Battery risk determination method and apparatus, electronic device, and readable storage medium

Abstract

The embodiment of the application provides a battery risk determination method and device, electronic equipment and readable storage medium, and relates to the technical field of computers. The embodiment of the application can obtain flow data generated by a target battery in a charging process, perform data filling on the flow data based on historical data corresponding to the target battery, determine target data corresponding to the flow data, and then determine a risk level corresponding to the target battery through the target data and a pre-trained battery risk determination model. Since the embodiment of the application can determine battery parameters not contained in the flow data in the historical data of the target battery to maximize the simulation of the state of the target battery in the current charging process, the embodiment of the application can ensure the real-time and accuracy of determining the battery risk at the same time.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, electronic device, and readable storage medium for determining battery risk. Background Technology

[0002] As a crucial component of the new energy sector, battery safety is a primary concern. In related technologies, battery parameters are typically used to determine whether any safety hazards exist.

[0003] However, since some battery parameters (such as battery capacity retention rate and depth of charge) can usually only be obtained after the battery has finished charging, there is a lag in obtaining these parameters, leading to delays or low accuracy in determining battery risks. In other words, ensuring both real-time performance and accuracy in determining battery risks is a pressing issue that needs to be addressed. Summary of the Invention

[0004] In view of this, embodiments of this application provide a battery risk determination method, apparatus, electronic device, and readable storage medium to simultaneously ensure the real-time nature and accuracy of battery risk determination.

[0005] Firstly, a method for determining battery risk is provided, the method comprising:

[0006] Acquire streaming data generated by the target battery during the charging process.

[0007] In response to the presence of missing data in the streaming data, the streaming data is filled with data based on the historical data corresponding to the target battery to determine the target data corresponding to the streaming data.

[0008] Based on the target data and the pre-trained battery risk determination model, the risk level corresponding to the target battery is determined.

[0009] In some embodiments, the step of filling the stream data with historical data corresponding to the target battery to determine the target data corresponding to the stream data includes:

[0010] Determine the types of missing data in the historical data and streaming data corresponding to the target battery.

[0011] Data fitting processing is performed on the portions of the historical data corresponding to the missing data types to determine the target filling data.

[0012] The stream data is populated with the target data to determine the target data corresponding to the stream data.

[0013] In some embodiments, the method further includes:

[0014] Obtain a training set, which includes at least multiple training samples, battery risk parameters corresponding to each training sample, and feature processing identifiers corresponding to each training sample. The feature processing identifiers are used to identify the feature extraction method used during feature extraction.

[0015] Based on the feature extraction method corresponding to the feature processing identifier, feature extraction is performed on each of the training samples to determine the sample feature set.

[0016] The battery risk determination model is trained by using the features in the sample feature set as input and the battery risk parameters corresponding to each training sample as training labels.

[0017] In some embodiments, the training samples include historical data corresponding to the target battery.

[0018] The method further includes:

[0019] Determine the feature subset corresponding to the historical data from the sample feature set.

[0020] Based on the correspondence between each feature in the feature subset and historical data, a feature mapping table corresponding to the target battery is generated.

[0021] In some embodiments, determining the risk level of the target battery based on the target data and a pre-trained battery risk determination model includes:

[0022] Based on the target data and the feature mapping table corresponding to the target battery, the target features corresponding to the target data are determined.

[0023] The target features are input into a pre-trained battery risk determination model to determine the target risk parameters corresponding to the target battery.

[0024] Based on the target risk parameters, the risk level corresponding to the target battery is determined.

[0025] In some embodiments, obtaining the training set includes:

[0026] Based on a pre-set ratio of positive to negative samples, positive and negative samples are extracted from the full feature factor library to form the training set.

[0027] In some embodiments, the method further includes:

[0028] In response to the completion of charging of the target battery, a set of single charging parameters corresponding to the target battery is generated.

[0029] Determine the battery risk parameters and feature processing identifiers corresponding to the single charging parameter set.

[0030] The set of single-charge parameters, the battery risk parameters corresponding to the set of single-charge parameters, and the feature processing identifier are stored in the full feature factor library.

[0031] In some embodiments, the training samples are a set of single-charge parameters corresponding to a single charge of the same battery. The feature processing identifier includes multiple sub-identifiers. The sub-identifiers are used to identify the feature extraction method used for each parameter in the single-charge parameter set during feature extraction. The sub-identifiers include one or more of the following combinations: Z-score identifier, binarization processing identifier, and equidistant bin discretization identifier.

[0032] In some embodiments, the target battery is a power battery in an electric vehicle, and the target data includes one or more of the following combinations: battery usage behavior parameters, single charge data parameters, time-domain variation parameters, safety hazard and fault perception parameters, and battery management system fault parameters.

[0033] Secondly, a battery risk determination device is provided, the device comprising:

[0034] The streaming data acquisition module is configured to acquire streaming data generated by the target battery during the charging process.

[0035] The data filling module is configured to respond to the presence of missing data in the streaming data by filling the streaming data with data based on historical data corresponding to the target battery, and to determine the target data corresponding to the streaming data.

[0036] The risk level determination module is configured to determine the risk level corresponding to the target battery based on the target data and a pre-trained battery risk determination model.

[0037] In some embodiments, the data population module is specifically configured to perform:

[0038] Determine the types of missing data in the historical data and streaming data corresponding to the target battery.

