition is satisfied, otherwise, the preset condition is not satisfied.

The fault early warning information is prompt information indicating that the battery is likely to be in fault. The fault early warning information can comprise the probability of the battery to fail, the type of the fault, fault emergency prompt and the like.

In some embodiments, the fault warning information may be generated by a processor and presented via a mobile terminal, vehicle terminal, or other device coupled to the vehicle. For example, the vehicle is prompted by any combination of vibration, text, voice and the like through a mobile phone, a vehicle-mounted large screen and a vehicle-mounted sound of the vehicle owner.

In some embodiments of the present disclosure, by collecting battery data, obtaining battery performance and predicting battery failure, the battery operation condition may be effectively monitored, the risk of sudden battery failure is greatly reduced, and a battery failure prevention mechanism is established.

Fig. 3 is a schematic diagram illustrating determination of battery performance and battery failure probability according to some embodiments of the present description. A related description of battery performance may be found in relation to fig. 2.

In some embodiments, battery performance may be determined based on how close the monitored voltage, current, etc. parameters are to rated parameters (standard voltage and current parameters) under comparable conditions (same temperature, vehicle state, etc.) over one or more cycles. The greater the proximity, the better the cell performance.

In some embodiments, battery performance may include battery degradation.

Battery degradation may refer to data used to characterize the degree of battery performance degradation. For example, a battery degradation of 10% represents a 10% degradation of the current performance of the battery compared to the standard performance of the battery, i.e., a 90% degradation of the current performance of the battery.

As shown in fig. 3, in flow 300, a battery attenuation 331 for one or more cycles may be determined by attenuation determination model 320 based on battery data 311 for the one or more cycles. A related description of battery data over one or more cycles may be found in connection with fig. 2. A related description of the attenuation degree determination model can be found in fig. 4.

In some embodiments, battery performance may also include cell uniformity.

For example, a cell uniformity of greater than 80% may represent a low degree of dispersion in the SOC of each cell in the battery, and a cell uniformity of less than 50% may represent a high degree of dispersion in the SOC of each cell in the battery.

As shown in fig. 3, in some embodiments, the SOC values of each battery cell in the battery pack at multiple times during one or more cycles may be obtained based on the battery data 311 during the one or more cycles, thereby determining the battery cell uniformity 332 during the one or more cycles.

The SOC value of each battery cell in the battery pack in one cycle may be determined in various ways. For example, in one cycle, the SOC value of each battery cell at a plurality of times is monitored; calculating the degree of dispersion of the SOC of each battery unit at each moment; the degree of dispersion of the SOC of each battery cell at a plurality of times is averaged to obtain the degree of dispersion of the SOC of each battery cell in the battery pack in the cycle. Wherein, a plurality of moments can be preset. The degree of dispersion can be determined by calculation methods such as polar difference, average difference and standard deviation.

The data value of the cell uniformity may be inversely related to the degree of dispersion of the SOC of each cell within the battery pack. Battery performance may be positively correlated with cell uniformity. For example, under the same conditions, the higher the cell uniformity, the higher the battery performance.

In some embodiments, a weighted average may be used when averaging the degree of dispersion of the SOC of each battery cell at a plurality of times. The weight value of the degree of dispersion of the SOC of the respective battery cells at each time may be correlated with the running state of the vehicle at that time, for example, may be positively correlated with the in-vehicle electric power consumption in the running state, or the like. For further description of the driving situation, see fig. 5 for a description.

In some embodiments of the present disclosure, by introducing a degree of cell consistency into the battery performance data, the battery performance data may be made to more closely reflect the battery performance; under different driving states, the electric quantity change rates of the battery units are different, the change rates are too fast, and the theoretical error is larger, so that the characteristic is introduced as a weight, and the determined consistency of the battery units can be more accurate.

In some embodiments, battery performance may also include outliers.

The abnormal value may refer to data reflecting the occurrence of an abnormal event of the battery. The abnormal event may include, but is not limited to, a charge jump, a rapid drop in charge, etc. In some embodiments, the outliers may be represented based on statistics of the occurrence times, frequency, etc. of the outliers over one or more periods. Battery performance may be inversely related to the magnitude of the outlier. For example, under the same conditions, the larger the abnormal value, the lower the battery performance.

