Power battery thermal runaway prediction method, device, electronic equipment and vehicle

Abstract

This application relates to a method, device, electronic device, and vehicle for predicting thermal runaway of a power battery. The method relates to the field of vehicle control technology and addresses at least the low accuracy of thermal runaway judgment in related technologies. The method comprises: determining time domain features based on battery status data, operating data, and environmental data of the vehicle's power battery's current charge and discharge cycle, wherein the battery status data includes voltage and temperature time series data of battery cells in the power battery; determining the highest cell voltage and temperature time series data of the power battery based on the voltage and temperature time series data; determining frequency domain features based on the highest cell voltage and temperature time series data; and predicting thermal runaway of the vehicle based on the time and frequency domain features. This application is intended to improve the accuracy of thermal runaway judgment.

Description

Technical Field

[0001] The present application relates to the field of vehicle control technology, and in particular to a method, device, electronic equipment and vehicle for predicting thermal runaway of a power battery. Background Art

[0002] With the rapid development of new energy vehicle technology, the safety of power batteries, as core components of new energy vehicles, has become a key issue in the industry. During use, power batteries may cause thermal runaway due to factors such as aging, manufacturing defects, and abuse, thus causing safety issues.

[0003] Related technologies analyze charging data from vehicles and construct fuzzy weighted neural network models to provide early warnings of thermal runaway in power batteries based on capacity degradation and voltage anomalies. This approach only considers charging data from the vehicle, but thermal runaway is influenced by a variety of data, not just charging data. Therefore, this technology has low accuracy in determining thermal runaway.

[0004] Another related technique determines the capacity difference by calculating the maximum and minimum voltages of fully charged cells, and then calculates the leakage current based on the capacity difference. Furthermore, by calculating the average voltage of the lowest-voltage cells in the battery pack during the interval between two consecutive full charges, the short-circuit resistance of the lowest-voltage cells in the battery pack during these intervals is calculated. This establishes a multi-parameter coupled early warning mechanism to provide early warning of thermal runaway in the power battery. This technical solution only considers relevant parameters of the power battery, but thermal runaway is affected by a variety of data, not just the power battery parameters. Therefore, this technique has low accuracy in determining thermal runaway. Summary of the Invention

[0005] The present application provides a method, device, electronic device and vehicle for predicting thermal runaway of a power battery. The purpose of the present application is to at least solve the technical problem of low accuracy in judging thermal runaway in the related art.

[0006] In order to achieve the above objectives, the technical solutions adopted in this application are as follows:

[0007] According to the first aspect provided by the present application, a method for predicting thermal runaway of a power battery is provided, and the method for predicting thermal runaway of a power battery includes: determining time domain characteristics based on battery status data, operating data, and environmental data of a current charge and discharge journey of a vehicle's power battery, wherein the battery status data includes voltage timing data and temperature timing data of battery cells in the power battery; determining the highest cell voltage timing data and the highest cell temperature timing data of the power battery based on the voltage timing data and the temperature timing data; determining frequency domain characteristics based on the highest cell voltage timing data and the highest cell temperature timing data; and predicting thermal runaway of the vehicle based on the time domain characteristics and the frequency domain characteristics.

[0008] According to the above technical means, when the vehicle is running, the present application can determine whether the vehicle's current running trip is a charging trip or a discharging trip, thereby determining the time domain characteristics of the vehicle when it is running based on the battery status data, operating data and environmental data of the current charging and discharging trip. And, further based on the voltage time series data and temperature time series data included in the battery status data, the highest cell voltage time series data and the highest cell temperature time series data of the power battery are determined. Then, based on the highest cell voltage time series data and the highest cell temperature time series data, the frequency domain characteristics of the vehicle when it is running are determined. In this way, by analyzing the data when the vehicle is running, the corresponding time domain characteristics and frequency domain characteristics are determined, and then based on the time domain characteristics and frequency domain characteristics, the operation of the vehicle is analyzed from multiple dimensions, and the thermal runaway of the vehicle is predicted. In this way, by analyzing the multi-dimensional data, the thermal runaway of the vehicle can be accurately predicted based on the comprehensive data when the vehicle is running, thereby improving the accuracy of predicting the thermal runaway of the vehicle.

[0009] In a possible embodiment, the battery status data also includes: current time series data, voltage time series data and resistance time series data of the power battery; the above-mentioned battery status data, operating data and environmental data based on the current charge and discharge journey of the vehicle power battery determine the time domain characteristics, including: determining statistical characteristics based on the current time series data, voltage time series data and resistance time series data of the power battery, and the voltage time series data and temperature time series data of the battery cell; determining the voltage correlation characteristics between battery cells based on the voltage time series data of the battery cell; determining the voltage travel change characteristics and temperature travel change characteristics of the battery cell based on the voltage time series data and temperature time series data of the battery cell; determining the time domain characteristics based on the operating data, environmental data, statistical characteristics, voltage correlation characteristics, voltage travel change characteristics and temperature travel change characteristics.

[0010] According to the above technical means, when determining the time domain characteristics, the present application can perform data analysis based on the current time series data, voltage time series data and resistance time series data of the power battery included in the battery status data, as well as the voltage time series data and temperature time series data of the battery cell to determine the statistical characteristics. And based on the voltage time series data of the battery cell, determine the voltage correlation characteristics between the battery cells. And, based on the voltage time series data and temperature time series data of the battery cell, determine the voltage travel change characteristics and temperature travel change characteristics of the battery cell. Therefore, based on the operating data, environmental data, statistical characteristics, voltage correlation characteristics, voltage travel change characteristics and temperature travel change characteristics, the time domain characteristics can be determined from a multi-dimensional perspective. In this way, by analyzing multi-dimensional data, the time domain characteristics can be accurately determined based on the comprehensive data during vehicle operation.

[0011] In a possible embodiment, the above-mentioned determination of statistical characteristics based on the current time series data, voltage time series data and resistance time series data of the power battery, as well as the voltage time series data and temperature time series data of the battery cell, includes: determining the cell voltage key value and the cell temperature key value of the power battery based on the voltage time series data and the temperature time series data of the battery cell, wherein the cell voltage key value includes the maximum value, minimum value, median value and maximum voltage difference of the cell voltage; the cell temperature key value includes the maximum value, minimum value, median value and maximum temperature difference of the cell temperature; determining the statistical characteristics of the current time series data, voltage time series data and resistance time series data of the power battery, as well as the cell voltage key value and the cell temperature key value of the power battery, wherein the statistical characteristics include at least one of the following: the maximum value, mean value, effective value, peak-to-peak value, difference ratio, skewness and peak factor in the current charge and discharge trip.

[0012] According to the above technical means, the present application determines the key values ​​of cell voltage and cell temperature of the power battery through the voltage time series data and temperature time series data of the battery cell, that is, the maximum value, minimum value, median value and maximum voltage difference of the cell voltage, as well as the maximum value, minimum value, median value and maximum temperature difference of the cell temperature can be determined. Furthermore, the statistical characteristics of the current time series data, voltage time series data and resistance time series data of the power battery, as well as the key values ​​of cell voltage and cell temperature of the power battery are determined, so as to determine the maximum value, mean value, effective value, peak-to-peak value, difference ratio, skewness and peak factor in the current charge and discharge cycle. In this way, the statistical characteristics in the time domain characteristics can be accurately determined.

[0013] In one possible implementation, the above-mentioned determination of the voltage correlation characteristics between battery cells based on the voltage timing data of the battery cells includes: determining the voltage mean of the battery cells in the current charge and discharge cycle based on the voltage timing data of the battery cells; and determining the voltage correlation characteristics between battery cells using a Pearson correlation coefficient calculation method based on the voltage timing data and the voltage mean of the battery cells.

[0014] Based on the above technical means, the present application determines the average voltage of the battery cells during the current charge and discharge cycle using the voltage time series data of the battery cells. Then, based on the voltage time series data and the voltage average of the battery cells, the Pearson correlation coefficient calculation method is used to accurately determine the voltage correlation characteristics between the battery cells.

[0015] In one possible implementation, the above-mentioned determination of the voltage range change characteristics and temperature range change characteristics of the battery cell based on the voltage time series data and temperature time series data of the battery cell includes: calculating the change characteristics of the cell voltage key value and the cell temperature key value of the power battery, wherein the change characteristics include at least one of the following: increase ratio, maximum change amount and change rate.

[0016] Based on the above technical means, this application can calculate the variation characteristics of the key cell voltage and temperature values ​​of the power battery based on the cell voltage and temperature time series data, thereby determining the voltage or temperature increase ratio, maximum change, and change rate. By determining these characteristics, the accuracy of thermal runaway prediction for the vehicle can be improved.

[0017] In one possible embodiment, the environmental data includes the outside temperature; the operating data includes at least one of the following: the ratio of the duration that the driving speed exceeds the preset driving speed to the duration that the driving speed does not exceed the preset driving speed, the duration of driving when the temperature exceeds the preset temperature threshold, the ratio of the duration that the acceleration exceeds the preset acceleration threshold to the total duration of the current charge and discharge stroke; the ratio of the duration that the deceleration exceeds the preset deceleration threshold to the total duration of the current charge and discharge stroke; the above-mentioned determination of the time domain characteristics based on the operating data, environmental data, statistical characteristics, voltage correlation characteristics, voltage stroke change characteristics and temperature stroke change characteristics includes: determining the outside temperature, various data included in the operating data, empirical characteristics, statistical characteristics, voltage correlation characteristics, voltage stroke change characteristics and temperature stroke change characteristics as time domain characteristics; wherein, the empirical characteristics include at least one of the following: charge and discharge depth, charging overcurrent ratio, the ratio of the duration when the state of charge is lower than the preset state of charge, the ratio of energy recovery time and the ratio of energy recovery current steps.

[0018] According to the above technical means, when determining the time domain characteristics, the present application can determine the external temperature, various data included in the operating data, empirical characteristics, statistical characteristics, voltage correlation characteristics, voltage range change characteristics and temperature range change characteristics as time domain characteristics. In this way, by determining the depth of charge and discharge, the proportion of charging overcurrent, the proportion of time when the state of charge is lower than the preset state of charge, the proportion of energy recovery time and the proportion of energy recovery current steps, as well as the external temperature, the proportion of time when the driving speed exceeds the preset driving speed and the proportion of time when the driving speed does not exceed the preset driving speed, the duration of driving when the temperature exceeds the preset temperature threshold, the proportion of time when the acceleration exceeds the preset acceleration threshold to the total duration of the current charge and discharge trip, and the proportion of time when the deceleration exceeds the preset deceleration threshold to the total duration of the current charge and discharge trip, these data are determined as time domain characteristics. When predicting thermal runaway of the vehicle, the accuracy of the prediction can be improved.