[0039] Data fitting processing is performed on the portions of the historical data corresponding to the missing data types to determine the target filling data.

[0040] The stream data is populated with the target data to determine the target data corresponding to the stream data.

[0041] In some embodiments, the apparatus further includes:

[0042] The training set acquisition module is configured to acquire a training set, which includes at least multiple training samples, battery risk parameters corresponding to each training sample, and feature processing identifiers corresponding to each training sample. The feature processing identifiers are used to identify the feature extraction method used during feature extraction.

[0043] The sample feature set determination module is configured to perform feature extraction on each of the training samples according to the feature extraction method corresponding to the feature processing identifier, and determine the sample feature set.

[0044] The training module is configured to train the battery risk determination model by taking features from the sample feature set as input and battery risk parameters corresponding to each training sample as training labels.

[0045] In some embodiments, the training samples include historical data corresponding to the target battery.

[0046] The device further includes:

[0047] The feature subset determination module is configured to determine the feature subset corresponding to the historical data in the sample feature set.

[0048] The feature mapping table generation module is configured to generate a feature mapping table corresponding to the target battery based on the correspondence between each feature in the feature subset and historical data.

[0049] In some embodiments, the risk level determination module is specifically configured to perform:

[0050] Based on the target data and the feature mapping table corresponding to the target battery, the target features corresponding to the target data are determined.

[0051] The target features are input into a pre-trained battery risk determination model to determine the target risk parameters corresponding to the target battery.

[0052] Based on the target risk parameters, the risk level corresponding to the target battery is determined.

[0053] In some embodiments, the training set acquisition module is specifically configured to perform:

[0054] Based on a pre-set ratio of positive to negative samples, positive and negative samples are extracted from the full feature factor library to form the training set.

[0055] In some embodiments, the apparatus further includes:

[0056] The single-charge parameter set generation module is configured to generate a single-charge parameter set corresponding to the target battery in response to the completion of charging of the target battery.

[0057] The parameter and identifier determination module is configured to determine the battery risk parameters and feature processing identifiers corresponding to the single charge parameter set.

[0058] The storage module is configured to store the single-charge parameter set, the battery risk parameters corresponding to the single-charge parameter set, and the feature processing identifier into the full feature factor library.

[0059] In some embodiments, the training samples are a set of single-charge parameters corresponding to a single charge of the same battery. The feature processing identifier includes multiple sub-identifiers. The sub-identifiers are used to identify the feature extraction method used for each parameter in the single-charge parameter set during feature extraction. The sub-identifiers include one or more of the following combinations: Z-score identifier, binarization processing identifier, and equidistant bin discretization identifier.

[0060] In some embodiments, the target battery is a power battery in an electric vehicle, and the target data includes one or more of the following combinations: battery usage behavior parameters, single charge data parameters, time-domain variation parameters, safety hazard and fault perception parameters, and battery management system fault parameters.

[0061] Thirdly, embodiments of this application provide an electronic device, including a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in the first aspect.

[0062] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions thereon, which, when executed by a processor, implement the method described in the first aspect.

[0063] Fifthly, embodiments of this application provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the method described in the first aspect.

[0064] This application embodiment can acquire streaming data generated by the target battery during charging, and fill the streaming data with data based on the historical data corresponding to the target battery to determine the target data corresponding to the streaming data. Then, the risk level of the target battery is determined by the target data and a pre-trained battery risk determination model. In this process, since the battery parameters of the same battery are relatively similar in recent charging times, this application embodiment can determine battery parameters not included in the streaming data from the historical data of the target battery. This maximizes the simulation of the target battery's state during the current charging process, ensuring the completeness of battery parameters and improving the accuracy of battery risk determination. Furthermore, since this application embodiment determines the risk level of the target battery in real time during charging, it can simultaneously guarantee both real-time performance and accuracy in determining battery risk. Attached Figure Description

[0065] The above and other objects, features and advantages of the present application will become clearer from the following description of embodiments of the present application with reference to the accompanying drawings, in which:

[0066] Figure 1 This is a flowchart of the battery risk determination method according to an embodiment of this application;

[0067] Figure 2 A flowchart for determining target data in embodiments of this application;

[0068] Figure 3 This is a flowchart illustrating the training of the battery risk determination model in an embodiment of this application.

[0069] Figure 4 A flowchart for updating the full feature factor library in the embodiments of this application;

[0070] Figure 5 A flowchart illustrating the generation of a feature mapping table for embodiments of this application;

[0071] Figure 6 This is a schematic diagram illustrating the correspondence between features of this application embodiment and historical data;

[0072] Figure 7 A flowchart for determining the risk level for embodiments of this application;

[0073] Figure 8 A flowchart illustrating the process of determining risk levels for embodiments of this application;

[0074] Figure 9 This is a schematic diagram of the battery risk determination device according to an embodiment of this application;

[0075] Figure 10 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0076] The present application is described below based on embodiments, but it is not limited to these embodiments. In the detailed description of the present application below, certain specific details are described in detail. Those skilled in the art can fully understand the present application without these details. To avoid obscuring the substance of the present application, well-known methods, processes, flows, elements, and circuits are not described in detail.