As shown in FIG. 3, in some embodiments, outliers 333 may be determined over one or more periods based on the frequency of occurrence 312 of the outlier event over the one or more periods. For example, the number of occurrence of an abnormal event in one cycle is 5, and then the abnormal value of the cycle is 5.

In some embodiments, the degree to which different anomaly events contribute to an anomaly value may be different. For example, the outliers may be determined by a weighted summation of the outliers, the weights being related to the severity of the outliers. For example, for a charge skip event, the power may be preset to skip from (80%, 90% ] to 100% with a weight value of 0.8, from (70%, 80% ] to 100% with a weight value of 0.9, from (60%, 70% ] to 100% with a weight value of 1.0, from (50%, 60% ] to 100% with a weight value of 1.1, and so on.

In some embodiments of the present disclosure, by introducing outliers into the battery performance data, the battery performance data may be made to more closely reflect the battery performance conditions; by setting the contribution degree of the abnormal event to the abnormal value, the determined abnormal value can be more in line with the actual requirement.

In some embodiments of the present disclosure, by introducing a battery degradation into the battery performance data, the battery performance data may be made to more closely reflect the battery performance; the battery attenuation degree is determined through the model, so that the efficiency of determining work is improved, and meanwhile, the accuracy of a determining result is improved.

As shown in fig. 3, in some embodiments, battery performance 330 over one or more periods corresponding to battery data 311 over the one or more periods may include battery degradation 331 over the one or more periods, battery cell uniformity 332 over the one or more periods, and outlier 333 over the one or more periods.

As shown in fig. 3, in some embodiments, a battery failure probability 350 may be determined by a failure probability determination model 340 based on battery performance 330 over one or more cycles. A description of the failure probability determination model is described with reference to fig. 5.

In some embodiments of the present disclosure, the probability of battery failure is determined by a model, which improves the efficiency of the determination work and also improves the accuracy of the probability determination result.

FIG. 4 is a block diagram of a model of attenuation level determination according to some embodiments of this disclosure.

In some embodiments, battery data for a period may be processed based on a degradation determination model to determine a battery degradation for the period. The degradation determination model may refer to a machine learning model for determining a battery degradation. In some embodiments, the attenuation degree determination model may include any one or combination of various possible models, including a recurrent neural network (Recurrent Neural Network, RNN) model, a deep neural network (Deep Neural Network, DNN) model, a convolutional neural network (Convolutional Neural Network, CNN) model, and the like.

In some embodiments, the attenuation degree determination model may include a plurality of processing layers. As shown in fig. 4, in structure 400, attenuation degree determination model 430 may include a first embedding layer 431 and a first determination layer 432.

The first embedding layer 431 may be configured to process the battery data 410 in a period and determine the first feature vector 440 corresponding to the period.

The first feature vector may refer to a feature vector constructed based on battery data. For example, the first feature vector of a certain period may be (13.5, 13.6,) 13.4,2.12,2.05,) 2.16, 30.4, 31.0,) 30.8, representing the meaning that the terminal voltages of the plurality of battery cells are 13.5V,13.6V, respectively, 13.4V, the currents of the plurality of battery cells are 2.12a,2.05a, respectively, 2.16A, the temperatures of the plurality of preset measurement points are 30.4 ℃,31.0 ℃, respectively.

The first determining layer 432 may be configured to process the first feature vector 440 corresponding to a period and determine the battery degradation 450 in the period. A related description of the degree of battery degradation can be found in fig. 2.

In some embodiments, the input to the first determination layer 432 may also include the current number of loops 420. The current cycle number 420 may include the number of times of the normal charge process and the normal discharge process of the battery pack, and the number of times of overcharge and overdischarge, which are accumulated from when the battery leaves the factory to when the battery is used at the current time, corresponding to the first feature vector 440.

In some embodiments of the present disclosure, by inputting the current cycle number, the attenuation degree of the battery output by the first determining layer may be made to more closely reflect the actual situation of the battery.

In some embodiments, the attenuation degree determination model may be obtained through joint training. The sample data for the joint training may include battery data and its corresponding current number of cycles of the sample over a plurality of sample periods, and the tag includes an actual battery decay over the plurality of sample periods. The multiple sample periods may be manually selected. The battery data in the plurality of sample periods and the corresponding sample current cycle times thereof, and the actual battery attenuation corresponding to the battery data in the plurality of sample periods can be obtained based on the relevant data of the battery history use record. The tag may be determined based on manual labeling.

