Battery thermal runaway anomaly detection method and battery management unit

Abstract

This invention provides a method for detecting battery thermal runaway anomalies and a battery management unit. During detection, operational data for each battery sub-module is acquired. Based on this data, the corresponding Channenzo entropy or Channenzo entropy change rate is obtained and compared with a set target threshold. When the parameter value in the sample to be detected exceeds the target threshold, it is considered thermal runaway anomaly data. Simultaneously, a Bayesian classifier is used for further data processing, and the abnormal data is output and an alert is issued, thereby effectively improving the accuracy of battery operational data detection and ensuring battery usability and safety performance.

Description

Technical Field

[0001] This invention relates to the field of battery technology, and in particular to a method for detecting abnormal thermal runaway in batteries and a battery management unit. Background Technology

[0002] With economic development and the continuous deepening of the new energy field, the battery industry has also developed and is meeting more of people's needs.

[0003] Compared to traditional energy industries, batteries, as a clean energy source, offer numerous advantages such as environmental friendliness and sustainable utilization, leading to their application in various fields. These include electric vehicles, electric cars, and electric control components, providing them with power. Under certain operating conditions and usage time, the temperature and performance of existing power batteries are controllable. However, when power batteries operate for extended periods or under complex conditions, such as collisions and / or overcharging or overload, they can become uncontrollable. Once thermal runaway occurs, the rapidly developing high temperatures and destructive force can burn through the battery pack cover, causing a fire and endangering passenger safety, resulting in a major accident. As market demands continue to expand, the energy density of lithium-ion and lead-acid batteries is constantly increasing to meet requirements such as longer driving range in automobiles. Therefore, the market demand for battery safety is also rising. Ensuring the safety and performance of different power batteries during operation, and real-time monitoring and alarming of abnormal heating points and thermal runaway within the battery, are among the most pressing issues that need to be addressed.

[0004] In summary, existing power batteries cannot detect and alarm for abnormal heating and thermal runaway within the battery in real time under complex operating conditions or during long-term operation, leading to safety issues during battery use. Summary of the Invention

[0005] This invention provides a method for detecting battery thermal runaway anomalies and a battery management unit to monitor internal thermal anomalies in real time, thereby ensuring the safety performance of the power battery during use and the safety of users.

[0006] To address the aforementioned technical problems, this invention provides a method for detecting abnormal thermal runaway in batteries, comprising the following steps:

[0007] Acquire the sample to be tested: detect and acquire the operating data of each battery sub-module in any battery pack;

[0008] Based on the operational data, extract and determine the aroma entropy or aroma entropy change rate corresponding to the sample to be detected;

[0009] A target threshold is set, and data in the sample to be detected is compared with the target threshold, if the data in the sample to be detected is greater than the target threshold, the data is thermal runaway abnormal data, and is output.

[0010] According to an embodiment of the present application, the step of "extracting and determining the aroma entropy or the aroma entropy change rate corresponding to the sample to be detected based on the operation data" further comprises the following steps:

[0011] A training sample set is generated based on the aroma entropy or the aroma entropy change rate and the first thermal runaway data;

[0012] A prediction model is set;

[0013] The training sample set is substituted into the prediction model for detection based on the prediction model, and probability values corresponding to the sample to be detected of different categories are obtained.

[0014] According to an embodiment of the present application, the prediction model comprises a Bayesian classifier, and the sample to be detected is detected based on the Bayesian classifier.

[0015] According to an embodiment of the present application, the thermal runaway abnormal probability values corresponding to the sample to be detected of different categories are extracted, and the thermal runaway abnormal probability values are compared with an abnormal reference value, if the thermal runaway abnormal probability value is greater than the abnormal reference value, the operation data is abnormal data.

[0016] According to an embodiment of the present application, the Bayesian classifier comprises a target function and a decision function;

[0017] The target function is:

[0018]

[0019] The decision function is:

[0020]

[0021] Wherein, c j is a category set formed by operation data of different categories in the battery sub-module, x i is a feature attribute set formed by operation data in different battery sub-modules, and the target function is a Gaussian distribution function.

[0022] According to an embodiment of the present application, when the Bayesian classifier is used to detect the sample to be detected, the following steps are further included:

[0023] Based on the target function and the decision function, the prior probability and the conditional probability corresponding to the different categories are calculated and obtained.