[0077] Furthermore, those skilled in the art should understand that the accompanying drawings provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0078] Unless the context explicitly requires it, words such as "including" or "contains" throughout the application documents should be interpreted as including rather than exclusive or exhaustive; that is, meaning "including but not limited to".

[0079] In the description of this application, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0080] The solutions described in this specification and embodiments, if involving the processing of personal information, will be processed only under the premise of having a legal basis (such as obtaining the consent of the personal information subject, or being necessary for the performance of a contract), and will only be processed within the scope stipulated or agreed upon. A user's refusal to process personal information beyond what is necessary for basic functions will not affect the user's use of basic functions.

[0081] Batteries, as a crucial component in the new energy field, are characterized by high system design difficulty, complex manufacturing, and numerous components. Furthermore, batteries are affected differently by various external environments and operating conditions. This leads to a wide range of causes for battery safety issues and significant uncertainty in the time and space of potential hazard development. Therefore, battery safety is currently a primary concern. In related technologies, the presence of safety hazards in a battery is generally determined by its parameters. However, the number of parameters used to characterize battery safety hazards is insufficient, the transmission granularity is low, and the data quality is poor. Therefore, it is difficult to determine the presence of safety hazards based on a single feature. In other words, the more battery parameters used to determine safety hazards, the higher the accuracy of the risk assessment.

[0082] However, since some parameters of the battery (such as battery capacity retention rate and depth of charge) can usually only be obtained after the battery has finished charging, there is a lag in obtaining these parameters.

[0083] In one scenario, if the relevant technology aims to ensure the accuracy of battery risk assessment, it needs to wait for the battery to finish charging before obtaining the aforementioned battery parameters, which are subject to lag. In this case, a lag problem arises in determining battery risk.

[0084] In another scenario, if the technology aims to ensure real-time performance in determining battery risk, it needs to discard the aforementioned battery parameters that exhibit lag. In this case, the accuracy of battery risk determination will be lower due to the limited number of parameters.

[0085] Therefore, in related technologies, the determination of battery risk still suffers from problems of lag or low accuracy. In other words, how to simultaneously ensure both real-time performance and accuracy in determining battery risk is a problem that urgently needs to be solved.

[0086] To address the aforementioned issues, this application provides a battery risk determination method. This method can be applied to electronic devices, which can be terminals or servers. Terminals can be smartphones, tablets, vehicle control terminals, or personal computers (PCs), etc. Servers can be single servers, distributed server clusters, or cloud servers. Specifically, the electronic device, through the battery risk determination method, can fill in the streaming data generated during the charging of the target battery, ensuring both the integrity of battery parameters and the real-time nature of battery risk determination.

[0087] Specifically, such as Figure 1 As shown, the battery risk determination method in this application embodiment may include the following steps:

[0088] In step S110, the stream data generated by the target battery during the charging process is acquired.

[0089] Streaming data can be used to characterize battery parameters generated in real time during the charging process of the target battery, which can be used to determine battery risk. For example, streaming data can be charging current, battery charging temperature, etc. The streaming data generated by the target battery during the charging process can be directly reported to electronic devices by the device where the target battery is located (such as electric vehicles), or it can be authorized to be sampled and reported to electronic devices by charging equipment (such as charging piles), or streaming data can be obtained by other applicable methods.

[0090] In step S120, in response to the existence of missing data in the streaming data, the streaming data is filled with data based on the historical data corresponding to the target battery to determine the target data corresponding to the streaming data.

[0091] Specifically, the battery parameters not included in the stream data with missing data are those that can only be obtained after the target battery has finished charging. Furthermore, since the battery parameters of the same battery are relatively similar in recent charging times, this embodiment of the application can determine the missing battery parameters in the stream data from the historical data of the target battery. This allows for the simulation of the target battery's state during the current charging process to the greatest extent possible, ensuring the integrity of the battery parameters and improving the accuracy of determining battery risks.

[0092] Specifically, in this application embodiment, the battery parameters corresponding to the last completed charge of the target battery can be directly used as fill data to fill the stream data, thereby determining the target data corresponding to the stream data. Alternatively, this application embodiment can first perform fitting processing on the battery parameters corresponding to the most recent multiple completed charges of the target battery, and use the fitting processing results as fill data to fill the stream data, thereby determining the target data corresponding to the stream data.

[0093] In one optional implementation, the target battery in this application embodiment is a power battery in an electric vehicle, and the target data may include one or more of the following combinations: battery usage behavior parameters, single charge data parameters, time domain variation parameters, safety hazard and fault perception parameters, and battery management system fault parameters.

[0094] Battery usage behavior parameters can include the initial battery capacity, the final battery capacity, the average depth of discharge, and the charging frequency. Single-charge data parameters can include temperature rise, temperature difference, maximum temperature, entropy dispersion, and fluctuation stability. Time-domain variation parameters can include the gradual increase in battery temperature difference, range variation, inconsistency, and frequency of abnormal triggers over a certain period (e.g., within one year, two years). Safety hazard and fault perception parameters can include parameters output by the lithium plating model, the bolt loosening identification model, and the value rate model. Battery management system fault parameters can include over-temperature warning parameters for electronic components, connector fault warning parameters, and insulation fault warning parameters. Additionally, target data can also include parameters of other applicable categories.

[0095] This application embodiment fills in the flow data to simulate the state of the target battery during the current charging process to the greatest extent, ensuring the integrity of battery parameters and improving the accuracy of determining battery risks.

