Power battery safety early warning method, device, equipment, medium and program product

Abstract

This invention provides a method, device, equipment, medium, and program product for power battery safety early warning, relating to the field of new energy vehicle technology. The power battery safety early warning method includes: obtaining a safety early warning result indicating whether the target power battery has an abnormal risk based on a power battery safety early warning model and the current charging behavior data of the target power battery; wherein the power battery safety early warning model is obtained by training on a synthetic charging behavior dataset and a first historical charging behavior dataset of multiple power batteries; the synthetic charging behavior dataset is obtained by synthesizing data from a second historical charging behavior dataset of power batteries with abnormal risks among the multiple power batteries. The solution of this invention, by combining user charging behavior for abnormal risk early warning of power batteries, significantly improves the prediction accuracy of abnormal risks of power batteries.

Description

Technical Field

[0001] This invention relates to the field of new energy vehicle technology, and in particular to a method, device, equipment, medium, and program product for safety early warning of power batteries. Background Technology

[0002] With the increasing number of new energy vehicles on the road and the extended service life of their batteries, fires involving new energy vehicles have become more and more frequent in recent years. Thermal runaway in new energy vehicles not only causes huge economic losses and affects user experience, but also negatively impacts the brand power and reputation of manufacturers.

[0003] In existing technologies, power battery safety early warning technologies mostly focus on the internal mechanisms of the battery cell, such as analyzing the battery's voltage difference and internal resistance to warn of abnormal risks. However, early warning of abnormal risks in power batteries based on the internal mechanisms of the battery cell suffers from problems such as unsatisfactory expected results and low prediction accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a method, device, equipment, medium, and program product for safety early warning of power batteries, in order to solve the problem of low accuracy of abnormal risk early warning of power batteries in the prior art.

[0005] To achieve the above objectives, the present invention is implemented as follows:

[0006] In a first aspect, embodiments of the present invention provide a power battery safety early warning method, comprising:

[0007] Based on the power battery safety early warning model and the current charging behavior data of the target power battery, obtain a safety early warning result to indicate whether there is an abnormal risk to the target power battery;

[0008] The power safety early warning model is obtained by training on a synthetic charging behavior dataset and a first historical charging behavior dataset of multiple power batteries; the synthetic charging behavior dataset is obtained by synthesizing data from a second historical charging behavior dataset of power batteries with abnormal risks among the multiple power batteries.

[0009] Optionally, the power battery safety early warning method further includes:

[0010] The power battery safety early warning model is obtained by following these steps:

[0011] Obtain the first historical charging behavior dataset of the plurality of power batteries; wherein, the number of power batteries with abnormal risks is less than the number of power batteries without abnormal risks.

[0012] Based on the first historical charging behavior dataset, a second historical charging behavior dataset of the power batteries with abnormal risks among the plurality of power batteries is obtained.

[0013] Data synthesis processing is performed on the second historical charging behavior dataset to obtain the synthesized charging behavior dataset;

[0014] A training dataset is obtained based on the synthetic charging behavior dataset and the first historical charging behavior dataset;

[0015] The preset classification model is trained based on the training dataset to obtain the power battery safety early warning model.

[0016] Optionally, the power battery safety early warning method includes, based on the first historical charging behavior dataset, obtaining a second historical charging behavior dataset of the power batteries with abnormal risks among the plurality of power batteries, including:

[0017] The first historical charging behavior dataset is cleaned to remove missing and abnormal data, resulting in a cleaned first historical charging behavior dataset.

[0018] The first historical charging behavior dataset after cleaning is normalized to obtain the normalized first historical charging behavior dataset.

[0019] Perform statistical analysis on the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after statistical analysis.

[0020] Based on the first historical charging behavior dataset after statistical analysis, a second historical charging behavior dataset is obtained for the power batteries among the multiple power batteries that have abnormal risks.