[0024] using a standard algorithm for standard normal distribution conversion, and looking up the probabilities corresponding to different category sets according to a normal distribution table;

[0025] obtaining posterior probabilities corresponding to the different category sets based on the decision function, and obtaining thermal runaway abnormal probability values corresponding to different categories.

[0026] According to an embodiment of the present application, when the thermal runaway abnormal probability value corresponding to different categories in the sample to be detected is greater than 0.7, it is a thermal runaway abnormality.

[0027] According to an embodiment of the present application, the step of "extracting and determining the shannon entropy or shannon entropy change rate corresponding to the operation data based on the operation data" further comprises:

[0028] determining a first shannon entropy corresponding to the operation data at a first time;

[0029] determining a second shannon entropy corresponding to the operation data at a second time;

[0030] According to the first shannon entropy and the second shannon entropy, the shannon entropy change rate at any two different times is obtained.

[0031] According to an embodiment of the present application, in the step of "detecting and obtaining operation data of each battery sub-module in any battery pack", the operation data includes parameter values of different category sets when each battery sub-module is working, wherein the parameter values of different category sets include voltage value, temperature value, current value, battery model, state of charge, and operation time.

[0032] According to a second aspect of the embodiment of the present application, a battery management unit is also provided, which comprises one or more programs that can be processed by one or more processors to implement the battery thermal runaway abnormality detection method provided in the embodiments of the present application.

[0033] The embodiment of the present application has the following beneficial effects: compared with the prior art, the embodiment of the present application provides a battery thermal runaway abnormality detection method. In the detection, the operation data of each battery sub-module is obtained, the shannon entropy or shannon entropy change rate is obtained based on the operation data, then a thermal runaway training sample is generated according to the shannon entropy or shannon entropy change rate, and compared with a set target threshold value, when the data in the thermal runaway training sample is greater than the target threshold value, it is thermal runaway abnormal data. At the same time, in the embodiment of the present application, the detection accuracy can be effectively improved, the operation data of the battery in the battery pack is judged, the abnormal data is output and warned, and the use and safety performance of the battery are ensured. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments or the prior art, the following will briefly introduce the drawings needed to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the application, and all other drawings obtained by those of ordinary skill in the art without creative work can also be obtained based on these drawings.

[0035] Figure 1 A structural schematic diagram of a battery module provided in the embodiments of the application;

[0036] Figure 2A A first battery thermal runaway abnormality detection method provided in the embodiments of the application;

[0037] Figure 2B Another thermal runaway abnormality flowchart provided in the embodiments of the application;

[0038] Figure 3 A detection step flowchart provided in the embodiments of the application;

[0039] Figure 4 A second battery thermal runaway abnormality detection method flowchart provided in the embodiments of the application;

[0040] Figure 5 A third battery thermal runaway abnormality detection method flowchart provided in the embodiments of the application;

[0041] Figure 6 A fourth battery thermal runaway abnormality detection method flowchart provided in the embodiments of the application;

[0042] Figure 7 A fifth battery thermal runaway abnormality detection method flowchart provided in the embodiments of the application;

[0043] Figure 8 A detection device and a corresponding battery management unit provided in the embodiments of the application. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the application, and the disclosure below provides different implementation manners or examples to realize different structures of the application. In order to simplify the application, the components and settings of specific examples are described below. In addition, the application provides various specific examples of processes and materials, and those of ordinary skill in the art can realize the application of other processes. All other embodiments obtained by those of ordinary skill in the art without creative work belong to the protection scope of the application.

[0045] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated.

[0046] With the continuous development of new energy technology, people have higher requirements for the performance and quality of batteries.

[0047] For new energy vehicles and other devices that require battery power, the performance of the battery will directly affect the working efficiency of the device. When new energy vehicles work for a long time or operate in complex working conditions, some abnormalities may occur inside the battery, such as voltage abnormalities, temperature abnormalities, and thermal runaway abnormalities. If these abnormalities cannot be detected and warned in time, it will cause serious harm to the device and even the user.

[0048] The present application provides a battery thermal runaway abnormality detection method and a battery management unit to effectively monitor the battery and ensure the performance and safety of the battery.

[0049] In the present application, the application scenario of the battery is illustrated by taking a power battery as an example, which can be applied in various fields, such as electric vehicles, electric transmission devices and other power equipment. Specifically, the power battery of an electric vehicle is taken as an example. The abnormality detection method of the battery in other devices or equipment is the same as that in the present application, which will not be described here.