[0096] In an optional implementation, embodiments of this application may use a fitting method to determine the filling data, specifically, as follows: Figure 2 As shown, step S120 above may include the following steps:

[0097] In step S121, the types of missing historical data and streaming data corresponding to the target battery are determined.

[0098] The historical data corresponding to the target battery can be the battery parameters corresponding to the target battery at the time of the most recent multiple charging completions. The missing data types can be pre-determined battery parameter types or real-time determined battery parameter types.

[0099] In one scenario, since the types of battery parameters used to determine battery risk are fixed, and the charging process of the target battery is essentially the same each time, the data that the target battery needs to fill in during each charging process is also essentially the same. Therefore, embodiments of this application can predetermine the types of data to be filled in. In another scenario, embodiments of this application can also query the types of missing data in the streaming data in real time after acquiring the streaming data.

[0100] In step S122, data fitting processing is performed on the parts corresponding to the missing data types in the historical data to determine the target filling data.

[0101] For example, the current charging task of the target battery can be represented as order0, and its corresponding stream data is [x1, x2, x3, x4...xn], where each element represents a type of battery parameter, and n represents the number of data types, with a value greater than or equal to 1. Furthermore, in this embodiment, with the charging order of order0 as 0, the historical charging tasks (i.e., historical data) queried in reverse time can be represented as [order1, order2, order3...order m], where m represents the number of historical charging tasks queried in reverse order, with a value greater than or equal to 1.

[0102] Furthermore, if x1, x2, and x3 in order0 are data types that do not need to be filled (i.e., the battery parameters corresponding to x1, x2, and x3 are real data, and x4-xn are missing data types that need to be filled), then this embodiment can perform fitting processing on the battery parameters in each historical charging task that are of the same type as the missing data to determine the target filling data. Taking x4 as an example, if this embodiment fills the data at position x4, it can first query each battery parameter (a total of m battery parameters) in order1-order m that are of the same type as x4, and then perform fitting processing on these m battery parameters to determine the target filling data corresponding to x4. By analogy, this embodiment can sequentially determine the target filling data corresponding to all missing data types.

[0103] It should be noted that, in the fitting process of this application embodiment, a weighted summation method can be used to determine the target filling data. The weights of each historical charging task (i.e., historical data) can be the same, or they can be determined according to the completion time of each historical charging task. That is to say, the closer the historical charging task is to the current charging task (order0), the closer the battery parameters in it are to the current state of the target battery. Therefore, this application embodiment can assign a larger weight to the recently completed historical charging tasks, and a smaller weight to the historical charging tasks that were completed a long time ago (for example, the weight corresponding to order1 is the largest, and the weight corresponding to order m is the smallest).

[0104] In step S123, the stream data is filled with data according to the target filling data to determine the target data corresponding to the stream data.

[0105] In other words, the target data in this application embodiment is the filled streaming data, which includes both the actual data corresponding to the target battery and the target filled data corresponding to the target battery. This application embodiment effectively fills the streaming data by performing data fitting processing on historical data, so that the streaming data can simulate the state of the target battery during this charging process to the greatest extent possible, ensuring the integrity of battery parameters and improving the accuracy of determining battery risks.

[0106] In step S130, the risk level corresponding to the target battery is determined based on the target data and the pre-trained battery risk determination model.

[0107] Among them, the battery risk determination model can directly output the risk level corresponding to the target battery based on the target data, or it can output the risk parameters corresponding to the target battery based on the target data, and then determine the risk level corresponding to the target battery based on the risk parameters.

[0108] This application embodiment can acquire streaming data generated by the target battery during charging, and fill the streaming data with data based on the historical data corresponding to the target battery to determine the target data corresponding to the streaming data. Then, the risk level of the target battery is determined by the target data and a pre-trained battery risk determination model. In this process, since the battery parameters of the same battery are relatively similar in recent charging times, this application embodiment can determine battery parameters not included in the streaming data from the historical data of the target battery. This maximizes the simulation of the target battery's state during the current charging process, ensuring the completeness of battery parameters and improving the accuracy of battery risk determination. Furthermore, since this application embodiment determines the risk level of the target battery in real time during charging, it can simultaneously guarantee both real-time performance and accuracy in determining battery risk.

[0109] In one alternative implementation, such as Figure 3 As shown, the process of training the battery risk determination model in this application embodiment may include the following steps:

[0110] In step S210, the training set is obtained.

[0111] The training set includes at least multiple training samples, battery risk parameters corresponding to each training sample, and feature processing identifiers corresponding to each training sample. Battery risk parameters are used to characterize the risk level of a battery. Different values ​​of battery risk parameters can correspond to different risk levels. For example, battery risk parameters can be values ​​greater than or equal to 0 and less than or equal to 1. If the battery risk parameter is 0, it indicates the lowest risk level (e.g., no risk). If the battery risk parameter is 1, it indicates the highest risk level (e.g., thermal runaway anomaly exists). It should be noted that the above is only an example of an embodiment of this application. In practical applications, the value of the battery risk parameter can be positively correlated with the risk level (i.e., the larger the battery risk parameter, the higher the risk level) or negatively correlated with the risk level (i.e., the smaller the battery risk parameter, the higher the risk level). Furthermore, embodiments of this application can also divide the range of battery risk parameters into multiple intervals, and set each interval to correspond to a different risk level.