[0019] In a possible embodiment, the above-mentioned determination of frequency domain characteristics based on the highest cell voltage timing data and the highest cell temperature timing data includes: for the target timing data in the highest cell voltage timing data and the highest cell temperature timing data, performing frequency domain transformation on the target timing data to obtain frequency domain data; determining the frequency amplitude and power spectrum based on the frequency domain data; determining the frequency domain characteristics based on the frequency amplitude and the power spectrum, wherein the frequency domain characteristics include at least one of the following: DC component, center of gravity frequency, frequency standard deviation and root mean square frequency.

[0020] Based on the above technical means, the present application can perform a frequency domain transformation on the highest cell voltage time series data or the highest cell temperature time series data, thereby obtaining the frequency domain data corresponding to the voltage or temperature. Then, based on the frequency domain data corresponding to the voltage or temperature, the frequency amplitude and power spectrum corresponding to the voltage or temperature are determined, and then the corresponding frequency domain characteristics are determined based on the frequency amplitude and power spectrum. In this way, by further performing a frequency domain transformation on the data to obtain frequency domain characteristics, the accuracy of the prediction can be further improved when predicting thermal runaway of the vehicle.

[0021] In a possible embodiment, the above-mentioned determination of frequency domain characteristics based on the highest monomer voltage timing data and the highest monomer temperature timing data includes: performing wavelet transform on the target timing data in the highest monomer voltage timing data and the highest monomer temperature timing data to obtain high-frequency coefficients and low-frequency coefficients; determining frequency domain characteristics based on the high-frequency coefficients and the low-frequency coefficients; wherein the frequency domain characteristics are energy proportions.

[0022] Based on the above technical means, the present application can also perform a wavelet transform on the highest cell voltage time series data or the highest cell temperature time series data to obtain high-frequency coefficients and low-frequency coefficients. Based on the high-frequency coefficients and low-frequency coefficients, the energy proportion is then determined as a frequency domain feature. In this way, frequency domain features can also be determined through wavelet transform, further improving the accuracy of vehicle thermal runaway prediction.

[0023] In a possible implementation, the above-mentioned thermal runaway prediction of the vehicle based on time domain features and frequency domain features includes: feature screening of time domain features and frequency domain features to obtain key features; and using the key features to predict thermal runaway of the vehicle.

[0024] Based on the above technical means, this application can also perform feature screening on time-domain and frequency-domain features to obtain key features, which can then be used to predict vehicle thermal runaway. In this way, the key features obtained after screening can further improve the accuracy of predicting vehicle thermal runaway.

[0025] In a possible implementation, the above-mentioned use of key features to predict thermal runaway of a vehicle includes: inputting the key features into a thermal runaway prediction model to obtain a thermal runaway risk level of the vehicle.

[0026] Based on the above technical means, this application can use a pre-trained thermal runaway prediction model to process the key features screened, thereby predicting the vehicle's thermal runaway risk level. In this way, the thermal runaway prediction model can efficiently and accurately determine the vehicle's thermal runaway risk level, improving early warning efficiency.

[0027] In one possible implementation, the training process of the thermal runaway prediction model includes: determining multiple historical charging and discharging trips of the sample vehicle based on historical data of the sample vehicle's full life cycle; determining the thermal runaway risk levels of the multiple historical charging and discharging trips; and training the thermal runaway prediction model based on the key features and thermal runaway risk levels of the multiple historical charging and discharging trips.

[0028] Based on the above technical means, when pre-training a thermal runaway prediction model, the present application can analyze historical data from the full lifecycle of a sample vehicle, thereby dividing the sample vehicle's full lifecycle into multiple historical charge and discharge cycles. Thus, by determining the thermal runaway risk level for each of these historical charge and discharge cycles and combining the key characteristics of these multiple historical charge and discharge cycles, a thermal runaway prediction model can be trained. By pre-training the thermal runaway prediction model, the efficiency and accuracy of predicting a vehicle's thermal runaway risk level can be improved.

[0029] In one possible implementation, the above-mentioned determination of multiple historical charging and discharging trips of the sample vehicle based on historical data of the full life cycle of the sample vehicle includes: determining multiple historical trips of the sample vehicle based on the historical data, wherein there is no historical data for the time period between two adjacent historical trips, and the time interval between the time periods is greater than a preset duration; and determining multiple historical charging and discharging trips based on the multiple historical trips.

[0030] Based on the above technical means, this application can analyze the historical data of the entire life cycle of a sample vehicle, thereby dividing the historical data into multiple historical trips based on the time difference between the data, and thus determining multiple historical charge and discharge trips. In this way, based on the generation time points of the vehicle charge and discharge data in the historical data, multiple historical charge and discharge trips can be determined. Then, by analyzing the data of multiple historical charge and discharge trips, corresponding features can be obtained.

[0031] In one possible implementation, the historical charge and discharge trips include a charging trip and a discharging trip. Based on the multiple historical trips, multiple historical charge and discharge trips are determined, including: for each historical trip, if the historical trip satisfies the condition that the sample vehicle is in a charging state and the mileage is less than a preset mileage, determining the historical trip as a charging trip; or, if the historical trip satisfies the condition that the sample vehicle is in a discharging state and the driving time is greater than a preset driving time, determining the historical trip as a discharging trip.

[0032] Based on the above technical means, the present application can determine whether a trip is a charging trip or a discharging trip based on the vehicle's charging status, mileage, discharging status, and driving duration in the data corresponding to each historical trip. Therefore, in subsequent data analysis, the corresponding characteristics of the charging trip and the discharging trip can be determined based on their characteristics.

[0033] In one possible implementation, the above-mentioned determination of the thermal runaway risk levels of multiple historical charge and discharge trips includes: determining the time interval between the multiple historical charge and discharge trips and the time when thermal runaway of the sample vehicle occurred; based on the time interval, determining the thermal runaway risk levels of the multiple historical charge and discharge trips; wherein, the closer the time interval, the higher the thermal runaway risk level of the corresponding historical charge and discharge trip.

[0034] According to the above technical means, when determining the thermal runaway risk level of multiple historical charging and discharging trips, the present application can determine the corresponding thermal runaway risk level based on the time interval between each historical charging and discharging trip and the moment when thermal runaway occurred in the sample vehicle, so that the closer the time interval, the higher the thermal runaway risk level of the historical charging and discharging trip.

[0035] According to a second aspect of the present application, a power battery thermal runaway prediction device is provided, the power battery thermal runaway prediction device comprising: a processing module and a prediction module;

[0036] The processing module is used to determine the time domain characteristics based on the battery status data, operating data and environmental data of the current charge and discharge journey of the vehicle's power battery, wherein the battery status data includes the voltage time series data and temperature time series data of the battery cells in the power battery; the processing module is also used to determine the highest cell voltage time series data and the highest cell temperature time series data of the power battery based on the voltage time series data and the temperature time series data; the processing module is also used to determine the frequency domain characteristics based on the highest cell voltage time series data and the highest cell temperature time series data; the prediction module is used to predict thermal runaway of the vehicle based on the time domain characteristics and the frequency domain characteristics.

[0037] In one possible embodiment, the battery status data also includes: current timing data, voltage timing data and resistance timing data of the power battery; a processing module, specifically used to determine statistical characteristics based on the current timing data, voltage timing data and resistance timing data of the power battery, and the voltage timing data and temperature timing data of the battery cell; a processing module, specifically used to determine the voltage correlation characteristics between battery cells based on the voltage timing data of the battery cell; a processing module, specifically used to determine the voltage travel change characteristics and temperature travel change characteristics of the battery cell based on the voltage timing data and temperature timing data of the battery cell; a processing module, specifically used to determine the time domain characteristics based on operating data, environmental data, statistical characteristics, voltage correlation characteristics, voltage travel change characteristics and temperature travel change characteristics.

[0038] In one possible embodiment, the processing module is specifically used to determine the cell voltage key value and the cell temperature key value of the power battery based on the voltage timing data and temperature timing data of the battery cell, wherein the cell voltage key value includes the maximum value, minimum value, median value and maximum voltage difference of the cell voltage; the cell temperature key value includes the maximum value, minimum value, median value and maximum temperature difference of the cell temperature; the processing module is specifically used to determine the current timing data, voltage timing data and resistance timing data of the power battery, as well as the statistical characteristics of the cell voltage key value and the cell temperature key value of the power battery, wherein the statistical characteristics include at least one of the following: the maximum value, mean value, effective value, peak-to-peak value, difference ratio, skewness and peak factor in the current charge and discharge cycle.

[0039] In one possible implementation, the processing module is specifically used to determine the voltage mean of the battery cells in the current charge and discharge cycle based on the voltage timing data of the battery cells; the processing module is specifically used to determine the voltage correlation characteristics between the battery cells by using the Pearson correlation coefficient calculation method based on the voltage timing data and the voltage mean of the battery cells.

[0040] In a possible implementation, the processing module is specifically configured to calculate change characteristics of a cell voltage key value and a cell temperature key value of the power battery, wherein the change characteristics include at least one of the following: an increase ratio, a maximum change amount, and a change rate.

[0041] In one possible embodiment, the environmental data includes the outside temperature; the operating data includes at least one of the following: the ratio of the duration that the driving speed exceeds the preset driving speed to the duration that the driving speed does not exceed the preset driving speed, the duration of driving when the temperature exceeds the preset temperature threshold, the ratio of the duration that the acceleration exceeds the preset acceleration threshold to the total duration of the current charging and discharging stroke, and the ratio of the duration that the deceleration exceeds the preset deceleration threshold to the total duration of the current charging and discharging stroke; the processing module is specifically used to determine the outside temperature, various data included in the operating data, empirical characteristics, statistical characteristics, voltage correlation characteristics, voltage stroke change characteristics and temperature stroke change characteristics as time domain characteristics; wherein the empirical characteristics include at least one of the following: charge and discharge depth, charging overcurrent ratio, the ratio of the duration when the state of charge is lower than the preset state of charge, the ratio of energy recovery time and the ratio of energy recovery current steps.

[0042] In one possible embodiment, the processing module is specifically used to perform frequency domain transformation on the target timing data in the highest cell voltage timing data and the highest cell temperature timing data to obtain frequency domain data; the processing module is specifically used to determine the frequency amplitude and power spectrum based on the frequency domain data; the processing module is specifically used to determine the frequency domain characteristics based on the frequency amplitude and power spectrum, wherein the frequency domain characteristics include at least one of the following: DC component, center of gravity frequency, frequency standard deviation and root mean square frequency.

