Dynamic thermal runaway early warning method and system based on multi-source data fusion

Abstract

The invention provides a dynamic thermal runaway early warning method and system based on multi-source data fusion. The dynamic thermal runaway early warning method comprises the steps that multi-dimensional operation data of a power battery system in a vehicle and operation state data of the vehicle are collected; extracting a multi-dimensional characteristic quantity representing a battery health state and a potential thermal runaway risk based on the multi-dimensional characteristic quantity, and then judging whether the battery state deviates from a normal range based on a pre-constructed dynamic threshold model to obtain a judgment result; if the judgment result triggers the first early warning, determining a target battery module which triggers the first early warning, and analyzing a thermal runaway precursor signal of the target battery module to obtain a thermal runaway precursor identification result; and finally, according to a judgment result of the dynamic threshold value model and a thermal runaway precursor identification result, outputting a comprehensive risk score and carrying out second early warning. According to the invention, early warning in dozens of seconds or even several minutes in advance can be realized by capturing early weak precursor signals, and by means of the multi-source data fusion and dynamic threshold adjustment technology, the false alarm rate and the missing report rate are remarkably reduced, and the early warning accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of electric vehicle safety, and more specifically, to a dynamic thermal runaway early warning method and system based on multi-source data fusion. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the safety issues of power batteries have become increasingly prominent. Among them, thermal runaway, due to its suddenness and destructiveness, has become the root cause of battery fires, explosions, and other accidents. Currently widely used early warning technologies still have significant shortcomings: First, most rely on single or a few parameters such as voltage and temperature for judgment, resulting in serious delays in early warning and making it difficult to identify weak precursor signals in the early stages of thermal runaway (such as internal short circuits or SEI film decomposition). Second, the common use of static threshold alarms fails to fully consider the normal fluctuation range of batteries under different operating conditions (such as rate, ambient temperature, and SOC), easily leading to false alarms or missed alarms. Third, existing technologies are mostly limited to "post-event alarms," ​​triggering alarms only when thermal runaway has occurred or is about to occur, resulting in an excessively short system response time window, failing to achieve effective early warning and proactive safety protection. Summary of the Invention

[0003] This application provides a dynamic thermal runaway early warning method and system based on multi-source data fusion. It can achieve early warning tens of seconds or even minutes in advance by capturing weak precursor signals in the early stage of thermal runaway. By using multi-source data fusion and dynamic threshold adjustment technology, it significantly reduces the false alarm and false alarm rates and improves the accuracy of early warning. It also enhances the adaptability of the early warning model to changes in environment and operating conditions, enabling it to achieve reliable adaptive early warning under different battery states and operating conditions, thereby solving the above-mentioned technical problems.

[0004] In a first aspect, embodiments of this application provide a dynamic thermal runaway early warning method based on multi-source data fusion, the method comprising: Collect multi-dimensional operational data of the power battery system in the vehicle and operational status data of the vehicle; Based on the multi-dimensional operating data and the operating status data, multi-dimensional feature quantities characterizing the battery health status and potential thermal runaway risk are extracted. Based on the pre-built dynamic threshold model and the multi-dimensional feature quantities, it is determined whether the battery status deviates from the normal range, and the judgment result is obtained. If the determination result triggers the first warning, the target battery module that triggered the first warning is identified, and the thermal runaway precursor signal of the target battery module is analyzed based on the multi-dimensional operating data, the operating status data and the multi-dimensional feature quantity to obtain the thermal runaway precursor identification result; Based on the judgment results of the dynamic threshold model and the identification results of the thermal runaway precursors, a comprehensive risk score is output and a second warning is issued.

[0005] In one possible implementation, the collection of multi-dimensional operational data of the vehicle's power battery system and the vehicle's operational status data includes: The system collects raw multi-category data of each individual cell, the battery module composed of multiple cells, and the entire battery pack of the power battery system in the vehicle, as well as the vehicle's raw operating parameter data; the raw multi-category data includes: voltage and current signals, temperature signals, and pressure signals; The first abnormal data and the second abnormal data that do not conform to the preset standard rules are cleaned from the original multi-category data and the original operating parameter data to obtain the preprocessed multi-category data and operating parameter data. The preprocessed multi-category data and the operating parameter data are normalized to obtain the multi-dimensional operating data and the operating status data.