[0112] Feature processing identifiers are used to identify the feature extraction method used during feature extraction. In other words, since there are many types of battery parameters (i.e., parameters in the training samples), this embodiment can set multiple feature processing identifiers according to the types of battery parameters, and select appropriate feature processing methods to process each battery parameter through the feature processing identifiers.

[0113] In one optional implementation, the training samples of this application embodiment are a set of single-charge parameters corresponding to a single charge of the same battery, and the feature processing identifier includes multiple sub-identifiers. The sub-identifiers are used to identify the feature extraction method used for each parameter in the single-charge parameter set during feature extraction, and the sub-identifiers may include one or more of the following combinations: Z-score identifier, binarization processing identifier, and equidistant bin discretization identifier.

[0114] Specifically, the parameters in the single-charge parameter set can include continuous feature parameters, judgment result parameters, and partially continuous parameters. Continuous feature parameters can include parameters such as temperature difference, range, charging frequency, and average depth of discharge. These parameters can be labeled using Z-scores and standardized using Z-scores during feature processing for feature extraction. Judgment result parameters can include parameters such as increasing voltage range, lithium plating model triggering, threshold exceeding triggering, and insulation fault (BST fault). These parameters can be labeled using binarization and extracted using binarization during feature processing. Partially continuous parameters can include parameters such as capacity retention rate and internal resistance. These parameters can be labeled using equidistant bin discretization and extracted using equidistant bin discretization during feature processing.

[0115] Furthermore, in this embodiment of the application, after setting different sub-identifiers for different parameters in the single charging parameter set, the feature processing method corresponding to the battery parameter can be directly determined based on the sub-identifier. During the training process, there is no need to search for the parameter type of the training sample, which reduces the time required for data processing and improves the efficiency of training.

[0116] In an optional implementation, step S210 can be specifically performed as follows: extracting positive and negative samples from the full feature factor library according to the pre-set positive and negative sample ratio to form a training set.

[0117] The full feature factor library is a pre-set database in this application embodiment, which can be constructed from historical data of battery charging tasks. For example, in this application embodiment, after each battery charging is completed, the dataset corresponding to this charging task can be stored in the full feature factor library as data in the full feature factor library.

[0118] In one alternative implementation, such as Figure 4 As shown, the process of updating the full feature factor library may include the following steps:

[0119] In step S310, in response to the completion of charging of the target battery, a set of single charging parameters corresponding to the target battery is generated.

[0120] The parameters in the single charging parameter set can include continuous characteristic parameters, judgment result parameters, and partially continuous parameters.

[0121] In step S320, the battery risk parameters and feature processing identifiers corresponding to the single charging parameter set are determined.

[0122] Among them, the battery risk parameter is the risk parameter corresponding to the entire set of parameters for a single charge, that is, the risk parameter of the battery corresponding to the set of parameters for a single charge. The feature processing identifier is used to identify the feature extraction method used during feature extraction. In this embodiment, multiple feature processing identifiers can be set according to the type of battery parameters in the set of parameters for a single charge, and appropriate feature processing methods can be selected to process each battery parameter through the feature processing identifiers.

[0123] In step S330, the single-charge parameter set, the battery risk parameters corresponding to the single-charge parameter set, and the feature processing identifier are stored in the full feature factor library.

[0124] The data in the full feature factor library can be used as training samples to train the battery risk determination model.

[0125] In step S220, features are extracted from each training sample according to the feature extraction method corresponding to the feature processing identifier to determine the sample feature set.

[0126] Since there are many types of battery parameters (i.e., parameters in the training samples), this embodiment of the application can set different feature processing identifiers for different parameters in the training samples, and directly determine the feature processing method corresponding to the battery parameters based on the feature processing identifiers. During the training process, there is no need to search for the parameter type of the training samples, which reduces the time required for data processing and improves the efficiency of training.

[0127] In step S230, the battery risk determination model is trained by using the features in the sample feature set as input and the battery risk parameters corresponding to each training sample as training labels.

[0128] In this embodiment, a battery risk determination model can be constructed based on extreme gradient boosting (XGBoost). The XGBRegressor module in XGBoost can solve the logistic regression problem for feature and risk quantification. The general parameters in the battery risk determination model can be iterated using general balanced trees (GBtrees) to enhance convergence learning performance. Regarding the loss function, this embodiment can select a binary logistic regression function by learning the objective parameter, enabling the battery risk determination model to quantify the risk of the target battery.

[0129] Furthermore, embodiments of this application can deploy the trained battery risk determination model at a predetermined period, where the predetermined period can be any applicable duration (e.g., one day, two days, one week, etc.). The deployed battery risk determination model can be used to determine the risk level corresponding to the target battery. Specifically, embodiments of this application can acquire streaming data generated by the target battery during charging, and fill the streaming data with data based on the historical data corresponding to the target battery to determine the target data corresponding to the streaming data. Then, the risk level corresponding to the target battery is determined by the target data and the pre-trained battery risk determination model. In this process, since the battery parameters of the same battery are relatively similar in the most recent charging data, embodiments of this application can determine the battery parameters not included in the streaming data from the historical data of the target battery, so as to simulate the state of the target battery in the current charging process to the greatest extent, ensuring the integrity of the battery parameters and improving the accuracy of determining battery risk. Moreover, since embodiments of this application determine the risk level corresponding to the target battery in real time during the charging process of the target battery, embodiments of this application can simultaneously ensure the real-time performance and accuracy of determining battery risk.