[0050] Specifically, as shown in Figure 1 In the present application, Figure 1 A structural diagram of the battery module provided in the present application is shown. In an electric vehicle, the interior of the electric vehicle 100 is provided with a plurality of battery packs, such as two or more battery packs. For example, a first battery pack 101 and a second battery pack 102, and each of the first battery pack 101 and the second battery pack 102 includes a plurality of battery sub-modules 1011. In the present application, the battery sub-modules 1011 in each battery pack can be arranged in sequence, such as forming a longitudinal arrangement structure in each battery pack, as shown in Figure 1The arrangement structure is shown in the figure. When the battery pack is working normally, the anomaly detection method provided in this embodiment is used for detection, and the abnormal data is output to ensure the safety and performance of the battery module.

[0051] Specifically, such as Figure 2A As shown, Figure 2A This application provides a first method for detecting battery thermal runaway anomalies. The method includes the following steps when detecting battery thermal runaway anomalies:

[0052] Acquire the sample to be tested: detect and acquire the operating data of each battery sub-module in any battery pack;

[0053] Based on the operational data, extract and determine the aroma entropy or aroma entropy change rate corresponding to the sample to be detected;

[0054] Set a target threshold and compare the data in the sample to be tested with the target threshold. If the data in the sample to be tested is greater than the target threshold, the data is thermal runaway abnormal data and is output.

[0055] Specifically, during testing, in combination with Figure 1 In the battery pack, the operating data of each battery sub-module within each battery pack is first acquired. In this embodiment, when the battery is operating, it generates parameter values ​​corresponding to various categories of operating data, and the aforementioned sample to be tested is acquired. Specifically, the parameter values ​​corresponding to different categories of operating data in this embodiment may include: voltage value, temperature value, current value, battery model, charging status, vehicle speed, operating time, vehicle identification number (VIN), etc. In the following embodiments, voltage and temperature values ​​are used as examples for illustration; other categories of parameter values ​​can be processed in the manner described in this application. The voltage value also includes the total voltage value and the individual cell voltage value, and the current value includes the total current value and the individual cell current value.

[0056] Furthermore, when collecting each parameter value, it can be collected according to a periodic pattern, such as once every 10 seconds, or the collection interval can be shortened, such as once every 5 seconds; details will not be elaborated here. Simultaneously, in this embodiment, after collection, the collected data can be directly stored in the storage module within the battery module, or the collected data can be directly transmitted to the cloud to expand its application scenarios.

[0057] Specifically, the voltage entropy and temperature entropy change rate of each battery sub-module in the first battery pack 101 are extracted, as well as the voltage entropy and temperature entropy change rate of each battery module in the second battery pack 102.

[0058] After obtaining the different rich and strong entropies, different rich and strong entropy change rates can be obtained according to the rich and strong entropies. Specifically, a first rich and strong entropy corresponding to the operation data at a first time is determined;

[0059] A second rich and strong entropy corresponding to the operation data at a second time is determined;

[0060] According to the first rich and strong entropy and the second rich and strong entropy, the rich and strong entropy change rate at any two different times is obtained.

[0061] Specifically, in the embodiment of the application, when different rich and strong entropies are extracted and determined, the following formula is used:

[0062]

[0063] Wherein, H(X) represents the rich and strong entropy, xi is each possible value of the random variable, P(xi) is the probability of the value, b is a parameter, and commonly used b can be selected as 2, 10, e 2 , e (natural logarithm).

[0064] Further, by measuring the rich and strong entropies corresponding to the first time t i-1 and the second time t i , the first rich and strong entropy H(X) i-1 at the first time and the second rich and strong entropy H(X) i at the second time are obtained, and the rich and strong entropy change rate is obtained as follows:

[0065]

[0066] In the embodiment of the application, when the operation data of each battery sub-module is extracted, the extracted feature attribute is a continuous value and obeys a Gaussian distribution, that is:

[0067]

[0068] According to the above rules, the following feature attribute corresponding set is finally obtained:

[0069]

[0070] And a category set c is obtained according to different categories of operation data:

[0071] c = {c1, c2, c3, c4, c5} = {battery thermal runaway abnormality, voltage collection abnormality, temperature collection abnormality, BMS system abnormality, normal}