[0043] In one possible embodiment, the processing module is specifically used to perform wavelet transform on the target timing data in the highest cell voltage timing data and the highest cell temperature timing data to obtain high-frequency coefficients and low-frequency coefficients; the processing module is specifically used to determine the frequency domain characteristics based on the high-frequency coefficients and the low-frequency coefficients; wherein the frequency domain characteristics are energy proportions.

[0044] In one possible implementation, the processing module is specifically used to perform feature screening on time domain features and frequency domain features to obtain key features; the prediction module is specifically used to use the key features to predict thermal runaway of the vehicle.

[0045] In a possible implementation, the processing module, specifically the prediction module, is specifically configured to input the key features into a thermal runaway prediction model to obtain a thermal runaway risk level of the vehicle.

[0046] In one possible implementation, the processing module is specifically used to determine multiple historical charging and discharging trips of the sample vehicle based on historical data of the sample vehicle's full life cycle; the processing module is specifically used to determine the thermal runaway risk levels of the multiple historical charging and discharging trips; and the processing module is specifically used to train a thermal runaway prediction model based on the key features and thermal runaway risk levels of the multiple historical charging and discharging trips.

[0047] In one possible implementation, the processing module is specifically configured to determine multiple historical trips of a sample vehicle based on historical data, wherein no historical data exists for a time period between two adjacent historical trips and the time interval between the time periods is greater than a preset duration; and the processing module is specifically configured to determine multiple historical charging and discharging trips based on the multiple historical trips.

[0048] In one possible implementation, the processing module is specifically configured to, for each historical trip, determine that the historical trip is a charging trip if the historical trip satisfies the requirement that the sample vehicle is in a charging state and the mileage is less than a preset mileage; or, the processing module is specifically configured to determine that the historical trip is a discharging trip if the historical trip satisfies the requirement that the sample vehicle is in a discharging state and the driving duration is greater than a preset driving duration.

[0049] In one possible implementation, the processing module is specifically configured to determine the time intervals between multiple historical charge and discharge trips and the moment when thermal runaway of a sample vehicle occurs; and based on the time intervals, determine the thermal runaway risk levels of the multiple historical charge and discharge trips; wherein, the closer the time interval, the higher the thermal runaway risk level of the corresponding historical charge and discharge trip.

[0050] According to the third aspect provided by the present application, an electronic device is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute instructions to implement the method of the above-mentioned first aspect and any possible implementation method thereof.

[0051] According to the fourth aspect provided by the present application, a computer-readable storage medium is provided. When the computer execution instructions stored in the computer-readable storage medium are executed by the processor of an electronic device, the electronic device executes the method of the above-mentioned first aspect and any possible implementation method thereof.

[0052] According to the fifth aspect provided by the present application, a computer program product is provided, which includes computer instructions. When the computer instructions are executed on an electronic device, the electronic device executes the method of the above-mentioned first aspect and any possible implementation method thereof.

[0053] According to the sixth aspect provided by the present application, a vehicle is provided, the vehicle including the power battery thermal runaway prediction device as described in the second aspect, and the vehicle is used to implement the method of the above-mentioned first aspect and any possible implementation method thereof.

[0054] It should be noted that the technical effects brought about by any implementation method in the second to sixth aspects can refer to the technical effects brought about by the corresponding implementation method in the first aspect, and will not be repeated here.

[0055] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification are used to explain the principles of the present application, and do not constitute an improper limitation on the present application.

[0057] Figure 1 is a schematic structural diagram of a power battery thermal runaway prediction system according to an exemplary embodiment;

[0058] Figure 2 is a flow chart showing a method for predicting thermal runaway of a power battery according to an exemplary embodiment;

[0059] Figure 3 is a flow chart showing a method for determining time domain features according to an exemplary embodiment;

[0060] Figure 4 is a flow chart showing a method of determining statistical features according to an exemplary embodiment;

[0061] Figure 5 is a flow chart showing a method for determining voltage correlation characteristics according to an exemplary embodiment;

[0062] Figure 6 is a flow chart showing a method for determining frequency domain features according to an exemplary embodiment;

[0063] Figure 7 is a flowchart illustrating another method of determining frequency domain features according to an exemplary embodiment;

[0064] Figure 8 is a flowchart showing a model construction method according to an exemplary embodiment;

[0065] Figure 9 is a flow chart showing a risk level prediction method according to an exemplary embodiment;

[0066] Figure 10is a block diagram of a power battery thermal runaway prediction device according to an exemplary embodiment;

[0067] Figure 11 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0068] In order to enable ordinary people in the art to better understand the technical solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0069] It should be noted that the terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish similar objects and are not used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0070] The power battery thermal runaway prediction method provided in the embodiment of the present application can be applied to the power battery thermal runaway prediction system. Figure 1 As shown, the power battery thermal runaway prediction system includes: a power battery 11 , a controller 12 and a sensor 13 .

[0071] Among them, the power battery 11 is used to provide electrical energy for the operation of the vehicle, the controller 12 can be a vehicle controller or a power battery management system, and the sensor 13 is used to collect relevant data of the vehicle during charging (such as the power battery's state of charge, battery temperature, etc.), or collect relevant data of the vehicle during operation (such as the power battery's temperature, voltage, current, vehicle speed, etc.).

[0072] Specifically, the controller 12 is used to determine the time domain characteristics based on the battery status data, operating data and environmental data of the current charge and discharge journey of the vehicle power battery 11, wherein the battery status data includes the voltage time series data and temperature time series data of the battery cells in the power battery 11.

[0073] Specifically, the controller 12 is configured to determine the highest cell voltage time series data and the highest cell temperature time series data of the power battery 11 based on the voltage time series data and the temperature time series data.

[0074] Specifically, the controller 12 is used to determine the frequency domain characteristics based on the highest cell voltage time series data and the highest cell temperature time series data; and then, predict thermal runaway of the vehicle based on the time domain characteristics and the frequency domain characteristics.

[0075] The sensor 13 can collect battery status data, operating data and environmental data of the current charging and discharging process.

[0076] For ease of understanding, the power battery thermal runaway prediction method provided by this application is specifically introduced below with reference to the accompanying drawings. The power battery thermal runaway prediction method is applied to the dual power battery thermal runaway predictor in the power battery thermal runaway prediction system, such as Figure 2 As shown, the power battery thermal runaway prediction method includes S201-S204:

[0077] S201 : Determine time domain characteristics based on battery status data, operating data, and environmental data of a current charge and discharge journey of a vehicle power battery.

[0078] The battery status data includes the voltage time series data and temperature time series data of the battery cells in the power battery.

[0079] In an embodiment of the present application, real-time operating data can be acquired while the vehicle is operating (charging or driving). Based on this acquired real-time operating data, battery status data, operating data, and environmental data can be determined. Data analysis is then performed based on the battery status data, operating data, and environmental data to determine time-domain and frequency-domain characteristics, which can then be used to predict thermal runaway in the vehicle.

[0080] In a possible implementation, when the vehicle is charging, the current charging and discharging stroke is the charging stroke, and when the vehicle is traveling, the current charging and discharging stroke is the discharging stroke.

[0081] In one possible implementation, the real-time operating data of the vehicle during operation may include: voltage list, temperature list, total current, insulation resistance, terminal time, vehicle charging status, vehicle speed, total mileage, state of charge (SOC), longitude and latitude, outside temperature and other signals.

[0082] In one possible implementation, the real-time running data of the vehicle can be sorted in ascending order by terminal time and total mileage, and pre-processed to remove abnormal data and handle missing values.

[0083] Specifically, rows where the total current exceeds a preset value (e.g., 10,000A, 9,000A, etc.) or where the SOC exceeds a threshold (e.g., 100) can be deleted. Missing values ​​are also handled. For example, if the voltage and temperature lists exist and are reasonable, and other data is missing for a continuous period exceeding a preset duration (e.g., 1 or 2 minutes), the rows containing the missing values ​​are deleted (i.e., all data corresponding to the missing data period is deleted). Otherwise, if other data is missing for a continuous period not exceeding the preset duration, the missing values ​​are filled in using an upward fill-up method.

[0084] In one possible implementation, linear interpolation can be used to calculate the battery's state of health (SOH) parameters based on the battery model and total mileage. The voltage and temperature lists are then separated, and the maximum voltage (temperature), minimum voltage (temperature), median voltage (temperature), and pressure (temperature) differential are calculated over time.

[0085] In addition, the total voltage and total current data are combined to further calculate the power and internal resistance parameters. These multi-source data are then integrated and analyzed to provide rich and effective information support for early warning of sudden thermal runaway.

[0086] S202 : Determine the highest cell voltage time series data and the highest cell temperature time series data of the power battery based on the voltage time series data and the temperature time series data.

[0087] In one possible implementation, the voltage time series data and the temperature time series data may be analyzed and processed to determine the highest cell voltage time series data and the highest cell temperature time series data of the multiple battery cells included in the power battery from the voltage time series data and the temperature time series data.

[0088] It can be understood that the maximum cell voltage time series data is the maximum value among the voltages of multiple battery cells included in the power battery at each moment; the maximum cell temperature time series data is the maximum value among the temperatures of multiple battery cells included in the power battery at each moment.

[0089] S203 : Determine frequency domain characteristics based on the highest cell voltage time series data and the highest cell temperature time series data.

[0090] Based on this, after determining the highest cell voltage timing data and the highest cell temperature timing data, frequency domain analysis can be performed on the highest cell voltage timing data and the highest cell temperature timing data to determine corresponding frequency domain features.

[0091] S204: Predict thermal runaway of the vehicle based on time domain characteristics and frequency domain characteristics.

[0092] In this way, after determining the time domain characteristics and frequency domain characteristics of the current charging and discharging journey, the time domain characteristics and frequency domain characteristics can be analyzed and processed to predict the thermal runaway of the vehicle and determine the thermal runaway risk level of the vehicle.