[0021] Optionally, the power battery safety early warning method includes performing statistical analysis on the normalized first historical charging behavior dataset to obtain the statistically analyzed first historical charging behavior dataset, including:

[0022] A correlation analysis is performed on the normalized first historical charging behavior dataset, and historical charging behavior data with a Pearson correlation coefficient equal to a preset value are removed from the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after correlation analysis.

[0023] The first historical charging behavior dataset after statistical analysis is obtained by removing historical charging behavior data with variance less than a threshold from the first historical charging behavior dataset after removing data with correlation analysis.

[0024] Optionally, the power battery safety early warning method, wherein the process of performing data synthesis processing based on the second historical charging behavior dataset to obtain the charging behavior synthetic dataset includes:

[0025] For each historical charging behavior data in the second historical charging behavior dataset, obtain at least one target data in the second historical charging behavior dataset that corresponds to the historical charging behavior data.

[0026] Obtain the absolute value of the difference between the historical charging behavior data and the target data;

[0027] Based on the product of the absolute value and the random number, and the sum of the historical charging behavior data, one of the charging behavior synthetic data in the charging behavior synthetic dataset is obtained.

[0028] Optionally, the power battery safety early warning method, wherein obtaining at least one target data corresponding to the historical charging behavior data in the second historical charging behavior dataset includes:

[0029] Obtain the Euclidean distance between the historical charging behavior data and the target historical charging behavior data; wherein, the target historical charging behavior data is the historical charging behavior data in the second historical charging behavior dataset that belongs to the same dimension as the historical charging behavior data;

[0030] Based on the Euclidean distance, obtain multiple nearest neighbor data of the historical charging behavior data;

[0031] According to the preset sampling rate, at least one target data is selected from the plurality of nearest neighbor data.

[0032] Secondly, embodiments of the present invention provide a power battery safety early warning device, comprising:

[0033] The first acquisition module is used to acquire a safety warning result indicating whether there is an abnormal risk in the target power battery, based on the power battery safety warning model and the current charging behavior data of the target power battery.

[0034] The power safety early warning model is obtained by training on a synthetic charging behavior dataset and a first historical charging behavior dataset of multiple power batteries; the synthetic charging behavior dataset is obtained by synthesizing data from a second historical charging behavior dataset of power batteries with abnormal risks among the multiple power batteries.

[0035] Thirdly, embodiments of the present invention provide a power battery safety warning device, including a processor, a memory, and a program or instructions stored in the memory and operable on the processor; when the processor executes the program or instructions, it implements the power battery safety warning method as described in the first aspect.

[0036] Fourthly, embodiments of the present invention provide a readable storage medium storing a program, which, when executed by a processor, implements the power battery safety warning method as described in the first aspect.

[0037] Fifthly, embodiments of the present invention provide a computer program product, including computer instructions, which, when executed by a processor, implement the power battery safety warning method as described in the first aspect.

[0038] The above-described technical solution of the present invention has at least the following beneficial effects:

[0039] The power battery safety early warning method described in this embodiment of the invention obtains a safety early warning result indicating whether the target power battery has any abnormal risks, based on a power battery safety early warning model and the current charging behavior data of the target power battery. The power battery safety early warning model is trained using a synthesized charging behavior dataset and a first historical charging behavior dataset of multiple power batteries. The synthesized charging behavior dataset is obtained by synthesizing data from a second historical charging behavior dataset of power batteries with abnormal risks. Thus, by training the power battery safety early warning model with user charging behavior data and predicting whether a power battery has abnormal risks based on the power battery safety early warning model, the prediction accuracy of abnormal risks of power batteries is effectively improved. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating the power battery safety early warning method according to an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram illustrating the principle of the support vector machine in an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram of the structure of the power battery safety early warning device according to an embodiment of the present invention. Detailed Implementation

[0043] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0044] This invention addresses the problem of low accuracy in early warning of abnormal risks in power batteries in the prior art by providing a power battery safety early warning method, device, equipment, medium, and program product.