An exemplary joint training process includes: inputting battery data in a plurality of sample periods into an initial first embedding layer to obtain a first feature vector output by the initial first embedding layer; the first feature vector output by the initial first embedding layer is used as training sample data, and is input into the first determining layer together with the current cycle times of samples corresponding to the battery data in a plurality of sample periods, so that the battery attenuation degree in a plurality of periods output by the initial first determining layer is obtained; the parameters of the initial first embedding layer and the initial first determination layer are updated synchronously based on the actual battery degradation over a plurality of sample periods and the battery degradation over a plurality of periods output by the initial first determination layer to construct a loss function. And when the loss function meets the preset condition, model training is completed, and a trained attenuation degree determination model is obtained. The preset condition may be that the loss function converges, the iteration number reaches an iteration number threshold, and the like.

In some embodiments of the present disclosure, the battery attenuation degree is determined through the model, so that accuracy of a determination result can be ensured, prediction efficiency is improved, and time cost is saved. Meanwhile, the accuracy of the output result of the attenuation degree determination model can be effectively improved by carrying out joint training on a plurality of processing layers of the attenuation degree determination model.

In some embodiments, battery performance over one or more cycles may be processed based on a probability of failure determination model to determine a probability of battery failure. The failure probability determination model may refer to a machine learning model for determining a failure probability of the battery. In some embodiments, the failure probability determination model may include any one or combination of various possible models, including a recurrent neural network model, a deep neural network model, a convolutional neural network model, and so forth.

In some embodiments, the failure probability determination model may include multiple processing layers. As shown in fig. 5, in the structure 500, the failure probability determination model 540 may include a second embedded layer 541, a third embedded layer 542, and a second determination layer 543.

The second embedded layer 541 may be configured to process the battery performance 510 over one or more periods to determine a second feature vector 550 corresponding to the one or more periods.

The second feature vector may refer to a feature vector constructed based on battery performance. For example, a certain second feature vector may be (88%, 84%, and/or the like, 82%,3.4,3.9, and/or the like, and 4.1, which means that the cell uniformity of the battery pack for a plurality of cycles is 88%,84%, and the like, and the abnormal value of the battery pack for a plurality of cycles is 3.4,3.9, and/or the like, and 4.1, respectively.

The third embedded layer 542 may be used to process the driving status 520 to determine a third feature vector 560.

The driving state may refer to a current driving state of the automobile, and may include an on-vehicle weight of the automobile when the automobile is running, a driving speed, a steering angle, a total power consumption of an in-vehicle facility (for example, an air conditioner), and the like at a plurality of consecutive times. For example, the content of a certain driving state may be (213, 75, 80, 77, 10, 70,8, 1500, 1250, 1300), which means that the vehicle weight is 213 when the vehicle is running, the running speeds at a plurality of times are 75km/h,80km/h and 77km/h, respectively, the steering angles at a plurality of times are 10 °,70 ° and 8 °, respectively, and the total power consumption of the facilities in the vehicle at a plurality of times is 1500w, 1250w and 1300w, respectively. The time intervals of the plurality of successive moments can be preset, for example, 10 seconds.

In some embodiments, the driving state may include a driving state of a current time period and a driving state of a future time period. The current time period may be preset, for example, the first 1 minute of the current time; the future time period may also be preset, for example, the last 2 minutes of the current time.

In some embodiments, the driving status of the future time period may be obtained by prediction. For example, the driving state of the future time period may be determined by processing the driving state of the current time period based on the driving state prediction model. The driving state prediction model may be a machine learning model, for example, a time series model, or the like. The input of the driving state prediction model may be the driving state of the current time period, and the output may be the driving state of the future time period.

In some embodiments, the driving state prediction model may be trained from a plurality of labeled training samples. For example, a plurality of training samples with labels may be input into the initial driving state prediction model, a loss function may be constructed from the labels and the results of the initial driving state prediction model, and parameters of the initial driving state prediction model may be iteratively updated based on the loss function. And when the loss function of the initial driving state prediction model meets the preset condition, model training is completed, and a trained driving state prediction model is obtained. The preset condition may be that the loss function converges, the number of iterations reaches a threshold value, etc.