[0072] Specifically, by performing multiple tests and extractions on multiple different battery sub-modules, multiple different aroma entropies or aroma entropy change rates are obtained. Further, based on the aforementioned aroma entropies and aroma entropy change rates, and different feature attributes, a sample to be tested is generated. This sample to be tested includes data of various categories, such as normal sample data and thermal runaway data. Simultaneously, in this embodiment, the sample to be tested is obtained and combined with the first thermal runaway data to generate a training sample set. Specifically, as shown in Table 1, Table 1 is the training sample set corresponding to different categories of data provided in this embodiment:

[0073] Table 1:

[0074]

[0075] In Table 1, the tags corresponding to IDs 1 and 2 are thermal runaway data, specifically the first thermal runaway data, which refers to data indicating a confirmed thermal runaway anomaly in the battery. This first thermal runaway data can be obtained directly from a database or from common knowledge in the field. Furthermore, the aforementioned first thermal runaway data, along with other detected aromatization entropy or aromatization entropy change rates, forms a training sample set. To improve the accuracy of this training sample set, this embodiment further includes the following steps during processing:

[0076] Set up a prediction model;

[0077] Based on the prediction model, the training sample set is substituted into the prediction model for detection, and the probability values ​​corresponding to the samples to be detected of different categories are obtained.

[0078] Specifically, in the following embodiments, the prediction model uses a Bayesian classifier for detection. Meanwhile, other models that meet the detection and training functions can also be used to detect the aforementioned samples.

[0079] Furthermore, such as Figure 2B As shown, Figure 2B This is a schematic diagram illustrating another thermal runaway anomaly provided in an embodiment of this application. Combined with... Figure 2A In the processing flow of this application embodiment, the first step is to acquire samples: vehicle frame number, message time, all cell voltages, and all cell temperatures. Specifically, in this application embodiment, the operating data of each battery sub-module in any battery pack is detected and acquired; then, feature extraction is performed: based on the operating data, the Shannon entropy or Shannon entropy change rate corresponding to the sample to be detected is extracted and determined; further, the sample to be detected is substituted into a Bayesian classifier, and the prior probability of different categories and the probability of the feature taking various values ​​under different category conditions are calculated. After the calculation is completed, the thermal runaway event is judged, and the abnormal thermal runaway data is output.

[0080] Preferably, after detecting the to-be-detected sample by the Bayesian classifier, the thermal runaway abnormal probability value corresponding to the to-be-detected sample of different categories is extracted, and the thermal runaway abnormal probability value is compared with an abnormal reference value. If the thermal runaway abnormal probability value is greater than the abnormal reference value, the operation data is abnormal data. In the embodiment of the present application, the abnormal reference value can be set according to different types of battery models. For example, the abnormal reference value is the same as the target threshold value. When the thermal runaway abnormal probability value or the maximum probability is greater than the abnormal reference value or the target threshold value, it indicates that the corresponding data is thermal runaway abnormal data, and the thermal runaway event will occur. Otherwise, it is normal data. At the same time, the abnormal reference value can be a threshold value set by comparison to determine abnormal data, which will not be described here.

[0081] In the Bayesian classifier, a target function and a decision function are included.

[0082] Specifically, the target function is:

[0083]

[0084] The decision function is:

[0085]

[0086] Wherein, c j is a category set formed by operation data of different categories in the battery sub-module, x i is a feature attribute set formed by operation data in different battery sub-modules, and the target function is a Gaussian distribution function. For example, when j = 1, c1 corresponding to it is battery thermal abnormality, when j = 2, c2 corresponding to it is voltage collection abnormality...

[0087] When detecting the thermal runaway training sample, as shown in Figure 3 , the detection steps provided in the embodiment of the present application are as follows: Figure 3

[0088] Based on the target function and the decision function, the prior probability and the conditional probability corresponding to the different category set are calculated and obtained;

[0089] The standard normal distribution conversion is performed by using the standard score algorithm, and the probabilities corresponding to different categories are found out according to the normal distribution table;

[0090] Based on the decision function, the posterior probability corresponding to the different categories is obtained, and the thermal runaway abnormal probability in the different categories is obtained.

[0091] ​Specifically, when calculating the prior probability, the following calculation formula is used:

[0092]

[0093]

[0094]

[0095]

[0096]

[0097] Meanwhile, when calculating the conditional probability, the following formula provided in the embodiments of the present application is used:

[0098]

[0099]

[0100] At this time, the standard normal distribution conversion is performed by using the standard score algorithm. Specifically, the conversion formula is as follows:

[0101] Since the above conversion conforms to the standard normal distribution, the specific probability value of the above p 模组1电压香浓熵变化率|热失控 can be directly obtained from the probability table corresponding to the standard normal distribution.