[0045] like Figure 1 As shown, this embodiment of the invention provides a power battery safety early warning method, including:

[0046] S101, Based on the power battery safety early warning model and the current charging behavior data of the target power battery, obtain a safety early warning result to indicate whether there is an abnormal risk in the target power battery;

[0047] The power safety early warning model is obtained by training on a synthetic charging behavior dataset and a first historical charging behavior dataset of multiple power batteries; the synthetic charging behavior dataset is obtained by synthesizing data from a second historical charging behavior dataset of power batteries with abnormal risks among the multiple power batteries.

[0048] It's understandable that charging behavior data refers to data related to a user's charging behavior. This data can be multi-dimensional, including current, temperature, and state of charge (SOC) dimensions, such as current, maximum voltage of a single cell, minimum voltage of a single cell, maximum temperature of a single cell, minimum temperature of a single cell, and SOC. It's also understandable that charging behavior data includes fast / slow charging indicators, vehicle identification numbers (VINs), and cumulative mileage.

[0049] It should be noted that, when providing early warning of abnormal risks of power batteries, the embodiments of the present invention fully consider the user's charging behavior and associate the user's charging behavior with the abnormal risks of power batteries. Compared with the prior art, which mainly provides early warning of abnormal risks of power batteries from the internal mechanism of the battery cell, the prediction accuracy is greatly improved.

[0050] As shown in Table 1 below, the current charging behavior data of the target power battery includes the percentage of fast charging times, the average state of charge at the end of charging, and the percentage of time spent charging for more than 40 degrees. The current charging behavior data of the target power battery is input into the power battery safety early warning model, and the safety early warning result is output. A safety early warning result of 1 indicates that the power battery has an abnormal risk, and a safety early warning result of 0 indicates that the power battery does not have an abnormal risk.

[0051]

[0052] Table 1: Examples of Security Warning Results

[0053] In one embodiment, optionally, the method further includes:

[0054] The power battery safety early warning model is obtained by following these steps:

[0055] Obtain the first historical charging behavior dataset of the plurality of power batteries; wherein, the number of power batteries with abnormal risks is less than the number of power batteries without abnormal risks.

[0056] Based on the first historical charging behavior dataset, a second historical charging behavior dataset of the power batteries with abnormal risks among the plurality of power batteries is obtained.

[0057] Data synthesis processing is performed on the second historical charging behavior dataset to obtain the synthesized charging behavior dataset;

[0058] A training dataset is obtained based on the synthetic charging behavior dataset and the first historical charging behavior dataset;

[0059] The preset classification model is trained based on the training dataset to obtain the power battery safety early warning model.

[0060] In this embodiment of the invention, a first historical charging behavior dataset of multiple vehicles' power batteries within a preset time period (e.g., the past year) can be obtained. This dataset includes historical charging behavior datasets of power batteries with abnormal risks (i.e., a second historical charging behavior dataset), as well as historical charging behavior datasets of power batteries without abnormal risks. It is understood that the number of power batteries with abnormal risks is less than the number of power batteries without abnormal risks.

[0061] Data synthesis processing is performed on the second historical charging behavior dataset to obtain a synthetic charging behavior dataset. This synthetic dataset and the first historical charging behavior dataset constitute a sample dataset. Using a random selection principle, the sample dataset is divided into a training dataset, a test dataset, and a validation dataset according to a preset ratio. For example, with a preset ratio of 7:2:1, 70% of the data randomly selected from the sample dataset is used as the training dataset, 20% as the test dataset, and 10% as the validation dataset.

[0062] Based on the training dataset, a power battery safety early warning model is built using a pre-defined classification model.

[0063] It should be noted that the preset classification model is a Support Vector Machine (SVM). The basic idea of ​​SVM learning is to find a separating hyperplane that correctly divides the training dataset and maximizes the geometric margin. Figure 2 As shown, w·x+b=0 is the separating hyperplane. For linearly separable datasets, there are infinitely many such hyperplanes (i.e., perceptrons), but the separating hyperplane with the largest geometric margin is unique.