In some embodiments, the training samples may include at least a plurality of driving states of historical time periods. The tag may be a driving state of a period after a history period corresponding to the history period. The tag may be obtained based on manual labeling.

In some embodiments of the present disclosure, by introducing a driving state, accuracy of the battery fault probability determined by the model may be made higher. By further introducing the future driving state, the driving situation of the vehicle can be reflected better, and the accuracy of model output can be improved.

The third feature vector may refer to a feature vector constructed based on the driving state. For example, a third feature vector may be ([ 213, 75, 80, 77, 10, 70,8, 1500, 1250, 1300], [213, 81, 78, 80,5,6, 20, 1400, 1350, 1380 ]), representing a driving state of the current time period as "vehicle weight 213, driving speeds of 75km/h,80km/h, and 77km/h, respectively, steering angles of 10 °,70 °, and 8 °, respectively, total power consumption of an in-vehicle facility (e.g., an air conditioner) of the plurality of continuous times as 1500w, 1250w, and 1300w", respectively, and driving states of the future time period (e.g., within 1 minute) as "vehicle weight 213," driving speeds of the plurality of continuous times as 81km/h, 78km/h, and 80km/h, respectively, steering angles of the plurality of continuous times as 5 °,6 °, and 20 °, respectively, total power consumption of the in-vehicle facility of the plurality of continuous times as 1400w, 1350w, and 1380w, respectively).

The second determination layer 543 may be configured to process the second feature vector 550, the third feature vector 560, and the abnormal frequent item overlap ratio 530 to determine the battery failure probability 570.

The abnormal frequent item may indicate an abnormal event whose occurrence times are frequent. The outlier frequent term may be determined based on the frequency of occurrence of the outlier term. For example, an abnormal item whose frequency of occurrence is higher than a frequent threshold may be determined as an abnormal frequent item, and the frequent threshold may be preset. For example, if the frequency threshold is preset to 0.9, according to the history data, "3 consecutive charge jumps, 2 consecutive power drops", then "3 consecutive charge jumps, 2 consecutive power drops", and then the occurrence frequency of "0.93 consecutive power drops" occur before 93 battery failures in 100 battery failure events, then the abnormal item may be determined as the abnormal frequent item.

The degree of coincidence of the abnormal frequent item may refer to the degree of closeness of the abnormal event to the abnormal frequent item. For example, the current abnormal event is "4 continuous charging jumps, 4 continuous power drops", the abnormal frequent item is "3 continuous charging jumps, 2 continuous power drops", and then the abnormal frequent item overlap ratio of the abnormal event is (3+2)/(4+4) =0.625.

In some embodiments, the failure probability determination model may be obtained by joint training. The sample data of the joint training may include battery performance, sample driving state and sample abnormal frequent item coincidence in one or more sample periods, and the label includes a sample battery fault condition, and may indicate yes or no occurrence of battery fault by 1 and 0, respectively. The sample battery fault condition may be a battery fault condition of one cycle or a battery fault condition of a plurality of continuous cycles. For example, if a battery failure occurs within one sample period, the label is 1, otherwise the label is 0; for another example, if a battery failure occurs in some or all of the sample periods, the flag is 1, and if no battery failure occurs in all of the sample periods, the flag is 0. In some embodiments, the one or more sample periods corresponding to the sample data may be the sample period closest to the current time. The multiple sample periods may be manually selected. The battery performance, sample driving conditions, and sample abnormal frequent item overlap for one or more sample periods may all be obtained based on the battery historical usage record and the related data of the historical driving records. The tag may be determined based on manual labeling.

An exemplary joint training process includes: inputting the battery performance in one or more sample periods into an initial second embedded layer to obtain a second feature vector output by the initial second embedded layer; inputting the driving state of the sample into an initial third embedded layer to obtain a third feature vector output by the initial third embedded layer; taking the second feature vector output by the initial second embedding layer and the third feature vector output by the initial third embedding layer as training sample data, and inputting the training sample data and the coincidence degree of abnormal frequent items of the samples into a second determining layer to obtain the fault probability of the battery output by the initial second determining layer; and constructing a loss function based on the sample battery fault condition and the battery fault probability output by the initial second determination layer, and synchronously updating parameters of the initial second embedding layer, the initial third embedding layer and the initial second determination layer. And when the loss function meets the preset condition, model training is completed, and a trained fault probability determination model is obtained. The preset condition may be that the loss function converges, the iteration number reaches an iteration number threshold, and the like.