[0093] In an embodiment of the present application, when the vehicle is running, the present application can determine whether the vehicle's current running trip is a charging trip or a discharging trip, thereby determining the time domain characteristics of the vehicle when it is running based on the battery status data, operating data and environmental data of the current charging and discharging trip. And, further based on the voltage time series data and temperature time series data included in the battery status data, the highest cell voltage time series data and the highest cell temperature time series data of the power battery are determined. Then, based on the highest cell voltage time series data and the highest cell temperature time series data, the frequency domain characteristics of the vehicle when it is running are determined. In this way, by analyzing the data when the vehicle is running, the corresponding time domain characteristics and frequency domain characteristics are determined, and then based on the time domain characteristics and frequency domain characteristics, the operation of the vehicle is analyzed from multiple dimensions, and the thermal runaway of the vehicle is predicted. In this way, by analyzing the multi-dimensional data, the thermal runaway of the vehicle can be accurately predicted based on the comprehensive data when the vehicle is running, thereby improving the accuracy of predicting the thermal runaway of the vehicle.

[0094] In some embodiments, the battery status data further includes: current time series data, voltage time series data, and resistance time series data of the power battery.

[0095] In some embodiments, as Figure 3 As shown, the above S201 may specifically include S301-S304:

[0096] S301 : Determine statistical features based on current time series data, voltage time series data, and resistance time series data of the power battery, and voltage time series data and temperature time series data of the battery cell.

[0097] In one possible implementation, statistical characteristics of total voltage, total current, maximum cell voltage, minimum cell voltage, median cell voltage, maximum cell temperature, minimum cell temperature, median cell temperature, voltage difference, temperature difference, and insulation resistance may be calculated respectively.

[0098] In some embodiments, as Figure 4 As shown, the above S301 may specifically include S401-S402:

[0099] S401 : Determine a cell voltage key value and a cell temperature key value of a power battery based on voltage time series data and temperature time series data of a battery cell.

[0100] Among them, the cell voltage key values ​​include the maximum value, minimum value, median value and maximum voltage difference of the cell voltage; the cell temperature key values ​​include the maximum value, minimum value, median value and maximum temperature difference of the cell temperature.

[0101] S402 : Determine statistical characteristics of current time series data, voltage time series data, and resistance time series data of the power battery, as well as key values ​​of cell voltage and cell temperature of the power battery.

[0102] The statistical features include at least one of the following: a maximum value, a mean value, an effective value (ie, a root mean square (RMS)), a peak-to-peak value, a difference ratio, a skewness, and a crest factor in the current charge and discharge process.

[0103] In a possible implementation, the effective value RMS can be calculated based on formula one.

[0104]

[0105] in, Indicates current time series data, voltage time series data and resistance time series data, as well as voltage time series data and temperature time series data of battery cells, any one of the maximum value, minimum value, median value and maximum voltage difference of cell voltage, and the maximum value, minimum value, median value and maximum temperature difference of cell temperature.

[0106] In a possible implementation, the peak-to-peak value can be calculated based on Formula 2.

[0107]

[0108] in, express The maximum value of express The minimum value of .

[0109] In a possible implementation, the difference ratio can be calculated based on Formula 3.

[0110]

[0111] In a possible implementation, the skewness can be calculated based on Formula 4.

[0112]

[0113] in, represents the mean of X.

[0114] In a possible implementation, the peak factor can be calculated based on Formula 5.

[0115]

[0116] In an embodiment of the present application, the present application determines the cell voltage key value and cell temperature key value of the power battery through the voltage time series data and temperature time series data of the battery cell, that is, the maximum value, minimum value, median value and maximum voltage difference of the cell voltage, as well as the maximum value, minimum value, median value and maximum temperature difference of the cell temperature can be determined. Further, the statistical characteristics of the current time series data, voltage time series data and resistance time series data of the power battery, as well as the cell voltage key value and cell temperature key value of the power battery are determined, so as to determine the maximum value, mean value, effective value, peak-to-peak value, difference ratio, skewness and peak factor in the current charge and discharge stroke. In this way, the statistical characteristics in the time domain characteristics can be accurately determined.

[0117] S302 : Determine voltage correlation characteristics between battery cells based on voltage time series data of the battery cells.

[0118] In some embodiments, as Figure 5 As shown, the above S302 may specifically include S501-S502:

[0119] S501 : Determine the average voltage of the battery cells in the current charge and discharge process based on the voltage time series data of the battery cells.

[0120] S502 : Based on the voltage time series data and voltage mean values ​​of the battery cells, a Pearson correlation coefficient calculation method is used to determine voltage correlation characteristics between the battery cells.

[0121] In a possible implementation, it is necessary to determine the voltage correlation characteristics between any two battery cells. The voltage correlation characteristics between the battery cells can be obtained by calculating the correlation coefficient using a Pearson correlation coefficient calculation method.

[0122] In a possible implementation, the Pearson correlation coefficient calculation method can be expressed by Formula 6.

[0123]

[0124] in, represents the voltage of the first battery cell of any two battery cells at time i, represents the voltage of the second battery cell of any two battery cells at time i, represents the average voltage of the first battery cell in any two battery cells, It represents the average voltage of the second battery cell among any two battery cells.

[0125] In one possible implementation, after determining the Pearson correlation coefficient between any two battery cells among the multiple battery cells included in the power battery, a consistency matrix can be obtained based on these coefficients. Then, the mean of the coefficients in each row of the consistency matrix is ​​determined to obtain a list with a length equal to the number of battery cells, which is recorded as a battery cell voltage correlation list and is used to describe the consistency between the battery cells.

[0126] In the embodiment of the present application, the present application determines the average voltage of the battery cells during the current charge and discharge cycle using the voltage time series data of the battery cells. Then, based on the voltage time series data and the voltage average of the battery cells, the Pearson correlation coefficient calculation method is used to accurately determine the voltage correlation characteristics between the battery cells.

[0127] S303 : Determine voltage travel variation characteristics and temperature travel variation characteristics of the battery cell based on the voltage time series data and temperature time series data of the battery cell.

[0128] In some embodiments, the above S303 may specifically include: calculating the change characteristics of the key values ​​of the cell voltage and the cell temperature of the power battery, wherein the change characteristics include at least one of the following: increase ratio, maximum change amount and change rate.

[0129] It should be noted that the increase ratio represents the ratio of the difference between the cell voltage (or cell temperature) at the next moment and the cell voltage (or cell temperature) at the previous moment to the cell voltage (or cell temperature) at the previous moment.

[0130] In an embodiment of the present application, the present application can calculate the variation characteristics of the key cell voltage and temperature values ​​of the power battery based on the cell voltage and temperature time series data, thereby determining the voltage or temperature increase ratio, maximum change, and change rate. By determining these characteristics, the accuracy of thermal runaway prediction for the vehicle can be improved.

[0131] In a possible implementation, by analyzing the variation characteristics of the cell voltage key value and the cell temperature key value, the voltage range variation characteristics and the temperature range variation characteristics of the battery cell can be determined.

[0132] S304 : Determine time domain features based on the operating data, environmental data, statistical features, voltage correlation features, voltage range change features, and temperature range change features.

[0133] In some embodiments, the environmental data includes the outside temperature; the operating data includes at least one of the following: the ratio of the duration that the driving speed exceeds the preset driving speed to the duration that the driving speed does not exceed the preset driving speed, the duration of driving when the temperature exceeds the preset temperature threshold, the ratio of the duration that the acceleration exceeds the preset acceleration threshold to the total duration of the current charging and discharging stroke, and the ratio of the duration that the deceleration exceeds the preset deceleration threshold to the total duration of the current charging and discharging stroke.

[0134] In some embodiments, the above-mentioned S304 may specifically include: determining the external temperature, various data included in the operating data, empirical characteristics, statistical characteristics, voltage correlation characteristics, voltage range change characteristics and temperature range change characteristics as time domain characteristics; wherein the empirical characteristics include at least one of the following: charge and discharge depth, charging overcurrent ratio, proportion of time when the state of charge is lower than the preset state of charge, proportion of energy recovery time and proportion of energy recovery current steps.

[0135] It should be noted that the charge and discharge depth is used to indicate the charging capacity of the charging stroke, the charging overcurrent ratio is used to indicate the proportion of the time when the current in the charging stroke exceeds the preset current to the duration of the entire charging stroke, the time ratio of the state of charge lower than the preset state of charge indicates the proportion of the time when the capacity (i.e., the state of charge) in the charging stroke (or discharge stroke) is lower than the preset capacity (i.e., the preset state of charge) to the duration of the entire stroke, the energy recovery time ratio indicates the proportion of the energy recovery time in the discharge stroke to the duration of the entire discharge stroke, and the energy recovery current step ratio indicates the proportion of the duration of each recovery stage in the energy recovery process to the duration of the entire discharge stroke.

[0136] In an embodiment of the present application, when determining time domain features, the present application may determine the external temperature, various data included in the operating data, empirical features, statistical features, voltage correlation features, voltage range change features, and temperature range change features as time domain features. In this way, by determining the depth of charge and discharge, the charging overcurrent ratio, the proportion of time when the state of charge is lower than the preset state of charge, the proportion of energy recovery time and the proportion of energy recovery current steps, as well as the external temperature, the proportion of time when the driving speed exceeds the preset driving speed and the proportion of time when the driving speed does not exceed the preset driving speed, the duration of driving when the temperature exceeds the preset temperature threshold, the proportion of time when the acceleration exceeds the preset acceleration threshold to the total duration of the current charge and discharge trip, and the proportion of time when the deceleration exceeds the preset deceleration threshold to the total duration of the current charge and discharge trip, these data are determined as time domain features. When predicting thermal runaway of a vehicle, the accuracy of the prediction can be improved.

[0137] In an embodiment of the present application, when determining the time domain characteristics, the present application can perform data analysis based on the current time series data, voltage time series data, and resistance time series data of the power battery included in the battery status data, as well as the voltage time series data and temperature time series data of the battery cell to determine the statistical characteristics. And based on the voltage time series data of the battery cell, determine the voltage correlation characteristics between the battery cells. And, based on the voltage time series data and temperature time series data of the battery cell, determine the voltage travel change characteristics and temperature travel change characteristics of the battery cell. Thus, based on the operating data, environmental data, statistical characteristics, voltage correlation characteristics, voltage travel change characteristics, and temperature travel change characteristics, the time domain characteristics can be determined from a multi-dimensional perspective. In this way, by analyzing multi-dimensional data, the time domain characteristics can be accurately determined based on the comprehensive data during vehicle operation.

[0138] In some embodiments, as Figure 6 As shown, the above S203 may specifically include S601-S603:

[0139] S601 : Perform frequency domain transformation on target time series data in the highest cell voltage time series data and the highest cell temperature time series data to obtain frequency domain data.