[0006] In one possible implementation, the step of extracting multi-dimensional features characterizing battery health status and potential thermal runaway risk based on the multi-dimensional operational data and the operational status data includes: The first feature quantity is extracted by separately analyzing the voltage and current signals, temperature signals, and pressure signals in the multi-dimensional operation data within a preset time period. By comprehensively analyzing the voltage and current signals, temperature signals, and pressure signals in the multi-dimensional operational data within a preset time period, the second feature quantity is extracted. The multi-dimensional feature quantity is generated based on the first feature quantity and the second feature quantity; the multi-dimensional feature quantity includes at least: features reflecting voltage consistency, features characterizing temperature distribution and uniformity, features reflecting temperature rise rate, and features revealing electro-thermal coupling relationship.

[0007] In one possible implementation, the multidimensional characteristic quantity includes at least one of the following: voltage standard deviation, maximum voltage difference, highest temperature, lowest temperature, temperature difference, temperature gradient, temperature rise rate, or voltage-temperature coupling characteristic; the temperature rise rate includes the temperature change rate of a single cell in the power battery system.

[0008] In one possible implementation, the step of determining whether the battery state deviates from the normal range based on a pre-built dynamic threshold model and the multi-dimensional feature quantities, and obtaining a determination result, includes: Based on historical health data, a dynamic threshold model is constructed by learning the data distribution pattern of the power battery system under normal operating conditions through machine learning algorithms. The multi-dimensional operating data and the multi-dimensional feature values ​​are input into the dynamic threshold model, and the dynamic threshold model is used to determine whether it deviates from the normal range. If so, the first warning result will be triggered.

[0009] In one possible implementation, the precursory signals of thermal runaway include a small voltage drop, an abnormal temperature rise, and weak oscillations: wherein, The micro voltage drop is defined as follows: when the target cell is in a static or charging phase, the voltage of the target cell undergoes an irreversible negative drift within a preset amplitude range for a period of at least time threshold T1, and the drift exceeds K times (K≥3) the average voltage change of other cells in the same module. The abnormal temperature rise is defined as follows: under the condition that the external heat source and the charging and discharging current are greater than the preset threshold, the temperature rise rate of the target battery cell exceeds the preset dynamic threshold and the duration of this state is greater than T2. ​​At the same time, the temperature difference between the cell and the reference point in the module continues to increase and exceeds the ΔT threshold. The weak oscillation is defined as follows: after detrending the voltage or temperature signal of the target unit, the signal power or amplitude is detected to be higher than the statistical baseline under normal conditions within a preset frequency band by short-time Fourier transform or wavelet transform, and the power enhancement phenomenon in this frequency band continues to exceed T3.

[0010] In one possible implementation, the step of outputting a comprehensive risk score and issuing a second early warning based on the determination result of the dynamic threshold model and the thermal runaway precursor identification result includes: Pre-trained risk assessment models; The judgment result of the dynamic threshold model is weighted and fused with the thermal runaway precursor identification result, and then input into the risk assessment model to trigger a second warning. The risk assessment model combines historical data, current operating conditions, and the results of thermal runaway precursor identification to output a comprehensive risk score. Based on the comprehensive risk score, a warning level is determined and the user is notified.

[0011] In one possible implementation, the warning levels include Level 1, Level 2, and Level 3 warnings; determining the warning level and notifying the user based on the comprehensive risk score includes: If the comprehensive risk score exceeds the first threshold, a level one warning is triggered and a first prompt message is generated to alert the user that the battery has a potential risk and recommend that it be repaired as soon as possible. If the comprehensive risk score exceeds the second threshold, a second warning is triggered to generate a second prompt message, which will remind the user that the risk of battery thermal runaway is high and that the user should stop the vehicle safely immediately. If the comprehensive risk score exceeds the third threshold, a level 3 early warning is triggered and execution measures are generated; wherein, the execution measures include: actively disconnecting the high-voltage circuit, starting the thermal management system to cool down, sending rescue information to the cloud, and issuing an audible and visual alarm; the third threshold is greater than the second threshold, and the second threshold is greater than the first threshold.

[0012] Secondly, embodiments of this application also provide a dynamic thermal runaway early warning system based on multi-source data fusion, including: The data acquisition module is used to collect multi-dimensional operating data of the power battery system in the vehicle and the operating status data of the vehicle. The extraction module is used to extract multi-dimensional feature quantities that characterize the battery health status and potential thermal runaway risk based on the multi-dimensional operating data and the operating status data. The judgment module is used to determine whether the battery state deviates from the normal range based on the pre-built dynamic threshold model and the multi-dimensional feature quantity, and to obtain the judgment result; The analysis module is used to determine the target battery module that triggered the first warning if the judgment result triggers the first warning, and to analyze the thermal runaway precursor signal of the target battery module based on the multi-dimensional operating data, the operating status data and the multi-dimensional feature quantity, so as to obtain the thermal runaway precursor identification result. The evaluation module is used to output a comprehensive risk score and issue a second early warning based on the judgment results of the dynamic threshold model and the thermal runaway precursor identification results.