[0064] It should also be noted that, optionally, the method further includes:

[0065] Based on the synthesized charging behavior dataset and the first historical charging behavior dataset, a test dataset is obtained;

[0066] A confusion matrix is ​​constructed using the test dataset and the power battery safety early warning model.

[0067] The power battery safety early warning model is evaluated based on the confusion matrix to obtain the model evaluation results.

[0068] The test dataset has been described in the above embodiments and will not be repeated here.

[0069] The test dataset is input into the power battery safety early warning model to obtain the safety early warning result (i.e., the predicted value). A confusion matrix is ​​constructed based on the actual value and the predicted value corresponding to the test dataset, as shown in Table 2. The symbols in Table 2 are explained below:

[0070] Category 1 indicates a safety warning result indicating an abnormal risk to the power battery;

[0071] Category 2 indicates a safety warning result indicating that the power battery does not pose any abnormal risks;

[0072] 'a' represents the number of predicted safety warning results indicating abnormal risks to the power battery, and the number of actual safety warning results indicating abnormal risks to the power battery, i.e., the number of predicted values ​​that match the actual values ​​(i.e., the predictions are correct).

[0073] b represents the number of predicted safety warnings that the power battery does not have any abnormal risks, and the actual value is the number of predicted safety warnings that the power battery has abnormal risks. In other words, the number of predicted values ​​that do not match the actual values ​​(i.e., the predictions are incorrect).

[0074] c represents the number of safety warning results where the predicted value indicates an abnormal risk to the power battery, and the actual value indicates a safety warning result where the power battery does not have an abnormal risk. In other words, the number of predicted values ​​that do not match the actual values ​​(i.e., the predictions are incorrect).

[0075] d represents the number of predicted values ​​that indicate no abnormal risks to the power battery and the number of actual values ​​that indicate no abnormal risks to the power battery. In other words, it represents the number of predicted values ​​that match the actual values ​​(i.e., the predictions are correct).

[0076]

[0077] Table 2: Example of a confusion matrix

[0078] Furthermore, the confusion matrix is ​​used to evaluate the power battery safety early warning model to obtain model evaluation results that indicate the predictive performance of the power battery safety early warning model.

[0079] The power battery safety early warning model is evaluated using at least one of the following evaluation methods:

[0080] Accuracy evaluation method:

[0081] Dividing the number of correct predictions by the total number of test datasets equals the model's prediction accuracy, where accuracy = (a+d) / (a+b+c+d). It's understandable that the closer the accuracy is to 1, the better the model.

[0082] Accuracy assessment method:

[0083] Also known as precision, it represents the proportion of samples that we predict to be the minority class, but which are actually in the minority class. Precision = a / (a+c). In SVM, higher precision means more minority classes are captured, and more accurate predictions of the minority class are made. Conversely, lower precision means more majority classes are falsely identified. Precision is a measure of the cost of misclassifying the majority class.

[0084] Recall rate assessment method:

[0085] Also known as sensitivity, true positive rate, or recall rate, it represents the proportion of samples that are correctly predicted out of all samples that are true to class 1. Recall rate = a / (a+b).

[0086] It's important to note that recall and precision have the same numerator; only the denominator differs. Furthermore, recall and precision are inversely related; the balance between them represents the need to capture the minority class and the need to minimize the impact on the majority class. Which side to prioritize depends on business needs.

[0087] In one implementation, optionally, a second historical charging behavior dataset is obtained from the first historical charging behavior dataset, showing the power batteries with abnormal risks among the plurality of power batteries, including:

[0088] The first historical charging behavior dataset is cleaned to remove missing and abnormal data, resulting in a cleaned first historical charging behavior dataset.

[0089] The first historical charging behavior dataset after cleaning is normalized to obtain the normalized first historical charging behavior dataset.