[0140] In one possible implementation, minor electrochemical anomalies within the power battery may manifest as subtle fluctuations in the time domain (i.e., fluctuations that do not exceed the anomaly threshold), making the anomaly unrecognizable. Therefore, frequency domain analysis can be combined to identify sudden thermal runaway, as the data fluctuations occur in the frequency domain.

[0141] In one possible implementation, the Fast Fourier Transform (FFT) can be used to convert time-domain data into the frequency domain, revealing the distribution and intensity of different frequency components in the data. FFT analysis can clearly capture periodic features in the data and identify anomalous frequency components.

[0142] In a possible implementation, the highest cell voltage timing data and the highest cell temperature timing data can be Fourier transformed using Formula 7 to obtain frequency domain data. .

[0143]

[0144] in, Indicates the highest cell voltage time series data or the highest cell temperature time series data at time i, Represents the frequency domain data of the kth frequency component.

[0145] S602: Determine the frequency amplitude and power spectrum based on the frequency domain data.

[0146] Furthermore, the frequency domain data after Fourier transformation is converted into Each value of the first half of the data in multiply by 2 and divide by n to get the final frequency amplitude .

[0147]

[0148] It should be noted that the frequency amplitude is determined When the complete frequency domain data is Extract the first half of the elements (that is, , multiply the extracted elements by 2, and then divide the result by the number of data after Fourier transform n to obtain the final normalized unilateral spectrum (i.e. frequency amplitude ).

[0149] This is because after Fourier transforming a real signal, a symmetrical spectrum is obtained, in which the 0th point corresponds to the DC component, the first half (from the 1st point to the (n / 2)-1th point) represents the positive frequency component, and the second half (from the n / 2th point to the n-1th point) represents the negative frequency component, and the negative frequency component is conjugate symmetrical with the positive frequency component. Therefore, in order to more conveniently analyze the spectrum, a one-sided spectrum is usually used, that is, only the DC component and the positive frequency component are retained, while the amplitude of the positive frequency component is doubled to keep the total energy unchanged. Finally, the result is divided by the number of data after the Fourier transform, n, in order to normalize the Fourier transform result, so that the frequency amplitude corresponds to the actual amplitude of the original signal.

[0150] Then, the power spectrum is calculated using the direct method using formula 9. .

[0151]

[0152] S603: Determine frequency domain features based on the frequency amplitude and power spectrum.

[0153] The frequency domain features include at least one of the following: a DC component, a center of gravity frequency, a frequency standard deviation, and a root mean square frequency.

[0154] In one possible implementation, based on the power spectrum and frequency amplitude , determine the characteristics such as DC component, center of gravity frequency, frequency standard deviation, and root mean square frequency.

[0155] It should be noted that the DC component is the first value in the FFT sequence.

[0156] In a possible implementation, the power spectrum can be calculated by formula 10. and frequency amplitude Determine the center of gravity frequency.

[0157]

[0158] In a possible implementation, the power spectrum can be calculated by formula 11: and frequency amplitude Determine the frequency standard deviation.

[0159]

[0160] Among them, FC represents the reference function, which also receives frequency f as input.

[0161] In a possible implementation, the power spectrum can be calculated by formula 12. and frequency amplitude Determine the frequency standard deviation.

[0162]

[0163] In an embodiment of the present application, the present application can perform a frequency domain transformation on the highest cell voltage time series data or the highest cell temperature time series data to obtain frequency domain data corresponding to the voltage or temperature. Then, based on the frequency domain data corresponding to the voltage or temperature, the frequency amplitude and power spectrum corresponding to the voltage or temperature are determined, and then the corresponding frequency domain characteristics are determined based on the frequency amplitude and power spectrum. In this way, by further performing a frequency domain transformation on the data to obtain frequency domain characteristics, the accuracy of the prediction can be further improved when predicting thermal runaway of the vehicle.

[0164] In some embodiments, as Figure 7 As shown, the above S203 may further include S701-S702:

[0165] S701 : Perform wavelet transform on target time series data in the highest cell voltage time series data and the highest cell temperature time series data to obtain high-frequency coefficients and low-frequency coefficients.

[0166] In a possible implementation, the high-frequency coefficient may be a coefficient corresponding to a frequency greater than a preset frequency, and the low-frequency coefficient may be a coefficient corresponding to a frequency less than or equal to the preset frequency.

[0167] It's important to note that the Wavelet Transform (WT) combines both time-domain and frequency-domain analysis capabilities. By selecting appropriate wavelet basis functions and performing multi-resolution decomposition on the data, it's possible to observe overall trends while also focusing on local details, capturing data characteristics at different time and frequency scales. For sudden thermal runaway events, the wavelet transform can capture subtle changes before a sudden abrupt change in the data.

[0168] In one possible implementation, based on Formula 13, multi-layer (e.g., 3-layer, 4-layer) wavelet transform can be performed on the highest cell voltage time series data and the highest cell temperature time series data respectively to obtain high-frequency coefficients and low-frequency coefficients of each layer.

[0169]

[0170] in, Represents the wavelet basis function, a and b represent the scale and displacement factor U respectively t It should be noted that, based on Formula 13, the high-frequency coefficient or low-frequency coefficient of a layer of wavelet transform can be determined each time, and the wavelet basis function when determining the high-frequency coefficient or low-frequency coefficient is different.

[0171] It should be noted that low-frequency coefficients capture the overall trend of the data, while high-frequency coefficients capture details in the data, such as mutations, edges, or noise.

[0172] S702: Determine frequency domain features based on the high-frequency coefficients and the low-frequency coefficients.

[0173] Among them, the frequency domain feature is the energy ratio.

[0174] In one possible implementation, the energy of each layer after wavelet transform can be calculated. The energy of each layer is equal to the sum of the squares of all coefficients of the layer. Finally, the energy proportion of each layer is calculated as the frequency domain feature to describe the frequency domain changes of the data.

[0175] In an embodiment of the present application, a wavelet transform can also be performed on the highest cell voltage time series data or the highest cell temperature time series data to obtain high-frequency coefficients and low-frequency coefficients. Based on the high-frequency coefficients and low-frequency coefficients, the energy proportion is then determined as a frequency domain feature. In this way, frequency domain features can also be determined through wavelet transform, further improving the accuracy of thermal runaway prediction for vehicles.

[0176] In some embodiments, the above S204 may specifically include S801-S802:

[0177] S801: Screen the time domain features and frequency domain features to obtain key features.

[0178] S802. Utilize key features to predict thermal runaway of the vehicle.

[0179] As you can understand, through the above calculations, we construct time-domain features using statistical, data transformation, and mechanistic methods, and frequency-domain features using wavelet and fast Fourier transforms. Ultimately, multiple features are constructed for each charging and discharging trip (e.g., 201 features for the charging trip and 247 features for the discharging trip). These features can be filtered to improve subsequent model performance, reduce the risk of overfitting, and accelerate predictions.

[0180] Specifically, a feature discretization method can be used to discretize the eigenvalues ​​of each feature in the time and frequency domains. This involves dividing the feature value range into several equally spaced intervals (for example, five or six intervals) and statistically analyzing the percentage of trips that fall within each interval. For each feature, the mean frequency of each trip is calculated. Based on these statistical results, key features are identified.

[0181] Alternatively, the feature splitting gain calculation method automatically selects important features through the tree model's splitting node mechanism. This method selects features based on the model, which takes into account the interactions between features.

[0182] In this way, feature discretization method and feature splitting gain calculation fusion method are used for feature selection, and finally important and effective charging and discharging features (key features) are screened out for subsequent thermal runaway prediction.

[0183] In some embodiments, the above S802 may specifically include: inputting key features into a thermal runaway prediction model to obtain a thermal runaway risk level of the vehicle.

[0184] It should be noted that the thermal runaway prediction model is a model trained based on historical data of the full life cycle of sample vehicles.

[0185] In an embodiment of the present application, a pre-trained thermal runaway prediction model can be used to process the screened key features to predict the vehicle's thermal runaway risk level. This allows the thermal runaway prediction model to efficiently and accurately determine the vehicle's thermal runaway risk level, improving early warning efficiency.

[0186] In an embodiment of the present application, the present application can also perform feature screening on time domain features and frequency domain features to obtain key features, and then use the key features to predict vehicle thermal runaway. In this way, the key features obtained after screening can further improve the accuracy of predicting vehicle thermal runaway.

[0187] In some embodiments, the training process of the thermal runaway prediction model includes S901-S903:

[0188] S901. Determine multiple historical charging and discharging trips of the sample vehicle based on historical data of the entire life cycle of the sample vehicle.

[0189] In one possible implementation, historical data of the entire life cycle of sample vehicles can be obtained through a cloud-based big data center platform, where the sample vehicles include: data on the entire life cycle of faulty vehicles (i.e., vehicles that have experienced thermal runaway) and normal vehicles (i.e., vehicles that have not experienced thermal runaway).

[0190] In one possible implementation, the historical data of the entire life cycle may include: voltage list, temperature list, total current, insulation resistance, terminal time, vehicle charging status, vehicle speed, total mileage, SOC, longitude and latitude, outside vehicle temperature, etc.

[0191] In one possible implementation, the historical data of the sample vehicle's full life cycle can be sorted in ascending order by terminal time and total mileage. The historical data of the sample vehicle's full life cycle can also be preprocessed to remove abnormal data and handle missing values.

[0192] Specifically, rows where the total current exceeds a preset value (e.g., 10,000A, 9,000A, etc.) or where the SOC exceeds a threshold (e.g., 100) can be deleted. Missing values ​​are also handled. For example, if the voltage and temperature lists exist and are reasonable, and other data is missing for a continuous period exceeding a preset duration (e.g., 1 or 2 minutes), the rows containing the missing values ​​are deleted (i.e., all data corresponding to the missing data period is deleted). Otherwise, if other data is missing for a continuous period not exceeding the preset duration, the missing values ​​are filled in using an upward fill-up method.

[0193] In one possible implementation, linear interpolation can be used to calculate the battery's state of health (SOH) parameters based on the battery model and total mileage. The voltage and temperature lists are then separated, and the maximum voltage (temperature), minimum voltage (temperature), median voltage (temperature), and pressure (temperature) differential are calculated over time.

[0194] In addition, the total voltage and total current data are combined to further calculate the power and internal resistance parameters. The fusion analysis of these multi-source data provides rich and effective information support for the early warning of sudden thermal runaway.

[0195] In some embodiments, S901 may specifically include: determining multiple historical trips of the sample vehicle based on historical data, and determining multiple historical charging and discharging trips based on the multiple historical trips, wherein no historical data exists for a time period between two adjacent historical trips, and the time interval between the time periods is greater than a preset duration.