[0013] In one possible implementation, the collection of multi-dimensional operational data of the vehicle's power battery system and the vehicle's operational status data includes: The system collects raw multi-category data of each individual cell, the battery module composed of multiple cells, and the entire battery pack of the power battery system in the vehicle, as well as the vehicle's raw operating parameter data; the raw multi-category data includes: voltage and current signals, temperature signals, and pressure signals; The first abnormal data and the second abnormal data that do not conform to the preset standard rules are cleaned from the original multi-category data and the original operating parameter data to obtain the preprocessed multi-category data and operating parameter data. The preprocessed multi-category data and the operating parameter data are normalized to obtain the multi-dimensional operating data and the operating status data.

[0014] This application provides a dynamic thermal runaway early warning method and system based on multi-source data fusion, including: collecting multi-dimensional operating data of the power battery system in a vehicle and operating status data of the vehicle; extracting multi-dimensional feature quantities characterizing the battery health status and potential thermal runaway risk based on these data; then, judging whether the battery status deviates from the normal range based on a pre-constructed dynamic threshold model to obtain a judgment result; if the judgment result triggers a first warning, identifying the target battery module that triggered the first warning and analyzing its thermal runaway precursor signals to obtain a thermal runaway precursor identification result; finally, outputting a comprehensive risk score and issuing a second warning based on the judgment result of the dynamic threshold model and the thermal runaway precursor identification result; this application can achieve early warning tens of seconds or even minutes in advance by capturing weak early warning signals, and significantly reduce false alarm and false negative rates and improve warning accuracy by using multi-source data fusion and dynamic threshold adjustment technology. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart of the first dynamic thermal runaway early warning method based on multi-source data fusion provided in this application embodiment is shown; Figure 2 A flowchart of a second dynamic thermal runaway early warning method based on multi-source data fusion provided in an embodiment of this application is shown; Figure 3 A flowchart of a third dynamic thermal runaway early warning method based on multi-source data fusion provided in an embodiment of this application is shown; Figure 4 The overall flowchart of the fourth dynamic thermal runaway early warning method based on multi-source data fusion provided in the embodiments of this application is shown; Figure 5 The diagram shows a schematic of a dynamic thermal runaway early warning system based on multi-source data fusion provided in an embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0018] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0019] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0020] refer to Figure 1 As shown in the flowchart, this application embodiment provides a dynamic thermal runaway early warning method based on multi-source data fusion, the method including: S101. Collect multi-dimensional operating data of the power battery system in the vehicle and the operating status data of the vehicle.

[0021] S102. Based on the multi-dimensional operating data and the operating status data, extract multi-dimensional feature quantities that characterize the battery health status and potential thermal runaway risk, and determine whether the battery status deviates from the normal range based on the pre-built dynamic threshold model and the multi-dimensional feature quantities to obtain the judgment result.

[0022] S103. If the determination result triggers the first warning, the target battery module that triggered the first warning is identified, and the thermal runaway precursor signal of the target battery module is analyzed based on the multi-dimensional operating data, the operating status data and the multi-dimensional feature quantity to obtain the thermal runaway precursor identification result.

[0023] S104. Based on the judgment result of the dynamic threshold model and the thermal runaway precursor identification result, output a comprehensive risk score and issue a second warning.

[0024] The dynamic thermal runaway early warning method based on multi-source data fusion provided in this application can achieve early warning tens of seconds or even minutes in advance by capturing weak precursor signals in the early stage of thermal runaway. With the help of multi-source data fusion and dynamic threshold adjustment technology, it can significantly reduce the false alarm and false alarm rates and improve the accuracy of early warning. It also enhances the adaptability of the early warning model to changes in environment and operating conditions, so that it can achieve reliable adaptive early warning under different battery states and operating conditions.