[0090] Perform statistical analysis on the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after statistical analysis.

[0091] Based on the first historical charging behavior dataset after statistical analysis, a second historical charging behavior dataset is obtained for the power batteries among the multiple power batteries that have abnormal risks.

[0092] In this embodiment of the invention, the first historical charging behavior dataset is first cleaned to remove missing and abnormal data, resulting in a cleaned first historical charging behavior dataset. Specifically, the data cleaning operation is as follows:

[0093] For missing data, if the percentage of missing data is less than 20%, the data is filled by taking the average value directly; if the percentage of missing data is greater than 20% but less than 50%, the missing data and the data without missing data are marked; if the percentage of missing data is greater than 50%, the data is deleted.

[0094] Abnormal data whose values ​​exceed the threshold range should be deleted directly.

[0095] Then, the first historical charging behavior dataset after cleaning is normalized to bring all data to the same level of measurement, resulting in the normalized first historical charging behavior dataset. Specifically, the data normalization operation is as follows:

[0096]

[0097] Where x represents any data in the first historical charging behavior dataset; x ′ This represents the corresponding normalized data; min represents the minimum value of the data; max represents the maximum value of the data.

[0098] Furthermore, statistical analysis is performed on the normalized first historical charging behavior dataset to obtain a statistically analyzed first historical charging behavior dataset. This allows for the extraction of a second historical charging behavior dataset from the first historical charging behavior dataset, identifying power batteries with abnormal risks among the multiple power batteries. It is understood that the first historical charging behavior dataset includes the second historical charging behavior dataset.

[0099] In one embodiment, optionally, statistical analysis is performed on the normalized first historical charging behavior dataset to obtain the statistically analyzed first historical charging behavior dataset, including:

[0100] A correlation analysis is performed on the normalized first historical charging behavior dataset, and historical charging behavior data with a Pearson correlation coefficient equal to a preset value are removed from the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after correlation analysis.

[0101] The first historical charging behavior dataset after statistical analysis is obtained by removing historical charging behavior data with variance less than a threshold from the first historical charging behavior dataset after removing data with correlation analysis.

[0102] In this embodiment of the invention, a correlation analysis is performed on the normalized first historical charging behavior dataset. The Pearson correlation coefficient between any two historical charging behavior data points in the dataset is calculated, and two historical charging behavior data points with a Pearson correlation coefficient equal to a preset value are removed. The Pearson correlation coefficient ranges from -1 to 1, where -1 indicates a completely negative correlation, 1 indicates a completely positive correlation, and 0 indicates no correlation. The closer the coefficient is to 0, the weaker the correlation. Here, the preset value can be a number greater than 0.8 or less than -0.8.

[0103] The Pearson correlation coefficient is shown in the following formula:

[0104]

[0105] Where r represents the Pearson correlation coefficient; x i and y i These represent two sets of historical charging behavior data.

[0106] Furthermore, the variance of the first historical charging behavior dataset after correlation analysis is less than a threshold is calculated to obtain the first historical charging behavior dataset after statistical analysis.

[0107] In one embodiment, optionally, data synthesis processing is performed based on the second historical charging behavior dataset to obtain the synthesized charging behavior dataset, including:

[0108] For each historical charging behavior data in the second historical charging behavior dataset, obtain at least one target data in the second historical charging behavior dataset that corresponds to the historical charging behavior data.

[0109] Obtain the absolute value of the difference between the historical charging behavior data and the target data;

[0110] Based on the product of the absolute value and the random number, and the sum of the historical charging behavior data, one of the charging behavior synthetic data in the charging behavior synthetic dataset is obtained.