[0196] In a possible implementation, the historical data of the entire life cycle of each faulty vehicle and normal vehicle included in the sample vehicles may be divided into individual charging trips or discharging trips (ie, multiple historical trips).

[0197] In one possible implementation, trips can be divided based on the time series of historical data from each vehicle's full lifecycle. If no data exists within a preset time interval, this data can be used to identify trips. Furthermore, the cumulative mileage before and after the data is collected is calculated to determine the change in total mileage, representing the change in total mileage based on the wheel speed sensors. Then, based on the time difference, mileage difference, and state of charge change, corresponding thresholds are set to classify trips as valid charge and discharge trips.

[0198] In a possible implementation, the difference in state of charge between the data before and after can also be calculated to represent the change in battery power in the time series data.

[0199] In an embodiment of the present application, the present application can analyze historical data from the entire life cycle of a sample vehicle, thereby dividing the historical data into multiple historical trips based on the time difference between the data, thereby determining multiple historical charge and discharge trips. In this way, based on the generation time points of the vehicle charge and discharge data in the historical data, multiple historical charge and discharge trips can be determined. Then, by analyzing the data of multiple historical charge and discharge trips, corresponding features can be obtained.

[0200] In some embodiments, the historical charge and discharge trips include charging trips and discharging trips; based on multiple historical trips, multiple historical charge and discharge trips are determined, including: for each historical trip, if the historical trip satisfies the condition that the sample vehicle is in a charging state and the mileage is less than a preset mileage, the historical trip is determined to be a charging trip; or, if the historical trip satisfies the condition that the sample vehicle is in a discharging state and the driving time is greater than a preset driving time, the historical trip is determined to be a discharging trip.

[0201] In one possible implementation, if the vehicle's charging state in a historical trip is charging, it is necessary to further determine whether the state of charge increases monotonically over time and whether the total mileage change (i.e., mileage) exceeds a preset mileage. This determines that the historical trip was a charging trip.

[0202] In one possible implementation, when the vehicle charging state in a historical trip is a discharging state, it is necessary to further determine whether the vehicle's driving time is greater than a preset driving time (e.g., 60 seconds, 100 seconds). If so, the historical trip is determined to be a discharging trip.

[0203] In an embodiment of the present application, the present application can determine whether a trip is a charging trip or a discharging trip based on the vehicle's charging status, mileage, discharging status, and driving duration in the data corresponding to each historical trip. Therefore, in subsequent data analysis, corresponding features can be determined based on the characteristics of the charging trip and the discharging trip, respectively.

[0204] S902: Determine thermal runaway risk levels for multiple historical charge and discharge cycles.

[0205] In some embodiments, the above S902 may specifically include: determining the time intervals between multiple historical charging and discharging trips and the time when thermal runaway of the sample vehicle occurs; based on the time intervals, determining the thermal runaway risk levels of the multiple historical charging and discharging trips; wherein, the closer the time interval, the higher the thermal runaway risk level of the corresponding historical charging and discharging trip.

[0206] In one possible implementation, a thermal runaway risk level can be determined for each trip of a faulty vehicle, including the sample vehicles, based on the time interval from the overheating event (i.e., the moment thermal runaway occurred). Thermal runaway risk levels can include: no risk, low risk, high risk, etc. For normal vehicles, each trip can be determined as no risk.

[0207] It can be understood that each trip of the faulty vehicle in which the duration between the trip and the overheating time is less than the first duration can be determined as a high-risk level, a trip in which the duration between the trip and the overheating time is greater than or equal to the first duration and less than the second duration can be determined as a low-risk level, and a trip in which the duration between the trip and the overheating time is greater than or equal to the second duration can be determined as a no-risk level.

[0208] In this way, time-frequency domain features can be constructed based on the vehicle's original data and multi-source time series fusion parameters in the travel dimension to characterize vehicle status, user behavior habits, external environment, etc.

[0209] In an embodiment of the present application, when determining the thermal runaway risk level of multiple historical charging and discharging trips, the present application can determine the corresponding thermal runaway risk level based on the time interval between each historical charging and discharging trip and the moment when thermal runaway occurred in the sample vehicle, so that the closer the time interval, the higher the thermal runaway risk level of the historical charging and discharging trip.

[0210] S903: Based on the key features and thermal runaway risk levels of multiple historical charge and discharge trips, train a thermal runaway prediction model.

[0211] In one possible implementation, the key features of the historical charge and discharge trip can refer to the above embodiment. After processing the battery status data, operation data and environmental data of the current charge and discharge trip, the obtained time domain features and frequency domain features are subjected to feature screening. The method for determining the obtained key features will not be repeated here.

[0212] In this way, based on the historical data of multiple charging and discharging trips throughout the life cycle of a sample vehicle, we constructed time-domain features using statistical, data transformation, and mechanistic methods, taking into account dimensions such as vehicle battery status, driving habits, and external environment. We also constructed frequency-domain features using wavelet transforms and fast Fourier transforms. This resulted in multiple features for each charging and discharging trip, and we selected these features to improve subsequent model performance, reduce the risk of overfitting, and accelerate training.

[0213] Specifically, a feature discretization method or a feature splitting gain calculation method may be used to determine key features of multiple historical charge and discharge trips.

[0214] In one possible implementation, to accurately test the model's effectiveness, the key features of multiple historical charge and discharge trips can be divided into training and test datasets based on the vehicle dimension. Specifically, 40% of faulty vehicles and 40% of normal vehicles can be randomly selected as the test set, and the remaining vehicle data can be used as the training set.

[0215] Furthermore, given that there are fewer samples of faulty vehicles, resulting in a serious imbalance in class samples, the adaptive synthetic oversampling method (ADASYN) and the auxiliary classifier generative adversarial network (ACGAN) can be used to solve the sample imbalance problem.

[0216] In this way, the generated training samples and real training samples are combined to form a new training set. The AI ​​model is trained based on this new training set, and the optimal hyperparameters of the model are selected using Bayesian search and grid search methods. The selected optimal model is then deployed in the cloud, and a front-end interface is developed for real-time monitoring. If a high-risk vehicle is found, a recall is immediately initiated, while low-risk vehicles are continuously tracked and observed.

[0217] Specifically, to achieve accurate early warning, the system sets a minimum risk probability threshold, a time window, the percentage of vehicles exceeding the minimum risk probability threshold, and the curvature and dispersion parameters of the travel probability curve within the time window. Based on these parameters, a three-level early warning mechanism is established, categorizing risk levels into high risk, low risk, and no risk. Appropriate measures are implemented for vehicles of different risk levels: high-risk vehicles are immediately recalled; low-risk vehicles are continuously tracked and monitored; and no-risk vehicles are left untouched.

[0218] For example, Figure 8 As shown in the figure, by acquiring raw vehicle data and multi-source time series fusion parameters and processing the acquired data, time-domain and frequency-domain features are derived. Time-domain features include statistical features, correlation features, travel variation features, and empirical features, while frequency-domain features include FFT-related features derived from the Fast Fourier Transform (FFT) and WT-related features derived from the Wavelet Transform (WT). A feature table is constructed based on these time-domain and frequency-domain features, and feature discretization and feature splitting gain calculation methods are used to filter these features and identify valid key features. A model is then constructed based on these valid key features.

[0219] For example, Figure 9 As shown, the thermal runaway prediction model may include a charging trip warning model and a discharging trip warning model. The charging trip warning model uses charging trip data from sample vehicles to determine charging characteristics and risk levels, obtaining a dataset. The dataset is then divided into a training dataset and a testing dataset. Data augmentation is then performed on the training dataset to train the model. The trained model is then tested on the testing dataset to obtain the charging trip warning model. The discharging trip warning model uses discharging trip data from sample vehicles to determine discharge characteristics and risk levels, obtaining a dataset. The dataset is then divided into a training dataset and a testing dataset. Data augmentation is then performed on the training dataset to train the model. The trained model is then tested on the testing dataset to obtain the discharging trip warning model. The charging and discharging trip warning models are then deployed in the cloud to determine the charging and discharging trip risk probabilities of the vehicle being predicted in real time. Furthermore, a risk probability trend chart for the charging and discharging trips is constructed based on a chronological order. Based on the three-level warning risk level, the risk level of the vehicle being predicted is determined as high risk, low risk, or no risk. In addition, high-risk vehicles can be recalled through invitation, low-risk vehicles can be continuously monitored, and risk-free vehicles can remain untouched.

[0220] In an embodiment of the present application, when pre-training a thermal runaway prediction model, the present application may analyze historical data from the full lifecycle of a sample vehicle, thereby dividing the full lifecycle of the sample vehicle into multiple historical charge and discharge cycles. Thus, by determining the thermal runaway risk level of each of these multiple historical charge and discharge cycles and combining the key features of these multiple historical charge and discharge cycles, a thermal runaway prediction model can be trained. By pre-training the thermal runaway prediction model, the efficiency and accuracy of predicting a vehicle's thermal runaway risk level can be improved.

[0221] This application embodiment provides an effective early warning method for sudden thermal runaway. By integrating long-term vehicle battery performance parameters, environmental operating parameters, and user behavior characteristics, this method constructs a multi-dimensional dynamic coupling model, enabling early warning within hours to days. This not only provides a key decision-making basis for vehicle recalls and maintenance, effectively preventing and controlling overheating risks, but also helps automakers improve their product safety reputation and lay a solid safety foundation for the high-quality development of the new energy vehicle industry.

[0222] The above mainly introduces the solution provided by the embodiment of the present application from the perspective of the method. In order to achieve the above functions, the power battery thermal runaway prediction device or electronic device includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0223] The embodiment of the present application can, according to the above method, exemplarily divide the functional modules of the power battery thermal runaway prediction device or electronic device. For example, the power battery thermal runaway prediction device or electronic device may include various functional modules corresponding to the various functional divisions, or two or more functions may be integrated into one processing module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present application is schematic and is only a logical functional division. There may be other division methods in actual implementation.

[0224] Reference Figure 10The power battery thermal runaway prediction device 1000 includes: a processing module 1001 and a prediction module 1002; the processing module 1001 is used to determine time domain characteristics based on battery status data, operating data and environmental data of the current charge and discharge journey of the vehicle power battery, wherein the battery status data includes voltage time series data and temperature time series data of the battery cells in the power battery; the processing module 1001 is also used to determine the highest cell voltage time series data and the highest cell temperature time series data of the power battery based on the voltage time series data and the temperature time series data; the processing module 1001 is also used to determine frequency domain characteristics based on the highest cell voltage time series data and the highest cell temperature time series data; the prediction module 1002 is used to predict thermal runaway of the vehicle based on the time domain characteristics and the frequency domain characteristics.