[0025] The above methods are applied to the power battery system in electric vehicles to provide a detailed explanation of S101-S104; The power battery system, also known as a "battery pack," is the complete set of hardware and software that provides electrical energy storage and management for a vehicle. It includes the following main components: Cells: The most basic energy storage units, like individual "small batteries," with many cells combined together. Modules: Intermediate structures composed of multiple cells (some power battery systems do not have modules). Battery Pack Casing: The physical structure that protects all internal components, providing waterproofing, dustproofing, and impact resistance. Thermal Management System: Responsible for cooling and heating the battery, ensuring it operates within a suitable temperature range. Electrical System: Includes high-voltage connections, contactors (equivalent to large switches), fuses, etc. Battery Management System (BMS).

[0026] S101. Collect multi-dimensional operating data of the power battery system in the vehicle and the operating status data of the vehicle.

[0027] In this embodiment, the BMS collects raw multi-category data of each cell, the battery module composed of multiple cells, and the entire battery pack of the power battery system in the vehicle, as well as the vehicle's raw operating parameter data. The raw multi-category data includes voltage and current signals, temperature signals, and pressure signals. Then, first abnormal data and second abnormal data that do not conform to preset standard rules are cleaned from the raw multi-category data and the raw operating parameter data to obtain preprocessed multi-category data and operating parameter data. Finally, the preprocessed multi-category data and the operating parameter data are normalized to obtain the multi-dimensional operating data and the operating status data.

[0028] Specifically, the raw multi-category data of the aforementioned power battery system includes: electrical signals (such as the voltage and current of each battery cell (or simply cell), and the total system voltage), temperature signals (such as the surface temperature of each cell, the internal temperature of the module, and the ambient temperature of the battery pack), optional pressure signals (such as changes in internal pressure of the module), and vehicle status signals (including vehicle speed, gear position, and air conditioning system status). Then, the collected raw multi-category data and vehicle status signals are cleaned, filtered, and normalized to eliminate the effects of noise interference and dimensional differences.

[0029] S102. Based on the multi-dimensional operating data and the operating status data, extract multi-dimensional feature quantities that characterize the battery health status and potential thermal runaway risk, and determine whether the battery status deviates from the normal range based on the pre-built dynamic threshold model and the multi-dimensional feature quantities to obtain the judgment result.

[0030] In this embodiment of the application, based on the obtained multi-dimensional operating data and the operating status data, multi-dimensional feature quantities that can characterize the battery health status and potential thermal runaway risk are further extracted. These mainly include: features reflecting voltage consistency (such as voltage standard deviation, maximum voltage difference); features characterizing temperature distribution and uniformity (such as maximum / minimum temperature, temperature difference and temperature gradient); features reflecting the rate of temperature rise (such as the rate of change of cell temperature dT / dt); and features revealing the electro-thermal coupling relationship (such as the ratio of voltage change rate to temperature change rate).

[0031] Specifically, based on historical health data, a dynamic threshold model is constructed by learning the data distribution pattern of the power battery system under normal operating conditions through machine learning algorithms. Then, the multi-dimensional operating data and the multi-dimensional feature quantities are input into the dynamic threshold model, and the dynamic threshold model is used to determine whether it deviates from the normal range. If so, the first warning result is triggered.

[0032] A dynamic threshold model is pre-established to determine whether the current battery state deviates from the normal range. This model is based on historical health data and learns the data distribution patterns of the battery under normal operating conditions through machine learning algorithms (such as Isolation Forest and Support Vector Machine). When real-time data points or their features exceed the confidence interval calculated by this dynamic model, a first warning is triggered. This first warning is usually a level one warning (risk alert).

[0033] S103. If the determination result triggers the first warning, the target battery module that triggered the first warning is identified, and the thermal runaway precursor signal of the target battery module is analyzed based on the multi-dimensional operating data, the operating status data and the multi-dimensional feature quantity to obtain the thermal runaway precursor identification result.

[0034] In this embodiment, for a battery cell that triggers a Level 1 warning, the system will activate a specialized time-series analysis algorithm to focus on capturing three types of early warning signals: first, a small, irreversible "micro-voltage drop" occurring during charging or resting; second, a slow, continuous "abnormal temperature rise" occurring in the absence of an external heat source or high-current discharge; and third, "weak oscillations" at specific frequencies in the voltage or temperature signals that may be caused by internal side reactions. This algorithm achieves accurate identification of early signs of thermal runaway by comprehensively analyzing the intensity, duration, and trends of these signals.