[0111] In one embodiment, optionally, at least one target data corresponding to the historical charging behavior data in the second historical charging behavior dataset is obtained, including:

[0112] Obtain the Euclidean distance between the historical charging behavior data and the target historical charging behavior data; wherein, the target historical charging behavior data is the historical charging behavior data in the second historical charging behavior dataset that belongs to the same dimension as the historical charging behavior data;

[0113] Based on the Euclidean distance, obtain multiple nearest neighbor data of the historical charging behavior data;

[0114] According to the preset sampling rate, at least one target data is selected from the plurality of nearest neighbor data.

[0115] In this embodiment of the invention, since the historical charging behavior dataset of power batteries with abnormal risks is less than that of power batteries without abnormal risks in the first historical charging behavior dataset, there is a problem of imbalanced training dataset samples. To solve this problem, the Synthetic Minority Oversampling Technique (SMOTE) is used. SMOTE is an improved scheme based on the random oversampling algorithm. Since random oversampling adopts a strategy of simply copying samples to increase minority class samples, it is easy to cause model overfitting, that is, the information learned by the model is too specific and not generalizable enough. The basic idea of ​​SMOTE is to analyze the minority class samples and artificially synthesize new samples based on the minority class samples and add them to the dataset.

[0116] Specifically, for each historical charging behavior data in the second historical charging behavior dataset, the Euclidean distance between the historical charging behavior data and the target historical charging behavior data is obtained, and multiple nearest neighbor data are obtained based on the Euclidean distance. A preset sampling rate is determined based on the sample imbalance ratio, and at least one target data corresponding to the historical charging behavior data is selected from the multiple nearest neighbor data.

[0117] Then, for each target data, charging behavior composite data is synthesized by combining it with the corresponding historical charging behavior data according to the following formula.

[0118] xnew=x+rand(0,1)*|x-xn|

[0119] Where xnew represents the synthetic data of charging behavior; x represents the charging behavior data; and xn represents the target data corresponding to x.

[0120] It should be noted that the charging behavior data in this embodiment of the invention involves at least one of the dimensions of current, temperature and state of charge, as shown in Table 3 below.

[0121]

[0122] Table 3: Examples of Charging Behavior Data

[0123] In summary, the power battery safety early warning method described in this embodiment of the invention incorporates user charging behavior data into the method, and the charging behavior data involves multiple dimensions, such as current, temperature, and state of charge, and constructs a power battery safety early warning model, effectively improving the prediction accuracy of abnormal risks of power batteries.

[0124] like Figure 3 As shown, this embodiment of the invention also provides a power battery safety warning device, including:

[0125] The first acquisition module 301 is used to acquire a safety warning result indicating whether there is an abnormal risk in the target power battery based on the power battery safety warning model and the current charging behavior data of the target power battery.

[0126] The power safety early warning model is obtained by training on a synthetic charging behavior dataset and a first historical charging behavior dataset of multiple power batteries; the synthetic charging behavior dataset is obtained by synthesizing data from a second historical charging behavior dataset of power batteries with abnormal risks among the multiple power batteries.

[0127] Optionally, the power battery safety warning device further includes:

[0128] The second acquisition module is used to acquire the power battery safety warning model through the following sub-modules:

[0129] The first acquisition submodule is used to acquire the first historical charging behavior dataset of the plurality of power batteries; wherein, the number of power batteries with abnormal risks is less than the number of power batteries without abnormal risks.

[0130] The second acquisition submodule is used to acquire the second historical charging behavior dataset of the power batteries with abnormal risks among the multiple power batteries based on the first historical charging behavior dataset.

[0131] The first acquisition submodule is used to perform data synthesis processing based on the second historical charging behavior dataset to obtain the charging behavior synthesized dataset.

[0132] The second acquisition submodule is used to obtain a training dataset based on the charging behavior synthetic dataset and the first historical charging behavior dataset;

[0133] The third acquisition submodule is used to train a preset classification model based on the training dataset to obtain the power battery safety early warning model.

[0134] Optionally, in the power battery safety warning device, the second acquisition submodule includes:

[0135] The cleaning unit is used to clean the first historical charging behavior dataset to remove missing and abnormal data, and obtain the cleaned first historical charging behavior dataset.

