Safety early warning method and equipment for power battery

Abstract

The invention relates to the technical field of power battery safety, in particular to a safety early warning method and device for a power battery, and the method specifically comprises the following steps: S1, marking a sample, and obtaining the basic information and operation data of the carried battery; s2, screening discharge condition data and cleaning the data; s3, setting a sliding window; s4, calculating an entropy value of voltage ranking of each monomer in each group; s5, calculating the entropy slope of the entropy of the voltage ranking of each monomer in each group along with the change of mileage, and preliminarily positioning the risk battery cell; s6, feature parameters of all samples are extracted and calculated; s7, training the model by adopting a random forest algorithm; and S8, collecting the data set marked as the to-be-evaluated vehicle data set, extracting the entropy slope of each battery cell of the to-be-analyzed battery, and judging an abnormal battery cell. According to the safety early warning method of the power battery provided by the invention, the abnormal single battery cell possibly having an internal short circuit risk can be identified in time, and the accuracy and reliability of safety early warning are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power battery safety, and in particular to a power battery safety early warning method and device. Background Art

[0002] With the popularity of electric vehicles and the continuous development of power battery technology, battery safety issues have gradually become the focus of attention in the new energy industry. As the main energy storage component of electric vehicles, the performance of lithium-ion batteries directly determines the safety, mileage and service life of the vehicle. However, the process problems existing in the production and manufacturing process of lithium-ion batteries and the attenuation of the comprehensive performance of batteries during use may lead to the occurrence of internal short circuit failures.

[0003] Internal short circuit faults have the characteristics of unclear early fault characteristics, difficult to detect, and difficult to warn. Once they occur, they may cause thermal runaway and even lead to serious consequences such as battery explosion, which seriously threatens user safety. By analyzing the thermal runaway battery data, it is found that in most cases, before the thermal runaway, the key monitoring indicators such as voltage, temperature, and internal resistance of the battery do not show significant abnormal signs, so the traditional early warning mechanism fails and cannot capture potential crises in time. In a small number of samples, data of abnormal temperature rise and abnormal voltage drop can be uploaded briefly at the moment of thermal runaway, but the appearance of these signals often means that thermal runaway has occurred, resulting in irreversible consequences. Existing battery safety early warning methods and technical solutions are mostly based on abnormal characteristic parameters such as voltage, temperature, capacity, and internal resistance as the research and development basis, and then combined with abnormal screening algorithms to achieve safety early warning. Although these strategies can effectively capture the evolutionary thermal runaway caused by significant deviation of parameters from normal values, they cannot identify the sudden thermal runaway of batteries whose voltage is within the normal range before thermal runaway, the pressure difference is small (within 50mV), and the temperature, capacity, internal resistance, insulation and other parameters are not obviously abnormal.

[0004] For example, CN202210089340.0 needs to calculate the voltage extreme difference, and then obtain the entropy value by dividing the maximum value of the extreme difference voltage by the minimum value. It is not applicable to thermal runaway batteries with very small pressure difference; CN202310446506.4 calculates the safety entropy of parameters such as voltage, temperature, capacity, internal resistance, and fault and obtains the total safety entropy. This scheme evaluates the overall safety status of the battery through multiple dimensions and lacks an accurate early warning mechanism for sudden thermal runaway; The CN202210386074.8 scheme uses voltage ranking to find the relative changes between different monomers. Although this characteristic parameter does not directly depend on the absolute pressure difference, for battery systems with small pressure difference, the voltage ranking fluctuations in different charging processes will be large, and only the change in slope will result in false alarms.

[0005] Therefore, developing more advanced and effective internal short-circuit detection technology is of great significance to improving the safety of power batteries. Summary of the invention

[0006] In order to overcome the problem of the inability to accurately predict the safety of power batteries in the prior art, the purpose of the present invention is to provide a safety warning method and device for power batteries, which predicts the safety of power batteries by calculating the entropy value of the frequency of occurrence of each single cell voltage ranking.