[0035] Specifically, the precursor signals of thermal runaway include micro voltage drop, abnormal temperature rise, and weak oscillation: wherein, the micro voltage drop is expressed as: when the target cell is in a static or charging phase, the voltage of the target cell undergoes an irreversible negative drift within a preset amplitude range for a period of at least time threshold T1, and the drift exceeds K times (K≥3) the average voltage change of other cells in the same module. The abnormal temperature rise is defined as follows: under the condition that the external heat source and the charging and discharging current are greater than the preset threshold, the temperature rise rate of the target battery cell exceeds the preset dynamic threshold and the duration of this state is greater than T2. ​​At the same time, the temperature difference between the cell and the reference point in the module continues to increase and exceeds the ΔT threshold. The weak oscillation is defined as follows: after detrending the voltage or temperature signal of the target unit, the signal power or amplitude is detected to be higher than the statistical baseline under normal conditions within a preset frequency band by short-time Fourier transform or wavelet transform, and the power enhancement phenomenon in this frequency band continues to exceed T3.

[0036] S104. Based on the judgment result of the dynamic threshold model and the thermal runaway precursor identification result, output a comprehensive risk score and issue a second warning.

[0037] In this embodiment, the dynamic threshold determination result of S102 and the early warning signal identification result of S103 are weighted and fused, and then input into a pre-trained deep learning risk assessment model (such as LSTM, GRU, etc.). The model integrates historical data, current operating conditions, and the severity of the warning signals to output a comprehensive risk score, and implements graded warnings accordingly: Level 1 warning (risk score exceeds the first threshold) prompts the user that "the battery has potential risks and it is recommended to have it repaired as soon as possible"; Level 2 warning (exceeds the second threshold) warns that "the risk of battery thermal runaway is high, please stop the vehicle safely immediately"; Level 3 warning (exceeds the third threshold) automatically triggers the highest level of safety measures, including actively disconnecting the high-voltage circuit, starting the thermal management system to cool down, sending rescue information to the cloud, and issuing an audible and visual alarm.

[0038] Furthermore, such as Figure 2As shown in the embodiments of this application, the dynamic thermal runaway early warning method based on multi-source data fusion, wherein the extraction of multi-dimensional feature quantities characterizing the battery health state and potential thermal runaway risk based on the multi-dimensional operating data and the operating status data includes: S201. Extract the first feature quantity by separately analyzing the voltage and current signals, temperature signals and pressure signals in the multi-dimensional operation data within a preset time period.

[0039] S202. By performing a comprehensive analysis of the voltage and current signals, temperature signals, and pressure signals in the multi-dimensional operating data within a preset time period, the second feature quantity is extracted.

[0040] S203. Generate the multi-dimensional feature quantity based on the first feature quantity and the second feature quantity; the multi-dimensional feature quantity includes at least: features reflecting voltage consistency, features characterizing temperature distribution and uniformity, features reflecting temperature rise rate, and features revealing electro-thermal coupling relationship.

[0041] Combining steps S201-S202, the BMS collects multi-dimensional operational data and operational status data, including voltage and current signals, temperature signals, and pressure signals. The BMS then analyzes these signals individually within a preset time period to extract first-level features. For example, the BMS calculates the highest temperature, maximum temperature difference, and temperature rise rate of each module in real time. Additionally, the BMS performs an overall analysis of these signals to extract second-level features, such as revealing characteristics of the electro-thermal coupling relationship (e.g., the ratio of voltage change rate to temperature change rate).

[0042] Then, by combining the first feature quantity and the second feature quantity, a multi-dimensional feature quantity can be obtained. Here, the first feature quantity and the second feature quantity can be combined to obtain the multi-dimensional feature quantity, or the first feature quantity and the second feature quantity can be combined and the required feature quantity can be selected as the multi-dimensional feature quantity. Here, the multi-dimensional feature quantity includes at least the following: features reflecting voltage consistency, features characterizing temperature distribution and uniformity, features reflecting temperature rise rate, and features revealing electro-thermal coupling relationship.

[0043] Specifically, the aforementioned multidimensional characteristic quantities include at least one of the following: characteristics reflecting voltage consistency (voltage standard deviation and maximum voltage difference), characteristics characterizing temperature distribution and uniformity (maximum temperature, minimum temperature, temperature difference and temperature gradient), characteristics reflecting the rate of temperature rise (i.e., the rate of change of temperature of a single unit, dT / dt), or characteristics revealing the electro-thermal coupling relationship (i.e., the ratio of the rate of change of voltage to the rate of change of temperature).