[0136] The normalization unit is used to normalize the cleaned first historical charging behavior dataset to obtain the normalized first historical charging behavior dataset.

[0137] The statistical analysis unit is used to perform statistical analysis on the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after statistical analysis.

[0138] The first acquisition unit is used to acquire, based on the statistically analyzed first historical charging behavior dataset, a second historical charging behavior dataset of the power batteries with abnormal risks among the plurality of power batteries.

[0139] Optionally, in the aforementioned power battery safety early warning device, the statistical analysis unit is specifically used for:

[0140] A correlation analysis is performed on the normalized first historical charging behavior dataset, and historical charging behavior data with a Pearson correlation coefficient equal to a preset value are removed from the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after correlation analysis.

[0141] The first historical charging behavior dataset after statistical analysis is obtained by removing historical charging behavior data with variance less than a threshold from the first historical charging behavior dataset after removing data with correlation analysis.

[0142] Optionally, in the power battery safety warning device, the first obtaining submodule includes:

[0143] The second acquisition unit is used to acquire at least one target data corresponding to each historical charging behavior data in the second historical charging behavior dataset for each historical charging behavior data in the second historical charging behavior dataset.

[0144] The third acquisition unit is used to acquire the absolute value of the difference between the historical charging behavior data and the target data;

[0145] The obtaining unit is used to obtain one of the charging behavior synthetic data in the charging behavior synthetic dataset by summing the product of the absolute value and the random number with the historical charging behavior data.

[0146] Optionally, in the aforementioned power battery safety warning device, the second acquisition unit is specifically used for:

[0147] Obtain the Euclidean distance between the historical charging behavior data and the target historical charging behavior data; wherein, the target historical charging behavior data is the historical charging behavior data in the second historical charging behavior dataset that belongs to the same dimension as the historical charging behavior data;

[0148] Based on the Euclidean distance, obtain multiple nearest neighbor data of the historical charging behavior data;

[0149] According to the preset sampling rate, at least one target data is selected from the plurality of nearest neighbor data.

[0150] The power battery safety warning device provided in this embodiment of the invention can execute the above-described power battery safety warning method embodiment. Its implementation principle and technical effect are similar, and will not be described again here.

[0151] Embodiments of the present invention also provide a power battery safety warning device, including a processor, a memory, and a program or instructions stored in the memory and executable on the processor; when the processor executes the program or instructions, it implements the power battery safety warning method as described above, which will not be repeated here.

[0152] A specific embodiment of the present invention also provides a computer-readable storage medium storing a computer program thereon. When the program is executed by a processor, it implements the steps in the above-described power battery safety warning method and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0153] In addition, embodiments of the present invention also provide a computer program product, including computer instructions, which, when executed by a processor, implement the above-described... Figure 1 The various processes of the method embodiments shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0154] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0155] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can be physically comprised separately, or two or more units can be integrated into one unit. The integrated unit described above can be implemented in hardware or in the form of hardware plus software functional units.

[0156] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions that cause a computer device (which may be a personal computer, server, or network device, etc.) to execute some steps of the transmission and reception methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0157] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for early warning of power battery safety, characterized in that, include: Based on the power battery safety early warning model and the current charging behavior data of the target power battery, obtain a safety early warning result to indicate whether there is an abnormal risk to the target power battery; The power safety early warning model is obtained by training on a synthetic charging behavior dataset and a first historical charging behavior dataset of multiple power batteries; the synthetic charging behavior dataset is obtained by data synthesis processing based on a second historical charging behavior dataset of power batteries with abnormal risks among the multiple power batteries. Data synthesis processing is performed on the second historical charging behavior dataset to obtain the synthesized charging behavior dataset, including: For each historical charging behavior data in the second historical charging behavior dataset, obtain at least one target data in the second historical charging behavior dataset that corresponds to the historical charging behavior data. Obtain the absolute value of the difference between the historical charging behavior data and the target data; Based on the product of the absolute value and the random number, and the sum of the historical charging behavior data, one of the charging behavior synthetic data in the charging behavior synthetic dataset is obtained; Obtaining at least one target data point corresponding to the historical charging behavior data from the second historical charging behavior dataset includes: Obtain the Euclidean distance between the historical charging behavior data and the target historical charging behavior data; wherein, the target historical charging behavior data is the historical charging behavior data in the second historical charging behavior dataset that belongs to the same dimension as the historical charging behavior data; Based on the Euclidean distance, obtain multiple nearest neighbor data of the historical charging behavior data; According to the preset sampling rate, at least one target data is selected from the plurality of nearest neighbor data.