In some embodiments of the present disclosure, the battery fault probability is determined through the model, so that accuracy of a determination result can be ensured, prediction efficiency is improved, and time cost is saved. Meanwhile, the accuracy of the output result of the fault probability determination model can be effectively improved by carrying out joint training on a plurality of processing layers of the fault probability determination model.

While the basic concepts have been described above, it will be apparent to those skilled in the art that the foregoing detailed disclosure is by way of example only and is not intended to be limiting. Although not explicitly described herein, various modifications, improvements, and adaptations to the present disclosure may occur to one skilled in the art. Such modifications, improvements, and modifications are intended to be suggested within this specification, and therefore, such modifications, improvements, and modifications are intended to be included within the spirit and scope of the exemplary embodiments of the present invention.

Meanwhile, the specification uses specific words to describe the embodiments of the specification. Reference to "one embodiment," "an embodiment," and/or "some embodiments" means that a particular feature, structure, or characteristic is associated with at least one embodiment of the present description. Thus, it should be emphasized and should be appreciated that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various positions in this specification are not necessarily referring to the same embodiment. Furthermore, certain features, structures, or characteristics of one or more embodiments of the present description may be combined as suitable.

Furthermore, the order in which the elements and sequences are processed, the use of numerical letters, or other designations in the description are not intended to limit the order in which the processes and methods of the description are performed unless explicitly recited in the claims. While certain presently useful inventive embodiments have been discussed in the foregoing disclosure, by way of various examples, it is to be understood that such details are merely illustrative and that the appended claims are not limited to the disclosed embodiments, but, on the contrary, are intended to cover all modifications and equivalent arrangements included within the spirit and scope of the embodiments of the present disclosure. For example, while the system components described above may be implemented by hardware devices, they may also be implemented solely by software solutions, such as installing the described system on an existing server or mobile device.

Likewise, it should be noted that in order to simplify the presentation disclosed in this specification and thereby aid in understanding one or more inventive embodiments, various features are sometimes grouped together in a single embodiment, figure, or description thereof. This method of disclosure, however, is not intended to imply that more features than are presented in the claims are required for the present description. Indeed, less than all of the features of a single embodiment disclosed above.

In some embodiments, numbers describing the components, number of attributes are used, it being understood that such numbers being used in the description of embodiments are modified in some examples by the modifier "about," approximately, "or" substantially. Unless otherwise indicated, "about," "approximately," or "substantially" indicate that the number allows for a 20% variation. Accordingly, in some embodiments, numerical parameters set forth in the specification and claims are approximations that may vary depending upon the desired properties sought to be obtained by the individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and employ a method for preserving the general number of digits. Although the numerical ranges and parameters set forth herein are approximations that may be employed in some embodiments to confirm the breadth of the range, in particular embodiments, the setting of such numerical values is as precise as possible.

Each patent, patent application publication, and other material, such as articles, books, specifications, publications, documents, etc., referred to in this specification is incorporated herein by reference in its entirety. Except for application history documents that are inconsistent or conflicting with the content of this specification, documents that are currently or later attached to this specification in which the broadest scope of the claims to this specification is limited are also. It is noted that, if the description, definition, and/or use of a term in an attached material in this specification does not conform to or conflict with what is described in this specification, the description, definition, and/or use of the term in this specification controls.

Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments of this specification. Other variations are possible within the scope of this description. Thus, by way of example, and not limitation, alternative configurations of embodiments of the present specification may be considered as consistent with the teachings of the present specification. Accordingly, the embodiments of the present specification are not limited to only the embodiments explicitly described and depicted in the present specification.