[0225] In one possible embodiment, the battery status data also includes: current time series data, voltage time series data and resistance time series data of the power battery; the processing module 1001 is specifically used to determine statistical characteristics based on the current time series data, voltage time series data and resistance time series data of the power battery, and the voltage time series data and temperature time series data of the battery cell; the processing module 1001 is specifically used to determine the voltage correlation characteristics between battery cells based on the voltage time series data of the battery cell; the processing module 1001 is specifically used to determine the voltage travel change characteristics and temperature travel change characteristics of the battery cell based on the voltage time series data and temperature time series data of the battery cell; the processing module 1001 is specifically used to determine the time domain characteristics based on the operating data, environmental data, statistical characteristics, voltage correlation characteristics, voltage travel change characteristics and temperature travel change characteristics.

[0226] In one possible implementation, the processing module 1001 is specifically used to determine the cell voltage key value and the cell temperature key value of the power battery based on the voltage timing data and temperature timing data of the battery cell, wherein the cell voltage key value includes the maximum value, minimum value, median value and maximum voltage difference of the cell voltage; the cell temperature key value includes the maximum value, minimum value, median value and maximum temperature difference of the cell temperature; the processing module 1001 is specifically used to determine the current timing data, voltage timing data and resistance timing data of the power battery, as well as the statistical characteristics of the cell voltage key value and the cell temperature key value of the power battery, wherein the statistical characteristics include at least one of the following: the maximum value, mean value, effective value, peak-to-peak value, difference ratio, skewness and peak factor in the current charge and discharge cycle.

[0227] In one possible implementation, the processing module 1001 is specifically used to determine the voltage mean of the battery cells in the current charge and discharge cycle based on the voltage timing data of the battery cells; the processing module 1001 is specifically used to determine the voltage correlation characteristics between the battery cells using the Pearson correlation coefficient calculation method based on the voltage timing data and the voltage mean of the battery cells.

[0228] In a possible implementation, the processing module 1001 is specifically configured to calculate change characteristics of a cell voltage key value and a cell temperature key value of the power battery, wherein the change characteristics include at least one of the following: an increase ratio, a maximum change amount, and a change rate.

[0229] In one possible embodiment, the environmental data includes the outside temperature; the operating data includes at least one of the following: the ratio of the duration that the driving speed exceeds the preset driving speed to the duration that the driving speed does not exceed the preset driving speed, the duration of driving when the temperature exceeds the preset temperature threshold, the ratio of the duration that the acceleration exceeds the preset acceleration threshold to the total duration of the current charging and discharging stroke, and the ratio of the duration that the deceleration exceeds the preset deceleration threshold to the total duration of the current charging and discharging stroke; the processing module 1001 is specifically used to determine the outside temperature, various data included in the operating data, empirical characteristics, statistical characteristics, voltage correlation characteristics, voltage stroke change characteristics and temperature stroke change characteristics as time domain characteristics; wherein, the empirical characteristics include at least one of the following: charge and discharge depth, charging overcurrent ratio, the ratio of the duration when the state of charge is lower than the preset state of charge, the ratio of energy recovery time and the ratio of energy recovery current steps.

[0230] In one possible embodiment, the processing module 1001 is specifically used to perform frequency domain transformation on the target timing data in the highest cell voltage timing data and the highest cell temperature timing data to obtain frequency domain data; the processing module 1001 is specifically used to determine the frequency amplitude and power spectrum based on the frequency domain data; the processing module 1001 is specifically used to determine the frequency domain characteristics based on the frequency amplitude and power spectrum, wherein the frequency domain characteristics include at least one of the following: DC component, center of gravity frequency, frequency standard deviation and root mean square frequency.

[0231] In one possible embodiment, the processing module 1001 is specifically used to perform wavelet transform on the target timing data in the highest cell voltage timing data and the highest cell temperature timing data to obtain high-frequency coefficients and low-frequency coefficients; the processing module 1001 is specifically used to determine the frequency domain characteristics based on the high-frequency coefficients and the low-frequency coefficients; wherein the frequency domain characteristics are energy proportions.

[0232] In a possible implementation, the processing module 1001 is specifically used to perform feature screening on time domain features and frequency domain features to obtain key features; the prediction module 1002 is specifically used to use the key features to predict thermal runaway of the vehicle.

[0233] In a possible implementation, the processing module 1001 is specifically configured to enable the prediction module 1002 to input key features into a thermal runaway prediction model to obtain a thermal runaway risk level of the vehicle.

[0234] In one possible implementation, the processing module 1001 is specifically used to determine multiple historical charging and discharging trips of the sample vehicle based on historical data of the full life cycle of the sample vehicle; the processing module 1001 is specifically used to determine the thermal runaway risk levels of the multiple historical charging and discharging trips; the processing module 1001 is specifically used to train a thermal runaway prediction model based on the key features and thermal runaway risk levels of the multiple historical charging and discharging trips.

[0235] In one possible implementation, the processing module 1001 is specifically configured to determine multiple historical trips of a sample vehicle based on historical data, wherein no historical data exists for a time period between two adjacent historical trips, and the time interval between the time periods is greater than a preset duration; the processing module 1001 is specifically configured to determine multiple historical charging and discharging trips based on the multiple historical trips.

[0236] In one possible implementation, the processing module 1001 is specifically configured to, for each historical trip, determine that the historical trip is a charging trip if the historical trip satisfies the conditions that the sample vehicle is in a charging state and the mileage is less than a preset mileage; or, the processing module 1001 is specifically configured to determine that the historical trip is a discharging trip if the historical trip satisfies the conditions that the sample vehicle is in a discharging state and the driving duration is greater than a preset driving duration.

[0237] In one possible implementation, the processing module 1001 is specifically configured to determine the time intervals between multiple historical charge and discharge trips and the moment when thermal runaway of a sample vehicle occurs; and based on the time intervals, determine the thermal runaway risk levels of the multiple historical charge and discharge trips; wherein, the closer the time interval, the higher the thermal runaway risk level of the corresponding historical charge and discharge trip.

[0238] Regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0239] like Figure 11 As shown, the electronic device 1100 includes but is not limited to: a processor 1101 and a memory 1102 .

[0240] The memory 1102 is configured to store executable instructions of the processor 1101. It is understood that the processor 1101 is configured to execute instructions to implement the power battery thermal runaway prediction method in the above embodiment.

[0241] It should be noted that those skilled in the art can understand that Figure 11 The structure of the electronic device 1100 shown in FIG. 1 does not limit the electronic device 1100. The electronic device 1100 may include Figure 11More or fewer components may be shown, or certain components may be combined, or the components may be arranged differently.

[0242] The processor 1101 is the control center of the electronic device 1100. It connects the various parts of the entire electronic device 1100 using various interfaces and lines. By running or executing software programs and / or modules stored in the memory 1102 and calling data stored in the memory 1102, it performs various functions of the electronic device 1100 and processes data, thereby monitoring the electronic device 1100 as a whole. The processor 1101 may include one or more processing units. Optionally, the processor 1101 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly handles wireless communications. It is understood that the above-mentioned modem processor may not be integrated into the processor 1101.

[0243] Memory 1102 can be used to store software programs and various data. Memory 1102 may primarily include a program storage area and a data storage area. The program storage area may store an operating system, application programs required by at least one functional module, and the like. Furthermore, memory 1102 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state memory device.

[0244] In an exemplary embodiment, a computer-readable storage medium including instructions is further provided, such as a memory 1102 including instructions. The above instructions can be executed by the processor 1101 of the electronic device 1100 to implement the power battery thermal runaway prediction method in the above embodiment.

[0245] In actual implementation, Figure 10 The functions of the processing module 1001 and the prediction module 1002 in Figure 11 The processor 1101 in the memory 1102 is used to call the computer program stored in the memory 1102. The specific execution process can be referred to the description of the power battery thermal runaway prediction method in the above embodiment, which will not be repeated here.

[0246] Optionally, the computer-readable storage medium may be a non-temporary computer-readable storage medium, for example, the non-temporary computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, etc.

[0247] In an exemplary embodiment, the present application also provides a computer program product including one or more instructions, which can be executed by the processor 1101 of the electronic device 1100 to complete the power battery thermal runaway prediction method in the above embodiment.

[0248] It should be noted that when the instructions in the above-mentioned computer-readable storage medium or one or more instructions in the computer program product are executed by the processor of the electronic device, the various processes of the above-mentioned power battery thermal runaway prediction method embodiment are implemented, and the same technical effect as the above-mentioned power battery thermal runaway prediction method can be achieved. To avoid repetition, they will not be repeated here.

[0249] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0250] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of modules or units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0251] Units described as separate components may or may not be physically separate, and components shown as units may be one physical unit or multiple physical units, that is, they may be located in one place or distributed in multiple places. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0252] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0253] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application, or the part that contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product. The software product is stored in a storage medium and includes a number of instructions for causing a device (which can be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, disk or optical disk, etc. Various media that can store program code.

[0254] The above are only specific embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or replacements within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for predicting thermal runaway of a power battery, characterized in that: The power battery thermal runaway prediction method includes: Determine the time domain characteristics based on the battery status data, operating data, and environmental data of the vehicle's power battery's current charge and discharge journey, wherein the battery status data includes voltage time series data and temperature time series data of the battery cells in the power battery; the environmental data includes the ambient temperature; and the operating data includes: the ratio of the duration of time when the driving speed exceeds a preset driving speed to the duration of time when the driving speed does not exceed the preset driving speed, the ratio of the duration of time when the acceleration exceeds a preset acceleration threshold to the total duration of the current charge and discharge journey, and the ratio of the duration of time when the deceleration exceeds a preset deceleration threshold to the total duration of the current charge and discharge journey; Determining maximum cell voltage time series data and maximum cell temperature time series data of the power battery based on the voltage time series data and the temperature time series data; the maximum cell voltage time series data is a maximum value among the voltages of a plurality of battery cells included in the power battery at each moment, and the maximum cell temperature time series data is a maximum value among the temperatures of a plurality of battery cells included in the power battery at each moment; determining frequency domain features based on the highest cell voltage time series data and the highest cell temperature time series data; in a case where the frequency domain features are obtained by frequency domain transformation, the frequency domain features include: centroid frequency, frequency standard deviation, and root mean square frequency; the frequency domain transformation is used to obtain frequency domain data, and determine the frequency domain features based on a frequency amplitude and a power spectrum determined by the frequency domain data, the frequency amplitude being obtained by multiplying each value of the first half of the frequency domain data by 2 and dividing by n; the power spectrum being obtained by multiplying the square of the frequency amplitude by 2 and dividing by n; the centroid frequency, the frequency standard deviation, and the root mean square frequency being determined based on the frequency amplitude and the power spectrum, respectively; and the frequency domain transformation includes Fourier transformation; or, in a case where the frequency domain features are obtained by wavelet transformation, the frequency domain features include energy proportion; the wavelet transformation is used to obtain high-frequency coefficients and low-frequency coefficients, and determine the frequency domain features based on the high-frequency coefficients and the low-frequency coefficients; the wavelet transformation includes multi-layer wavelet transformation, and the energy proportion of each wavelet transform layer is determined based on the high-frequency coefficients and the low-frequency coefficients included in each wavelet transform layer; Based on the time domain characteristics and the frequency domain characteristics, thermal runaway prediction is performed on the vehicle.