[0130] In one optional implementation, the training samples in this application embodiment may include historical data corresponding to the target battery.

[0131] Furthermore, such as Figure 5 As shown, embodiments of this application may further include the following steps:

[0132] In step S410, the feature subset corresponding to the historical data is determined in the sample feature set.

[0133] In step S420, a feature mapping table corresponding to the target battery is generated based on the correspondence between each feature in the feature subset and historical data.

[0134] The historical data refers to the historical data corresponding to the target battery. In other words, if the training samples include the historical data corresponding to the target battery, this embodiment of the application can determine the feature subset corresponding to the historical data from the sample feature set after determining the sample feature set, and generate a feature mapping table corresponding to the target battery based on the correspondence between each feature in the feature subset and the historical data.

[0135] For example, such as Figure 6 As shown, taking historical data corresponding to a certain type (Y) of the target battery as an example, if the value range corresponding to Y (i.e., the value range corresponding to historical data without feature processing) is 0-100, then this embodiment of the application can divide the value range and determine the four numerical intervals corresponding to the value range (i.e., Figure 6 The ranges are 0-25, 25-50, 50-75, and 75-100. Furthermore, in this embodiment, after determining the feature subset corresponding to Y, one or more features included in each numerical interval can be determined based on the correspondence between each feature in the feature subset and historical data. Furthermore, this embodiment can perform fitting processing on each feature in each numerical interval to determine the fitting feature corresponding to each numerical interval (i.e.,...). Figure 6 (T1, T2, T3, and T4 in the original text). Further, the feature mapping table generated in this embodiment can be shown in Table 1 below:

[0136] Table 1

[0137]

[0138]

[0139] Furthermore, this feature mapping table can be used in the process of determining battery risk. Specifically, in this embodiment of the application, after obtaining the streaming data of the target battery and determining the target data through data filling, the features corresponding to the battery parameters in the target data can be directly determined according to the above feature mapping table, eliminating the feature extraction process and improving the efficiency of determining battery risk.

[0140] In one alternative implementation, such as Figure 7 As shown, step S130 above may include the following steps:

[0141] In step S131, the target features corresponding to the target data are determined based on the target data and the feature mapping table corresponding to the target battery.

[0142] The feature mapping table can be as shown in Table 1 above. Through the feature mapping table, the embodiments of this application can directly determine the features corresponding to the battery parameters in the target data, eliminating the feature extraction process and improving the efficiency of determining battery risk.

[0143] Furthermore, in one scenario, since the target data is used to characterize the battery parameters of the target battery during this charging process, each type of parameter in the target data includes only a single data point. Moreover, because some feature processing methods in this embodiment (such as equidistant bin discretization) require several data points, this can negatively impact the feature processing of the target data. To address this, this embodiment can directly determine the features corresponding to the battery parameters in the target data based on the aforementioned feature mapping table, eliminating the feature extraction process and avoiding the problem of not being able to perform feature processing on individual data points.

[0144] In step S132, the target features are input into the pre-trained battery risk determination model to determine the target risk parameters corresponding to the target battery.

[0145] In step S133, the risk level corresponding to the target battery is determined based on the target risk parameters.

[0146] The target risk parameter is used to characterize the risk level of the target battery. Different values ​​of the target risk parameter can correspond to different risk levels. For example, the target risk parameter can be a value greater than or equal to 0 and less than or equal to 1. If the target risk parameter is 0, it can characterize the target battery with the lowest risk level (e.g., no risk). If the target risk parameter is 1, it can characterize the target battery with the highest risk level (e.g., thermal runaway anomaly exists). It should be noted that the above is only an example of an embodiment of this application. In practical applications, the value of the target risk parameter can be positively correlated with the risk level (i.e., the larger the target risk parameter, the higher the risk level) or negatively correlated with the risk level (i.e., the smaller the target risk parameter, the higher the risk level). In addition, the embodiment of this application can also divide the value range of the target risk parameter into multiple intervals and set each interval to correspond to a different risk level.

[0147] For example, such as Figure 8 As shown, this embodiment of the application can set four numerical ranges (0-B1, B1-B2, B2-B3, B3-1) for the target risk parameter, and assign a corresponding risk level to each numerical range. That is, the risk level corresponding to 0-B1 is no risk, the risk level corresponding to B1-B2 is Level 1 risk, the risk level corresponding to B2-B3 is Level 2 risk, and the risk level corresponding to B3-1 is Level 3 risk. Among them, the risk level corresponding to no risk is the lowest, and the risk level corresponding to Level 3 risk is the highest.

[0148] Furthermore, after determining the target risk parameter A, this embodiment can determine the numerical range to which the target risk parameter A belongs, and determine the risk level corresponding to the target risk parameter A based on the numerical range to which the target risk parameter A belongs. This risk level is also the risk level corresponding to the target battery. In the process of determining the risk level, since the battery parameters of the same battery are relatively similar in recent charging times, this embodiment can identify battery parameters not included in the streaming data from the historical data of the target battery. This maximizes the simulation of the target battery's state during the current charging process, ensuring the completeness of the battery parameters and improving the accuracy of determining battery risk. Moreover, since this embodiment determines the risk level of the target battery in real time during the charging process, it can simultaneously guarantee both real-time performance and accuracy in determining battery risk.