Claims

1. A method for early warning of a battery failure of an automobile, the method comprising:

collecting battery data in one or more periods in a preset observation period, wherein the battery data comprises terminal voltage and current of one or more battery units in a battery pack and temperature of one or more preset measurement points;

determining battery performance over the one or more periods based on the battery data over the one or more periods;

determining a battery failure probability based on battery performance over the one or more cycles;

the determining a battery failure probability based on the battery performance over the one or more cycles includes: processing the battery performance and the abnormal frequent item coincidence degree in one or more periods through a fault probability determining model to determine the fault probability of the battery;

the coincidence degree of the abnormal frequent items is determined based on the similarity degree of the abnormal event and the abnormal frequent items; the abnormal frequent item is an abnormal event indicating frequent occurrence times;

generating fault early warning information in response to the battery fault probability meeting a preset condition;

the battery performance comprises battery attenuation degree, battery unit consistency degree and abnormal value;

the battery attenuation degree is obtained through prediction of an attenuation degree determination model, the attenuation degree determination model is a machine learning model, and the attenuation degree determination model comprises a first embedded layer and a first determination layer; the input of the first embedded layer comprises the battery data in one period, and the output comprises a first feature vector; the input of the first determining layer comprises the first characteristic vector and the current cycle number, and the output comprises the battery attenuation degree in the period;

the determining battery performance over the one or more cycles based on the battery data over the one or more cycles further comprises: monitoring SOC values of each battery cell at a plurality of moments based on the battery data in the one or more periods; calculating the discrete degree of the SOC value of each battery unit at a plurality of moments; taking the average value or weighted average value of the discrete degree of the SOC value of each battery unit at the plurality of moments as the discrete degree of the SOC value of each battery unit in the battery pack at the plurality of moments in the one or more periods, and determining the consistency of the battery units; the degree of dispersion of the SOC value of each battery cell in the battery pack at a plurality of times in the one or more periods is related to the running state of the vehicle at the corresponding time;

the outliers within the one or more periods are determined based on a weighted determination of the frequency of occurrence of the anomaly event within the one or more periods.

2. The method of claim 1, wherein the failure probability determination model is a machine learning model, the failure probability determination model comprising a second embedding layer and a second determination layer.

3. The method of claim 2, wherein the battery failure probability is related to a driving state, wherein the input of the failure probability determination model includes the driving state including one or more of a speed, a steering, a vehicle weight, and an in-vehicle power consumption of the vehicle while the vehicle is traveling.

4. The automobile battery fault early warning system comprises an acquisition module, a first determination module, a second determination module and a generation module;

the acquisition module is used for acquiring battery data in one or more periods in a preset observation period, wherein the battery data comprises terminal voltage and current of one or more battery units in the battery pack and temperature of one or more preset measurement points;

the first determining module is configured to determine battery performance during the one or more periods based on the battery data during the one or more periods;

the first determining module is further configured to: predicting the battery attenuation degree through an attenuation degree determination model, wherein the attenuation degree determination model is a machine learning model and comprises a first embedded layer and a first determination layer; the input of the first embedded layer comprises the battery data in one period, and the output comprises a first feature vector; the input of the first determining layer comprises the first characteristic vector and the current cycle number, and the output comprises the battery attenuation degree in the period;

monitoring SOC values of each battery cell at a plurality of moments based on the battery data in the one or more periods; calculating the discrete degree of the SOC value of each battery unit at a plurality of moments; taking the average value or weighted average value of the discrete degree of the SOC value of each battery unit at the plurality of moments as the discrete degree of the SOC value of each battery unit in the battery pack at the plurality of moments in the one or more periods, and determining the consistency of the battery units; the degree of dispersion of the SOC value of each battery cell in the battery pack at a plurality of times in the one or more periods is related to the running state of the vehicle at the corresponding time;

determining the outliers in the one or more periods by weighting based on the frequency of occurrence of the outliers in the one or more periods;

the second determining module is configured to determine a battery failure probability based on battery performance in the one or more periods;

the generation module is used for responding to the battery fault probability to meet a preset condition and generating fault early warning information.

5. The system of claim 4, wherein the failure probability determination model is a machine learning model, the failure probability determination model comprising a second embedding layer and a second determination layer.

6. The system of claim 5, wherein the probability of battery failure is related to a driving condition, wherein the input to the failure probability determination model comprises the driving condition comprising one or more of a speed, a steering, a weight on board, and a power consumption in the vehicle when the vehicle is driving.

7. An automotive battery fault warning device comprising a processor for executing the automotive battery fault warning method of any one of claims 1 to 3.

8. A computer-readable storage medium storing computer instructions that, when read by a computer in the storage medium, the computer performs the car battery failure warning method according to any one of claims 1 to 3.