2. The power battery safety early warning method according to claim 1, characterized in that, The method further includes: The power battery safety early warning model is obtained by following these steps: Obtain the first historical charging behavior dataset of the plurality of power batteries; wherein, the number of power batteries with abnormal risks is less than the number of power batteries without abnormal risks. Based on the first historical charging behavior dataset, a second historical charging behavior dataset of the power batteries with abnormal risks among the plurality of power batteries is obtained. Data synthesis processing is performed on the second historical charging behavior dataset to obtain the synthesized charging behavior dataset; A training dataset is obtained based on the synthetic charging behavior dataset and the first historical charging behavior dataset; The preset classification model is trained based on the training dataset to obtain the power battery safety early warning model.

3. The power battery safety early warning method according to claim 2, characterized in that, Based on the first historical charging behavior dataset, a second historical charging behavior dataset is obtained for the power batteries among the plurality of power batteries that have abnormal risks, including: The first historical charging behavior dataset is cleaned to remove missing and abnormal data, resulting in a cleaned first historical charging behavior dataset. The first historical charging behavior dataset after cleaning is normalized to obtain the normalized first historical charging behavior dataset. Perform statistical analysis on the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after statistical analysis. Based on the first historical charging behavior dataset after statistical analysis, a second historical charging behavior dataset is obtained for the power batteries among the multiple power batteries that have abnormal risks.

4. The power battery safety early warning method according to claim 3, characterized in that, Perform statistical analysis on the normalized first historical charging behavior dataset to obtain the statistically analyzed first historical charging behavior dataset, including: A correlation analysis is performed on the normalized first historical charging behavior dataset, and historical charging behavior data with a Pearson correlation coefficient equal to a preset value are removed from the normalized first historical charging behavior dataset to obtain the first historical charging behavior dataset after correlation analysis. The first historical charging behavior dataset after statistical analysis is obtained by removing historical charging behavior data with variance less than a threshold from the first historical charging behavior dataset after removing data with correlation analysis.

5. A power battery safety early warning device, characterized in that, The device is used to implement the power battery safety warning method according to any one of claims 1 to 4, and the device comprises: The first acquisition module is used to acquire a safety warning result indicating whether there is an abnormal risk in the target power battery, based on the power battery safety warning model and the current charging behavior data of the target power battery. The power safety early warning model is obtained by training on a synthetic charging behavior dataset and a first historical charging behavior dataset of multiple power batteries; the synthetic charging behavior dataset is obtained by data synthesis processing based on a second historical charging behavior dataset of power batteries with abnormal risks among the multiple power batteries.

6. A power battery safety early warning device, comprising a processor, a memory, and a program or instructions stored in the memory and operable on the processor; characterized in that, When the processor executes the program or instructions, it implements the power battery safety early warning method as described in any one of claims 1 to 4.

7. A readable storage medium, characterized in that, The readable storage medium stores a program that, when executed by a processor, implements the power battery safety warning method as described in any one of claims 1 to 4.

8. A computer program product, characterized in that, It includes computer instructions, which, when executed by a processor, implement the power battery safety warning method as described in any one of claims 1 to 4.

Images