[0102] According to the above training rules and calculation scheme, the probabilities of different feature attributes under different category sets can be obtained in sequence.

[0103] Further, when the above probability values and corresponding different parameters are obtained, the posterior probability corresponding to different categories is obtained based on the decision function.

[0104] Specifically, the calculation formula of the posterior probability is as follows:

[0105]

[0106]

[0107]

[0108]

[0109]

[0110] In the embodiments of the present application, when the data samples corresponding to different categories in the above to-be-detected sample are detected, multiple different probabilities can be obtained. ​

[0111] The different probabilities are screened. Specifically, when screening, the thermal runaway abnormal probability obtained by the same category is extracted. When extracting, the following screening formula is used for screening:

[0112] g 热失控 (x)>g 电压采集异常 (x)

[0113] g 热失控 (x)>g 温度采集异常 (x)

[0114] g 热失控 (x)>g BMS系统异常 (x))

[0115] g 热失控 (x)>g 电池正常 (x)。

[0116] g 热失控(x) In actual working conditions, when g thermal runaway (x) ≥ 0.7, the corresponding operation data is thermal runaway abnormal data. When the above thermal runaway abnormal probability value is greater than the target threshold 0.7, the corresponding battery point in the battery pack is prone to thermal runaway, and the thermal runaway event occurs. Therefore, the abnormal data point, such as thermal runaway abnormal data, is obtained by the above detection method in the embodiment of the application. And output, thereby effectively warning the abnormality of the battery and ensuring the safety performance of the battery module. In the embodiment of the application, the above thermal runaway and thermal runaway abnormality are the same concept, both indicating that the thermal runaway event occurs, which will not be described here.

[0117] Further, in the embodiment of the application, when setting the above target threshold, it can also be set according to different products. For example, when the sensitivity of the equipment to the battery performance is high, the target threshold can be lowered. When part of the thermal runaway abnormal probability detected is greater than the target threshold, it is considered as abnormal data. At the same time, combined with actual consideration, in order to improve the accuracy, an abnormal continuous event can be set. For example, based on the current data thermal abnormality probability A, it is determined that the battery pack has a thermal runaway abnormality. In the embodiment of the application, it is considered that a thermal abnormality or thermal runaway occurs when the thermal runaway abnormality probability value exceeds 70%. Specifically, according to actual experience, the above condition can be triggered, such as 2 times in a row or 30S, triggering thermal runaway warning.

[0118] In the embodiment of the application, the model is trained by the above thermal runaway training sample, thereby effectively improving the accuracy of detection and ensuring the safety performance of the battery.

[0119] Specifically, as shown in Figure 4 and Figure 5 , Figure 4A flowchart of a second battery thermal runaway anomaly detection method provided in an embodiment of the present application is shown in FIG. 2. Figure 5 A flowchart of a third battery thermal runaway anomaly detection method provided in an embodiment of the present application is shown in FIG. 3.

[0120] In the method, the change rate of the Gini entropy is extracted and detected. Figure 4 In the method, the change rate of the Gini entropy is extracted and detected. Figure 5 In the method, the change rate of the Gini entropy is extracted and detected.

[0121] Specifically, when the Gini entropy corresponding to each different operating data in the battery module is extracted, the Gini entropy of different categories in the sample to be detected can be compared with the target threshold value in the manner of the change rate of the Gini entropy. When the Gini entropy corresponding to the different operating data is greater than the target threshold value, it is determined that the thermal runaway anomaly data occurs.

[0122] In the embodiment of the present application, when the target threshold value is determined, the data set by the test vehicle can be used for determination. For example, when the Gini entropy of the voltage in the same module in the battery pack is greater than or equal to the target threshold value 1.2 and the Gini entropy of the temperature in the module is greater than or equal to the target threshold value 0.8, it is determined that the thermal runaway event occurs.

[0123] Meanwhile, in the embodiment of the present application, when the change rate of the Gini entropy is used for detection, the data set by the test vehicle can be used for determination. For example, when the change rate of the Gini entropy of the voltage in the same module in the battery pack and the change rate of the Gini entropy of the temperature in the module are both greater than the target threshold value, it is determined that the thermal runaway event occurs.