[0007] The present invention adopts the following technical solutions:

[0008] A safety early warning method for a power battery comprises the following steps:

[0009] S1: Mark samples, which include abnormal vehicle data sets, normal vehicle data sets, and vehicle data sets to be evaluated, and obtain basic information and operating data of the batteries: battery rated capacity, rated voltage, maximum allowable charging voltage, material system, number of series and parallel connections, number of temperature probes, historical operating data, message time, maximum value of battery cell voltage, maximum battery cell voltage number, minimum battery cell voltage value, minimum battery cell voltage number, voltage value of each cell, temperature value of each temperature sensor, current, total voltage, cumulative mileage, SOC, and charge and discharge status.

[0010] S2: Filter the discharge condition data and perform data cleaning, delete abnormal values ​​and duplicate values ​​such as voltage, current, mileage, etc., and extract the voltage value of each battery cell;

[0011] S3: Set the sliding window S to group all the discharge condition data, that is, the first frame to the Sth frame is the first group, the S+1th frame to the 2Sth frame is the second group, and so on, the (n-1)S+1th frame to the nSth frame is the nth group, and the nS+1th to the last frame is the n+1th group, with groups as units;

[0012] S4: Calculate the entropy value H of the voltage ranking of each monomer i in each group i , and obtain the change law of entropy value with mileage in the whole life cycle. The specific operation is: according to the sorting algorithm, the voltage ranking value VRcell of each cell is calculated from small to large i , calculate the number of times each monomer i appears in each ranking value j T ij and frequency F ij , record the number of cells N, entropy value H i The calculation formula is:

[0013]

[0014] S5: Calculate the entropy slope k of the voltage ranking of each cell i in each group as the mileage changes i, normalize the slope values ​​of different cells at each mileage point, and calculate the Z-score to record the entropy slope of each cell as Z i ,

[0015] The specific operation is to calculate the average value μ and standard deviation σ of the entropy slope of all cells, and normalize the entropy slope When Z i When the minimum value is lower than the preset value, abnormal cells R are preliminarily screened out.

[0016] Furthermore, the method further comprises the following steps:

[0017] S6: extract and calculate the characteristic parameters of all samples;

[0018] S7: Collect data sets marked as abnormal vehicles and normal vehicles, extract corresponding key feature parameters, and use random forest algorithm to train the model;

[0019] S8: Collect the vehicle data set marked as to be evaluated, extract the lowest value of the entropy slope Z-score of each battery cell to be analyzed, if it is less than the preset value, extract the corresponding feature parameters and input them into the random forest model trained in step S7 to output the abnormal vehicle probability value, when the predicted probability of the abnormal vehicle exceeds the preset threshold MT, the battery cell is identified as a risky battery cell.

[0020] Furthermore, the characteristic parameters of step S6 include: the minimum value Z of the normalized entropy slope of the battery cell R Rmin , the minimum value Z of the normalized entropy slope of all cells except cell R imin 、The rate value V at which the maximum value of the cell R voltage ranking drops to the minimum value R , the difference between the maximum voltage value and the voltage of the battery cell R and the rate of change with mileage C R , the rate value Vk of the normalized entropy slope value of the battery R from the maximum value to the minimum value R , the duration of the minimum value of the normalized entropy slope of the battery cell R, and the minimum value number of the single cell voltage.

[0021] Furthermore, the specific operation of using the random forest algorithm to train the model in step S7 is:

[0022] S701: further divide the data set into a training set and a test set, the training set is used for model training, and the test set is used to evaluate the performance of the model;

[0023] S702: Use the random forest algorithm to train the training set data. During the training process, it is necessary to adjust the parameters of the model: the number of trees and the maximum depth of the trees to optimize the model performance;

[0024] S703: Use the test set data to evaluate the performance of the trained model.

[0025] Furthermore, the entropy slope preset value of step S5 and step S8 is set to -6σ.

[0026] The preset threshold MT of step S8 is 0.8.

[0027] The number of trees in the random forest algorithm is set to 100 and the maximum depth of the tree is set to 15.

[0028] The present invention also discloses another technical solution:

[0029] A safety warning device for a power battery, comprising a memory and a processor,

[0030] The memory stores a safety warning program for the power battery.