2. The method for predicting thermal runaway of a power battery according to claim 1, characterized in that: The battery status data also includes: current time series data, voltage time series data, and resistance time series data of the power battery; the time domain characteristics are determined based on the battery status data, operating data, and environmental data of the current charge and discharge journey of the vehicle power battery, including: Determining statistical features based on the current time series data, voltage time series data, and resistance time series data of the power battery, and the voltage time series data and temperature time series data of the battery cell; determining voltage correlation characteristics between battery cells based on the voltage time series data of the battery cells; Determining voltage travel change characteristics and temperature travel change characteristics of the battery cell based on the voltage time series data and temperature time series data of the battery cell; The time domain feature is determined based on the operating data, the environmental data, the statistical feature, the voltage correlation feature, the voltage range change feature, and the temperature range change feature.

3. The method for predicting thermal runaway of a power battery according to claim 2, characterized in that: The determining of statistical features based on the current time series data, voltage time series data, and resistance time series data of the power battery, and the voltage time series data and temperature time series data of the battery cell, includes: Determine the cell voltage key value and cell temperature key value of the power battery based on the voltage time series data and temperature time series data of the battery cell, wherein the cell voltage key value includes the maximum value, minimum value, median value and maximum voltage difference of the cell voltage; the cell temperature key value includes the maximum value, minimum value, median value and maximum temperature difference of the cell temperature; Determine statistical characteristics of the current time series data, voltage time series data, and resistance time series data of the power battery, as well as the key values ​​of cell voltage and cell temperature of the power battery, wherein the statistical characteristics include at least one of the following: the maximum value, mean value, effective value, peak-to-peak value, difference ratio, skewness, and crest factor in the current charge and discharge trip.

4. The method for predicting thermal runaway of a power battery according to claim 2, wherein: The determining of voltage correlation characteristics between battery cells based on the voltage time series data of the battery cells includes: Determining a voltage average of the battery cell in a current charge and discharge cycle based on the voltage time series data of the battery cell; Based on the voltage time series data of the battery cells and the voltage mean, a Pearson correlation coefficient calculation method is adopted to determine the voltage correlation characteristics between the battery cells.

5. The method for predicting thermal runaway of a power battery according to claim 3, wherein: The determining of the voltage travel variation characteristics and the temperature travel variation characteristics of the battery cell based on the voltage time series data and the temperature time series data of the battery cell includes: Calculate the change characteristics of the key values ​​of the cell voltage and the cell temperature of the power battery, wherein the change characteristics include at least one of the following: an increase ratio, a maximum change amount, and a change rate.

6. The method for predicting thermal runaway of a power battery according to claim 2, characterized in that: The determining the time domain feature based on the operating data, the environmental data, the statistical feature, the voltage correlation feature, the voltage range change feature, and the temperature range change feature includes: The external temperature, the various data included in the operating data, the empirical characteristics, the statistical characteristics, the voltage correlation characteristics, the voltage range change characteristics and the temperature range change characteristics are determined as the time domain characteristics; wherein the empirical characteristics include at least one of the following: charge and discharge depth, charging overcurrent ratio, proportion of time when the state of charge is lower than the preset state of charge, proportion of energy recovery time and proportion of energy recovery current steps.

7. The method for predicting thermal runaway of a power battery according to claim 1, wherein: The determining of frequency domain characteristics based on the highest cell voltage time series data and the highest cell temperature time series data includes: For target time series data in the highest cell voltage time series data and the highest cell temperature time series data, performing frequency domain transformation on the target time series data to obtain frequency domain data; determining a frequency amplitude and a power spectrum based on the frequency domain data; Frequency domain features are determined based on the frequency amplitude and the power spectrum, wherein the frequency domain features include at least one of the following: a direct current component, a center of gravity frequency, a frequency standard deviation, and a root mean square frequency.

8. The method for predicting thermal runaway of a power battery according to claim 7, characterized in that: The determining of frequency domain characteristics based on the highest cell voltage time series data and the highest cell temperature time series data includes: For target time series data in the highest cell voltage time series data and the highest cell temperature time series data, performing wavelet transform on the target time series data to obtain high-frequency coefficients and low-frequency coefficients; Based on the high-frequency coefficients and the low-frequency coefficients, a frequency domain feature is determined; wherein the frequency domain feature is an energy proportion.

9. The method for predicting thermal runaway of a power battery according to claim 1, wherein: The predicting of thermal runaway of the vehicle based on the time domain features and the frequency domain features includes: Performing feature screening on the time domain features and the frequency domain features to obtain key features; Thermal runaway prediction is performed on the vehicle using the key features.

10. The method for predicting thermal runaway of a power battery according to claim 9, characterized in that: The method of predicting thermal runaway of the vehicle by utilizing the key features includes: The key features are input into a thermal runaway prediction model to obtain a thermal runaway risk level of the vehicle.

11. The method for predicting thermal runaway of a power battery according to claim 10, characterized in that: The training process of the thermal runaway prediction model includes: Determining a plurality of historical charging and discharging trips of the sample vehicle based on historical data of the full life cycle of the sample vehicle; determining thermal runaway risk levels of the plurality of historical charge and discharge trips; The thermal runaway prediction model is trained based on a plurality of key features and thermal runaway risk levels of the historical charge and discharge trips.

12. The method for predicting thermal runaway of a power battery according to claim 11, characterized in that: The determining of a plurality of historical charge and discharge trips of the sample vehicle based on historical data of the full life cycle of the sample vehicle includes: Determining, based on the historical data, a plurality of historical trips of the sample vehicle, wherein no historical data exists for a time period between two adjacent historical trips, and a time interval between the time periods is greater than a preset duration; Based on the plurality of historical trips, a plurality of historical charge and discharge trips are determined.

13. The method for predicting thermal runaway of a power battery according to claim 12, characterized in that: The historical charge and discharge trips include a charge trip and a discharge trip; and determining the multiple historical charge and discharge trips based on the multiple historical trips includes: For each of the historical trips, if the historical trips satisfy the condition that the sample vehicle is in a charging state and the mileage is less than a preset mileage, determining that the historical trip is a charging trip; Alternatively, when the historical trip satisfies the condition that the sample vehicle is in a discharging state and the driving time is greater than a preset driving time, the historical trip is determined to be a discharging trip.

14. The method for predicting thermal runaway of a power battery according to claim 11, characterized in that: Determining the thermal runaway risk levels of the plurality of historical charge and discharge trips includes: Determining the time interval between a plurality of the historical charge and discharge trips and the moment when thermal runaway of the sample vehicle occurs; Based on the time interval, the thermal runaway risk levels of the plurality of historical charge and discharge trips are determined; wherein, the closer the time interval, the higher the thermal runaway risk level of the corresponding historical charge and discharge trip.

15. A power battery thermal runaway prediction device, characterized in that: The power battery thermal runaway prediction device includes: a processing module and a prediction module; The processing module is configured to determine time domain characteristics based on battery status data, operating data, and environmental data of a current charge and discharge trip of a vehicle power battery, wherein the battery status data includes voltage time series data and temperature time series data of battery cells in the power battery; the environmental data includes the ambient temperature; and the operating data includes: a ratio of a duration when the driving speed exceeds a preset driving speed to a duration when the driving speed does not exceed the preset driving speed, a ratio of a duration when the acceleration exceeds a preset acceleration threshold to a total duration of the current charge and discharge trip, and a ratio of a duration when the deceleration exceeds a preset deceleration threshold to a total duration of the current charge and discharge trip; The processing module is further configured to determine, based on the voltage time series data and the temperature time series data, maximum cell voltage time series data and maximum cell temperature time series data of the power battery; the maximum cell voltage time series data is a maximum value among the voltages of a plurality of battery cells included in the power battery at each moment, and the maximum cell temperature time series data is a maximum value among the temperatures of a plurality of battery cells included in the power battery at each moment; The processing module is further used to determine the frequency domain characteristics based on the highest cell voltage time series data and the highest cell temperature time series data; when the frequency domain characteristics are obtained by frequency domain transformation, the frequency domain characteristics include: center of gravity frequency, frequency standard deviation and root mean square frequency, and the frequency domain transformation is used to obtain frequency domain data, so as to determine the frequency domain characteristics based on the frequency amplitude and power spectrum determined by the frequency domain data, wherein the frequency amplitude is obtained by multiplying each value of the first half of the frequency domain data by 2 and dividing by n, and the power spectrum is obtained by multiplying the square of the frequency amplitude by 2 and dividing by n. The centroid frequency, the frequency standard deviation, and the root mean square frequency are determined based on the frequency amplitude and the power spectrum, respectively, and the frequency domain transform includes a Fourier transform; or, in a case where the frequency domain feature is obtained by a wavelet transform, the frequency domain feature includes an energy proportion, the wavelet transform is used to obtain high-frequency coefficients and low-frequency coefficients, and to determine the frequency domain feature based on the high-frequency coefficients and the low-frequency coefficients, the wavelet transform includes a multi-layer wavelet transform, and the energy proportion of each layer of wavelet transform is determined based on the high-frequency coefficients and the low-frequency coefficients included in each layer of wavelet transform; The prediction module is used to predict thermal runaway of the vehicle based on the time domain characteristics and the frequency domain characteristics.

16. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the power battery thermal runaway prediction method according to any one of claims 1 to 14.

17. A vehicle, characterized in that: The vehicle includes the power battery thermal runaway prediction device according to claim 15, and the vehicle is used to implement the power battery thermal runaway prediction method according to any one of claims 1 to 14.

Images