[0124] Further, as shown in FIG. 4, Figures 6-7 Figure 6 A flowchart of a fourth battery thermal runaway anomaly detection method provided in an embodiment of the present application is shown in FIG. 4. Figure 7 A flowchart of a fifth battery thermal runaway anomaly detection method provided in an embodiment of the present application is shown in FIG. 5. Specifically, in the embodiment of the present application, first, the sample to be detected is obtained: the operating data of each battery sub-module in the battery pack is obtained: for example, the vehicle frame number, the message time, and the operating data of all single cell temperature values of the battery sub-module of the sample to be detected are detected and obtained.

[0125] After the operating data is obtained, the features are extracted based on the operating data: in the embodiment of the present application, the Gini entropy or the change rate of the Gini entropy corresponding to the operating data is extracted and determined. For example, the temperature Gini entropy of all modules in the battery pack and the change rate of the temperature Gini entropy of all modules in the battery pack are obtained.

[0126] ​Further, a training sample set is generated according to the data extracted from the above features and the first thermal runaway data, and a to-be-detected sample in the training sample set is brought into the Bayesian classifier for detection. In the embodiment of the present application, when detection is performed in the Bayesian classifier, the prior probability corresponding to the running data of different categories in the to-be-detected sample and the probability when other features take various values are calculated.

[0127] After the training is completed, the probability values corresponding to different categories can be obtained, such as the thermal runaway abnormal probability value, the temperature collection abnormal probability value, and the BMS system abnormal probability value. At the same time, the different probability values are compared with the target threshold value, and if the probability value is greater than the target threshold value, it is regarded as abnormal data, so as to be screened and output.

[0128] Further, in the Figure 6 , in combination with the flowchart in Figure 5 , in the embodiment of the present application, after the features corresponding to the running data in the battery are extracted, preliminary screening is performed, such as setting a first target threshold value or a plurality of different first target threshold values, and different first target threshold values can correspond to running data of different categories. For example, the voltage entropy in the same module corresponds to a first target threshold value, and the temperature entropy in the same module corresponds to another first target threshold value. The size relationship between the extracted voltage entropy or temperature entropy and the first target threshold value is judged: when the voltage entropy and the temperature entropy are both greater than the first target threshold value, or when the voltage entropy change rate and the temperature entropy change rate are both greater than another target threshold value, the data meeting the conditions are detected by the Bayesian classifier. If the voltage entropy and the temperature entropy are not greater than the first target threshold value, the battery running data is re-detected. Then the judgment is performed again until the conditions are met. Then the thermal runaway abnormal probability value corresponding to the to-be-detected sample of different categories is obtained, and the thermal runaway probability value is compared with a second target threshold value, which corresponds to the minimum value of the abnormal probability of the battery running data in the embodiment of the present application. Whether the corresponding running data is thermal runaway abnormal data is determined.

[0129] In the embodiment of the present application, according to the experimental vehicle data, in order to improve the accuracy, the first target threshold value and the second target threshold value can be adjusted, such as the first target threshold value corresponding to the module voltage entropy ≥ 1, the first target threshold value corresponding to the module temperature entropy ≥ 0.5, the relationship between the thermal runaway abnormal probability and the target threshold value is determined after the Bayesian decision, and whether the thermal runaway event occurs is determined. In combination with the practice, in order to reduce the false positives, it is determined that the thermal runaway abnormal event occurs only when the thermal abnormal probability event probability is maximum and the thermal abnormal event probability ≥ the second target threshold value 0.7.

[0130] In the embodiment of the present application, the above-mentioned Bayesian classifier is used to detect the sample to be detected, thereby effectively improving the accuracy of the screening data and ensuring the efficiency of the anomaly detection.

[0131] Further, as shown in Figure 8 , Figure 8 A detection device and a corresponding battery management unit are provided in the embodiment of the present application. Specifically, the battery management unit 610 can include a first battery pack 101 and a plurality of battery sub-modules 1011 arranged in the first battery pack 101. Meanwhile, a processor 60 is arranged in the battery management unit, which can be arranged at the edge end of the first battery pack 101. The processor 60 is connected with a corresponding detection terminal. After the detection terminal completes the detection of the data in the battery pack, the detected data can be directly transmitted to the processor 60. In the embodiment of the present application, the processor 60 can further include a storage module. The above-mentioned data can be stored through the storage module, or the above-mentioned data can be directly transmitted to the cloud through the processor 60, and the above-mentioned data can be stored and operated in the cloud.