[0149] Based on the same technical concept, embodiments of this application also provide a battery risk determination device, such as... Figure 9 As shown, the device includes: a streaming data acquisition module 91, a data filling module 92, and a risk level determination module 93.

[0150] The streaming data acquisition module 91 is configured to acquire streaming data generated by the target battery during the charging process.

[0151] The data filling module 92 is configured to respond to the existence of missing data in the streaming data by filling the streaming data with data based on the historical data corresponding to the target battery, and to determine the target data corresponding to the streaming data.

[0152] The risk level determination module 93 is configured to determine the risk level corresponding to the target battery based on the target data and a pre-trained battery risk determination model.

[0153] In some embodiments, the data filling module 92 is specifically configured to perform:

[0154] Determine the types of missing data in the historical data and streaming data corresponding to the target battery.

[0155] Data fitting processing is performed on the portions of the historical data corresponding to the missing data types to determine the target filling data.

[0156] The stream data is populated with the target data to determine the target data corresponding to the stream data.

[0157] In some embodiments, the apparatus further includes:

[0158] The training set acquisition module is configured to acquire a training set, which includes at least multiple training samples, battery risk parameters corresponding to each training sample, and feature processing identifiers corresponding to each training sample. The feature processing identifiers are used to identify the feature extraction method used during feature extraction.

[0159] The sample feature set determination module is configured to perform feature extraction on each of the training samples according to the feature extraction method corresponding to the feature processing identifier, and determine the sample feature set.

[0160] The training module is configured to train the battery risk determination model by taking features from the sample feature set as input and battery risk parameters corresponding to each training sample as training labels.

[0161] In some embodiments, the training samples include historical data corresponding to the target battery.

[0162] The device further includes:

[0163] The feature subset determination module is configured to determine the feature subset corresponding to the historical data in the sample feature set.

[0164] The feature mapping table generation module is configured to generate a feature mapping table corresponding to the target battery based on the correspondence between each feature in the feature subset and historical data.

[0165] In some embodiments, the risk level determination module 93 is specifically configured to perform:

[0166] Based on the target data and the feature mapping table corresponding to the target battery, the target features corresponding to the target data are determined.

[0167] The target features are input into a pre-trained battery risk determination model to determine the target risk parameters corresponding to the target battery.

[0168] Based on the target risk parameters, the risk level corresponding to the target battery is determined.

[0169] In some embodiments, the training set acquisition module is specifically configured to perform:

[0170] Based on a pre-set ratio of positive to negative samples, positive and negative samples are extracted from the full feature factor library to form the training set.

[0171] In some embodiments, the apparatus further includes:

[0172] The single-charge parameter set generation module is configured to generate a single-charge parameter set corresponding to the target battery in response to the completion of charging of the target battery.

[0173] The parameter and identifier determination module is configured to determine the battery risk parameters and feature processing identifiers corresponding to the single charge parameter set.

[0174] The storage module is configured to store the single-charge parameter set, the battery risk parameters corresponding to the single-charge parameter set, and the feature processing identifier into the full feature factor library.

[0175] In some embodiments, the training samples are a set of single-charge parameters corresponding to a single charge of the same battery. The feature processing identifier includes multiple sub-identifiers. The sub-identifiers are used to identify the feature extraction method used for each parameter in the single-charge parameter set during feature extraction. The sub-identifiers include one or more of the following combinations: Z-score identifier, binarization processing identifier, and equidistant bin discretization identifier.

[0176] In some embodiments, the target battery is a power battery in an electric vehicle, and the target data includes one or more of the following combinations: battery usage behavior parameters, single charge data parameters, time-domain variation parameters, safety hazard and fault perception parameters, and battery management system fault parameters.

[0177] This application embodiment can acquire streaming data generated by the target battery during charging, and fill the streaming data with data based on the historical data corresponding to the target battery to determine the target data corresponding to the streaming data. Then, the risk level of the target battery is determined by the target data and a pre-trained battery risk determination model. In this process, since the battery parameters of the same battery are relatively similar in recent charging times, this application embodiment can determine battery parameters not included in the streaming data from the historical data of the target battery. This maximizes the simulation of the target battery's state during the current charging process, ensuring the completeness of battery parameters and improving the accuracy of battery risk determination. Furthermore, since this application embodiment determines the risk level of the target battery in real time during charging, it can simultaneously guarantee both real-time performance and accuracy in determining battery risk.

[0178] Figure 10 This is a schematic diagram of an electronic device according to an embodiment of this application. For example... Figure 10 As shown, Figure 10The illustrated electronic device is a general address lookup device, comprising a general computer hardware architecture, including at least a processor 1001 and a memory 1002. The processor 1001 and memory 1002 are connected via a bus 1003. The memory 1002 is adapted to store instructions or programs executable by the processor 1001. The processor 1001 can be a standalone microprocessor or a collection of one or more microprocessors. Thus, the processor 1001 executes the instructions stored in the memory 1002, thereby performing the method flow described in the embodiments of this application to process data and control other devices. The bus 1003 connects the aforementioned components together, and also connects these components to a display controller 1004, a display device, and an input / output (I / O) device 1005. The input / output (I / O) device 1005 can be a mouse, keyboard, modem, network interface, touch input device, motion-sensing input device, printer, and other devices known in the art. Typically, the input / output device 1005 is connected to the system via an input / output (I / O) controller 1006.