[0031] The processor executes the power battery safety warning method of the first technical solution according to the power battery safety warning program in the memory.

[0032] The present invention has at least the following beneficial effects:

[0033] 1. The present invention provides a safety early warning method for a power battery. By real-time monitoring of the changes in the voltage ranking of each cell of the power battery and calculating the entropy value, the method can timely identify abnormal cells that may have internal short circuit risks. This early warning mechanism significantly improves the safety of electric vehicles and helps prevent serious safety accidents caused by battery failures, such as thermal runaway, fire, and even explosion.

[0034] 2. The safety warning method for power batteries provided by the present invention does not rely on significant abnormalities in parameters such as voltage, temperature, capacity, and internal resistance to issue an alarm. Instead, it captures abnormal cells with small voltage differences but potential internal short circuit risks by analyzing the entropy changes in voltage ranking, thereby improving the accuracy and reliability of safety warnings;

[0035] 3. The present invention provides a safety early warning method for power batteries, which can extend the service life of batteries, reduce maintenance costs, and improve the overall performance and reliability of electric vehicles by identifying and intervening in potential safety hazards in advance. In addition, this method can also provide valuable data for battery manufacturers to help them improve battery design and production processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A flow chart of a safety early warning method for a power battery provided by the present invention;

[0037] Figure 2 A random forest model training flow chart of a power battery safety early warning method provided by the present invention;

[0038] Figure 3 This is a result diagram of the entropy value of the voltage ranking of each monomer calculated in the second embodiment of the present invention;

[0039] Figure 4 This is a graph showing the entropy slope results of each monomer calculated in Example 2 of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] Embodiment 1

[0042] Combination Figure 1 and Figure 2 As shown, the present invention discloses a safety early warning method for a power battery, and the specific operation steps are as follows:

[0043] S1: Mark samples, which include abnormal vehicle data sets, normal vehicle data sets, and vehicle data sets to be evaluated, and obtain basic information and operating data of the batteries: battery rated capacity, rated voltage, maximum allowable charging voltage, material system, number of series and parallel connections, number of temperature probes, historical operating data, message time, maximum value of battery cell voltage, maximum battery cell voltage number, minimum battery cell voltage value, minimum battery cell voltage number, voltage value of each cell, temperature value of each temperature sensor, current, total voltage, cumulative mileage, SOC, and charge and discharge status.

[0044] S2: Filter the discharge condition data and perform data cleaning, delete abnormal values ​​and duplicate values ​​such as voltage, current, mileage, etc., and extract the voltage value of each battery cell;

[0045] S3: Set the sliding window S to group all the discharge condition data, that is, the first frame to the Sth frame is the first group, the S+1th frame to the 2Sth frame is the second group, and so on, the (n-1)S+1th frame to the nSth frame is the nth group, and the nS+1th to the last frame is the n+1th group, with groups as units;

[0046] S4: Calculate the entropy value H of the voltage ranking of each monomer i in each group i , and obtain the change law of entropy value with mileage in the whole life cycle. The specific operation is: according to the sorting algorithm, the voltage ranking value VRcell of each cell is calculated from small to large i , calculate the number of times each monomer i appears in each ranking value j T ij and frequency Fij , Record the number of cells N and entropy value H i The calculation formula is:

[0047]

[0048] S5: Calculate the entropy slope k of the voltage ranking of each cell i in each group as the mileage changes i , normalize the slope values ​​of different cells at each mileage point, and calculate the Z-score to record the entropy slope of each cell as Z i ,

[0049] The specific operation is to calculate the average value μ and standard deviation σ of the entropy slope of all cells.