[0132] In the embodiment of the present application, the operating data of the battery detected is transmitted to the cloud, and different processing is performed in the cloud, thereby effectively improving the use environment and protecting the data, preventing the data loss caused by the problems of the battery itself.

[0133] In the embodiment of the present application, one or more programs can be arranged in the battery management unit. The above-mentioned programs are used to control the processor and detect the battery, and the abnormal operating data is processed according to the requirements, thereby ensuring the use performance and safety performance of the battery.

[0134] In summary, the above describes the battery thermal runaway anomaly detection method and the battery management unit provided by the embodiment of the present application in detail. The principle and implementation mode of the present application are described by using specific examples. The above embodiment is only used to help understand the technical scheme of the present application and its core idea. Although the above-mentioned preferred embodiment is disclosed, the above-mentioned preferred embodiment is not used to limit the present application. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present application. Therefore, the protection scope of the present application is based on the range defined by the claims.

Claims

1. A method for detecting abnormal thermal runaway in batteries, characterized in that, Includes the following steps: Acquire the sample to be tested: detect and acquire the operating data of each battery sub-module in any battery pack; Based on the operational data, extract and determine the aroma entropy or aroma entropy change rate corresponding to the sample to be detected; generate a training sample set based on the aroma entropy or aroma entropy change rate and the first thermal runaway data, set a prediction model, and substitute the training sample set into the prediction model for detection based on the prediction model to obtain the probability values ​​corresponding to the sample to be detected of different categories. Set a target threshold and compare the data in the sample to be tested with the target threshold. If the data in the sample to be tested is greater than the target threshold, the data is thermal runaway abnormal data and is output.

2. The method for detecting battery thermal runaway anomalies according to claim 1, characterized in that, The prediction model includes a Bayesian classifier, which is used to detect the sample to be detected.

3. The method for detecting battery thermal runaway anomalies according to claim 2, characterized in that, Extract the thermal runaway anomaly probability values ​​corresponding to the samples to be tested of different categories, and compare the thermal runaway anomaly probability values ​​with the anomaly reference values. If the thermal runaway anomaly probability value is greater than the anomaly reference value, then the running data is abnormal data.

4. The method for detecting battery thermal runaway anomalies according to claim 2, characterized in that, The Bayesian classifier includes an objective function and a decision function; The objective function is: The decision function is: Among them, c j x is the set of categories formed by different types of operating data in the battery sub-module. i The set of characteristic attributes formed by the operating data within different battery sub-modules, and the objective function is a Gaussian distribution function.

5. The method for detecting battery thermal runaway anomalies according to claim 4, characterized in that, When using the Bayesian classifier to detect the sample to be detected, the following steps are also included: Based on the objective function and the decision function, the prior probability and conditional probability corresponding to the different categories are calculated and obtained; The standard score algorithm is used to transform the standard normal distribution, and the probabilities corresponding to different class sets are found according to the normal distribution table. Based on the decision function, the posterior probabilities corresponding to the different category sets are obtained, and the thermal runaway anomaly probability values ​​corresponding to the different categories are obtained.

6. The method for detecting battery thermal runaway anomalies according to claim 5, characterized in that, When the probability value of different categories of thermal runaway anomalies in the sample to be tested is greater than 0.7, it is considered a thermal runaway anomaly.

7. The method for detecting battery thermal runaway anomalies according to claim 1, characterized in that, The step "based on the operational data, extracting and determining the entropy or rate of change of entropy corresponding to the operational data" further includes: Determine the first entropy corresponding to the running data at the first moment; Determine the second entropy corresponding to the running data at the second time point; Based on the first aroma entropy and the second aroma entropy, the rate of change of the aroma entropy at any two different times is obtained.

8. The method for detecting battery thermal runaway anomalies according to claim 1, characterized in that, In the step of "detecting and acquiring the operating data of each battery sub-module in any battery pack", the operating data includes different categories of parameter values ​​corresponding to each battery sub-module when it is working. The different categories of parameter values ​​include: voltage value, temperature value, current value, battery model, charging status, and running time.

9. A battery management unit, characterized in that, The battery management unit includes one or more programs, which can be processed by one or more processors to implement the battery thermal runaway anomaly detection method according to any one of claims 1 to 8.

Images