[0179] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus (devices), or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0180] This application is described with reference to flowchart illustrations of methods, apparatus (devices), and computer program products according to embodiments of this application. It should be understood that each step in the flowchart can be implemented by computer program instructions.

[0181] These computer program instructions may be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction means, the implementation process of which is described in the instruction means. Figure 1 The function specified in one or more processes.

[0182] These computer program instructions may also be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, produce instructions for implementing processes. Figure 1 A device for a function specified in one or more processes.

[0183] Another embodiment of this application relates to a non-volatile storage medium for storing a computer-readable program for use by a computer to execute some or all of the above-described method embodiments.

[0184] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program specifying the relevant hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0185] Another embodiment of this application relates to a computer program product, including a computer program / instructions that, when executed by a processor, can implement some or all of the above-described method embodiments.

[0186] That is, those skilled in the art will understand that the embodiments of this application can specify related hardware (including the processor itself) by having the processor execute a computer program product (computer program / instructions) to implement all or part of the steps in the methods of the above embodiments.

[0187] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A battery risk determination method, characterized by, The method comprises: obtaining flow data generated by a target battery during charging; in response to the flow data having missing data, filling the missing data in the flow data according to historical data corresponding to the target battery, wherein the missing battery parameters in the flow data cannot be obtained before the target battery is fully charged; and determining a risk level corresponding to the target battery according to the target data and a pre-trained battery risk determination model.

2. The method of claim 1, wherein, The filling of the missing data in the flow data according to the historical data corresponding to the target battery comprises: determining the historical data corresponding to the target battery and the missing data types of the flow data; performing data fitting processing on the part of the historical data corresponding to the missing data types to determine target filling data; and filling the missing data in the flow data according to the target filling data to determine the target data corresponding to the flow data.

3. The method of claim 1, wherein, The method further comprises: obtaining a training set, wherein the training set comprises at least a plurality of training samples, battery risk parameters corresponding to each of the training samples, and feature processing identifiers corresponding to each of the training samples, wherein the feature processing identifiers are used to identify the feature extraction method adopted during feature extraction; performing feature extraction on each of the training samples according to the feature extraction method corresponding to the feature processing identifier to determine a sample feature set; and training the battery risk determination model by taking the features in the sample feature set as input and taking the battery risk parameters corresponding to each of the training samples as training labels.

4. The method of claim 3, wherein, The training samples comprise historical data corresponding to the target battery. The method further comprises: determining a feature sub-set corresponding to the historical data in the sample feature set; and generating a feature mapping table corresponding to the target battery according to the correspondence between each feature in the feature sub-set and the historical data.

5. The method of claim 4, wherein, The determination of the risk level corresponding to the target battery according to the target data and the pre-trained battery risk determination model comprises: determining target features corresponding to the target data according to the target data and the feature mapping table corresponding to the target battery; inputting the target features into the pre-trained battery risk determination model to determine target risk parameters corresponding to the target battery; and determining the risk level corresponding to the target battery according to the target risk parameters.

6. The method of claim 3, wherein, The obtaining of the training set comprises: extracting positive samples and negative samples from a full feature factor library according to a pre-set positive sample ratio and negative sample ratio to form the training set.

7. The method of claim 6, wherein, The method further comprises: in response to the completion of charging of a target battery, generating a single charging parameter set corresponding to the target battery; determining battery risk parameters and feature processing identifiers corresponding to the single charging parameter set; and storing the single charging parameter set, the battery risk parameters corresponding to the single charging parameter set, and the feature processing identifiers into the full feature factor library.

8. The method according to claim 3 or 7, characterized in that, The training sample is a single charging parameter set corresponding to a single charging of the same battery, the feature processing identifier includes multiple sub-identifiers, the sub-identifier is used to identify the feature extraction method adopted by each parameter in the single charging parameter set during feature extraction, and the sub-identifier includes one or more combinations of the following: Z-score identifier, binary processing identifier, and equidistant binning discretization identifier.

9. The method of claim 1, wherein, The target battery is a power battery in an electric vehicle, and the target data includes one or more combinations of the following: battery usage behavior parameters, single charging data parameters, time domain variation parameters, safety hazard and fault awareness parameters, and battery management system fault parameters.

10. A battery risk determination apparatus, characterized by, The device includes: a flow data acquisition module configured to acquire flow data generated by the target battery during charging; a data filling module configured to, in response to the flow data having missing data, fill the flow data with historical data corresponding to the target battery, determine target data corresponding to the flow data, and acquire missing battery parameters in the flow data before the target battery is fully charged; and a risk level determination module configured to determine a risk level corresponding to the target battery according to the target data and a pre-trained battery risk determination model. 11.An electronic device, comprising a memory and a processor, wherein, The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method of any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, The computer program is stored in the computer readable storage medium, and the computer program is executed by the processor to implement the method of any one of claims 1-9.

13. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instruction is executed by the processor to implement the method of any one of claims 1-9.

Images