[0050] in,

[0051] Normalized entropy slope When Z i When the minimum value is lower than -6σ, the abnormal cell R is initially screened out; otherwise, it returns to S1 and adds new operating data to start again;

[0052] S6: Extract and calculate the characteristic parameters of all samples, including the minimum slope value Z of the normalized entropy value of the battery cell R Rmin , the minimum value Z of the normalized entropy slope of all cells except cell R imin 、The rate value V at which the maximum value of the cell R voltage ranking drops to the minimum value R , the difference between the maximum voltage value and the voltage of the battery cell R and the rate of change with mileage C R , the rate value Vk of the normalized entropy slope value of the battery R from the maximum value to the minimum value R , the duration of the minimum value of the normalized entropy slope of the battery cell R, and the number of the minimum value of the single cell voltage;

[0053] S701: Collect data sets marked as abnormal vehicles and normal vehicles, and extract corresponding key feature parameters;

[0054] S702: further dividing the data set into a training set and a test set, where the training set is used for model training and the test set is used for evaluating the performance of the model;

[0055] S703: Use the random forest algorithm to train the training set data. During the training process, it is necessary to adjust the parameters of the model: the number of trees and the maximum depth of the trees to optimize the model performance;

[0056] S704: Use the test set data to evaluate the performance of the trained model.

[0057] S8: Collect the vehicle data set marked as to be evaluated, extract the lowest value of the entropy slope Z-score of each battery cell to be analyzed, if it is less than -6σ, extract the corresponding feature parameters and input them into the random forest model trained in step S704 to output the abnormal vehicle probability value. When the predicted probability of an abnormal vehicle exceeds the preset threshold of 0.8, the battery cell is identified as a risky battery cell. When the predicted probability of an abnormal vehicle does not exceed MT, the newly added operating data returns to step S1 to start a new round of early warning.

[0058] Embodiment 2

[0059] This embodiment performs a safety warning for vehicle No. 1 according to the steps of the first embodiment, and the specific operation steps are as follows:

[0060] Step 1: Collect the abnormal vehicle data set, normal vehicle data set, and vehicle data set to be evaluated of vehicle No. 1, and obtain the basic information and operating data of the battery carried by vehicle No. 1;

[0061] Step 2: Filter the discharge condition data, perform data cleaning, and delete abnormal and duplicate values ​​such as voltage, current, and mileage;

[0062] Step 3: Extract the voltage value of each battery cell, set the sliding window to 500, and group all the discharge data, that is, the first frame to the 500th frame are the first group, the 501st frame to the 1000th frame are the second group, and so on, proceed to the next step in groups;

[0063] Step 4: Calculate the entropy value H of the voltage ranking of each cell in each group, and obtain the change law of entropy value with mileage in the whole life cycle. The calculated result is as follows Figure 3 As shown;

[0064] Step 5: Calculate the entropy slope k of the entropy value of each cell voltage ranking in each group as the mileage changes, and normalize the slope values ​​of different cells at each mileage point. By calculating the Z-score, the entropy slope of each cell is recorded as Z. The calculation results are as follows: Figure 4 As shown, when the lowest value of Z is lower than -6σ, the abnormal cell is initially screened out and recorded as 50#;

[0065] Step 6: Extract and calculate the characteristic parameters of all samples. The values ​​of each parameter are as follows:

[0066] <![CDATA[Z Rmin ]]> <![CDATA[Z imin ]]> <![CDATA[V R ]]> <![CDATA[C R ]]> <![CDATA[Vk R ]]> t Minimum value of single cell voltage -9.19 -3.48 92 15 0.2 196d 50

[0067] Step 7: Collect data sets marked as abnormal vehicles and normal vehicles, extract corresponding key feature parameters, and use random forest algorithm to train the model;

[0068] Step 8: Collect the vehicle data set marked as to be evaluated, extract the lowest value of the entropy slope Z-score of each battery cell to be analyzed, if it is less than -6σ, extract the corresponding feature parameters and input them into the random forest model trained in step 7 to output the abnormal vehicle probability value. When the predicted probability of an abnormal vehicle exceeds the preset threshold of 0.8, the battery cell is identified as a risky battery cell and needs to be disposed of immediately.

[0069] Embodiment 3

[0070] The present invention also discloses a safety warning device for a power battery, comprising a memory and a processor.

[0071] The memory stores a safety warning program for the power battery.

[0072] The processor executes the power battery safety warning method of the first embodiment according to the power battery safety warning program in the memory.

[0073] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A safety early warning method for a power battery, characterized in that: The following steps are involved: S1: Mark samples to obtain basic information and operating data of the batteries carried. The samples include abnormal vehicle data sets, normal vehicle data sets, and vehicle data sets to be evaluated; S2: Filter the discharge condition data and perform data cleaning, delete abnormal values ​​and duplicate values, and extract the voltage value of each battery cell; S3: Set the sliding window to group all the discharge condition data into groups; S4: Calculate the entropy value H of the voltage ranking of each monomer i in each group i , we can get the change law of entropy value with mileage in the whole life cycle; S5: Calculate the entropy slope k of the voltage ranking of each cell i in each group as the mileage changes i , normalize the slope values ​​of different cells at each mileage point, and calculate the Z-score to record the entropy slope of each cell as Z i , when Z i When the minimum value is lower than the preset value, abnormal cells R are preliminarily screened out.

2. A safety warning method for a power battery according to claim 1, characterized in that: The following steps are also included: S6: extract and calculate the characteristic parameters of all samples; S7: Collect data sets marked as abnormal vehicles and normal vehicles, extract corresponding key feature parameters, and use random forest algorithm to train the model; S8: Collect the vehicle data set marked as to be evaluated, extract the lowest value of the entropy slope Z-score of each battery cell to be analyzed, if it is less than the preset value, extract the corresponding feature parameters and input them into the random forest model trained in step S7 to output the abnormal vehicle probability value, when the predicted probability of the abnormal vehicle exceeds the preset threshold MT, the battery cell is identified as a risky battery cell.

3. A safety warning method for a power battery according to claim 1, characterized in that: The basic information and operation data of the battery carried in step S1 include: battery rated capacity, rated voltage, maximum allowable charging voltage, material system, number of series and parallel connections, number of temperature probes, historical operation data, message time, maximum value of battery cell voltage, maximum voltage number of battery cell, minimum value of battery cell voltage, minimum voltage number of battery cell, voltage value of each cell, temperature value of each temperature sensor, current, total voltage, accumulated mileage, SOC, and charge and discharge status.

4. A safety warning method for a power battery according to claim 2, characterized in that: The specific operation of step S5 is: calculating the average value μ and standard deviation σ of the entropy value slopes of all cells, normalizing the entropy value slope When Z i When <-6σ, cell i is an abnormal cell R.

5. A safety warning method for a power battery according to claim 4, characterized in that: The characteristic parameters of step S6 include: the minimum value Z of the normalized entropy slope of the battery cell R Rmin , the minimum value Z of the normalized entropy slope of all cells except cell R imin , the rate value V at which the maximum value of the cell R voltage ranking drops to the minimum value R , the difference between the maximum voltage value and the voltage of the battery cell R and the rate of change with mileage C R , the rate value Vk of the normalized entropy slope value of the battery R from the maximum value to the minimum value R , the duration of the minimum value of the normalized entropy slope of the battery cell R, and the minimum value number of the single cell voltage.

6. A safety warning method for a power battery according to claim 4, characterized in that: The step S7 of using the random forest algorithm to train the model specifically includes: S701: further divide the data set into a training set and a test set, the training set is used for model training, and the test set is used to evaluate the performance of the model; S702: Use the random forest algorithm to train the training set data. During the training process, it is necessary to adjust the parameters of the model: the number of trees and the maximum depth of the trees to optimize the model performance; S703: Use the test set data to evaluate the performance of the trained model.

7. A safety warning method for a power battery according to claim 2, characterized in that: The preset threshold MT of step S8 is 0.

8.

8. A safety warning method for a power battery according to claim 6, characterized in that: The number of trees in the random forest algorithm is set to 100 and the maximum depth of the tree is set to 15.

9. A power battery safety warning method according to claim 6, characterized in that: The entropy slope of step S8 is preset to -6σ.

10. A safety warning device for a power battery, characterized in that: It comprises a memory and a processor, wherein the memory stores a safety warning program for a power battery, and the processor executes the power battery safety warning method according to any one of claims 1 to 9 according to the power battery safety warning program in the memory.

Images