[0044] Furthermore, such as Figure 3 As shown in the embodiment of this application, the dynamic thermal runaway early warning method based on multi-source data fusion, wherein the step of outputting a comprehensive risk score and providing a second early warning based on the judgment result of the dynamic threshold model and the thermal runaway precursor identification result includes: S301, Pre-trained risk assessment model; S302. The judgment result of the dynamic threshold model and the thermal runaway precursor identification result are weighted and fused, and then input into the risk assessment model to trigger the second warning. S303. By combining the risk assessment model with historical data, current operating conditions, and the results of thermal runaway precursor identification, a comprehensive risk score is output. S304. Based on the comprehensive risk score, determine the warning level and notify the user.

[0045] Combining S301-S304, the dynamic threshold determination result of S102 and the early precursor signal identification result of S103 are weighted and fused, and then input into a pre-trained deep learning risk assessment model (such as LSTM, GRU, etc.). The model integrates historical data, current operating conditions, and the severity of precursor signals to output a comprehensive risk score, and implements tiered early warnings accordingly. The early warning levels include Level 1, Level 2, and Level 3. Determining the early warning level and alerting the user based on the comprehensive risk score includes: if the comprehensive risk score exceeds a first threshold, a Level 1 early warning is triggered and a first alert message is generated to inform the user that "the battery has a potential risk and it is recommended to have it inspected as soon as possible"; if the comprehensive risk score exceeds a second threshold, a Level 2 early warning is triggered and a second alert message is generated to inform the user that "the battery has a high risk of thermal runaway, please stop the vehicle safely immediately"; if the comprehensive risk score exceeds a third threshold, a Level 3 early warning is triggered and execution measures are generated. These execution measures include: actively disconnecting the high-voltage circuit, activating the thermal management system for cooling, sending rescue information to the cloud, and issuing an audible and visual alarm. The third threshold is greater than the second threshold, and the second threshold is greater than the first threshold.

[0046] The dynamic thermal runaway early warning method and system based on multi-source data fusion provided in this application can achieve early warning tens of seconds or even minutes in advance by capturing weak precursor signals in the early stage of thermal runaway. By using multi-source data fusion and dynamic threshold adjustment technology, it can significantly reduce the false alarm and false alarm rates and improve the accuracy of early warning. It also enhances the adaptability of the early warning model to changes in environment and operating conditions, enabling it to achieve reliable adaptive early warning under different battery states and operating conditions.

[0047] Combination Figure 4The overall flowchart illustrates the dynamic thermal runaway early warning method based on multi-source data fusion in a specific application scenario according to the embodiments of this application: This embodiment applies to a hybrid SUV. Its battery pack consists of four modules, each containing 39 ternary lithium-ion battery cells. The system operates according to the following steps; as follows: Figure 4 As shown: Step 1: Data Acquisition: The voltage and temperature data of all 156 cells are acquired at a frequency of 1Hz through the BMS (Battery Management System), while the total battery current and vehicle speed signal on the vehicle CAN bus are also acquired.

[0048] Step 2, Feature Extraction: BMS calculates the highest temperature, maximum temperature difference, and temperature rise rate of each unit in real time for each module.

[0049] Step 3: Dynamic threshold judgment: The isolated forest model in the system background judged that the temperature rise rate of the 5th module exceeded its dynamic threshold, triggering a level 1 warning.

[0050] Step 4, Early Warning Analysis: The system immediately focused on Module 5 and found that a single cell inside it experienced a small voltage drop (about 5mV) that lasted for about 30 seconds at the end of charging, and its temperature continued to rise slowly without any obvious external cause.

[0051] Step 5: Comprehensive Assessment and Early Warning: The LSTM risk assessment model integrates two key precursors, "exceeding the temperature rise rate limit" and "minor voltage drop," calculating a risk score of 85 points (out of 100). This score exceeds the level 2 warning threshold (80 points), and the system immediately sends a warning to the vehicle display screen stating "High risk of battery thermal runaway; please stop the vehicle immediately for safety." Simultaneously, the data is sent to the user's mobile app and the backend monitoring center via the vehicle network to alert the user and maintenance personnel.

[0052] In this embodiment, the focus is on: real-time acquisition of multi-source data from the power battery system, including the voltage, temperature, and current of at least one battery cell; preprocessing the multi-source data and extracting characteristic quantities representing the battery state; determining whether a first-level warning risk exists based on the characteristic quantities using a dynamic threshold model; when a first-level warning risk is determined to exist, analyzing whether the target battery cell exhibits at least one early warning signal, such as a slight voltage drop, abnormal temperature rise, or temperature oscillation; and conducting a comprehensive assessment of thermal runaway risk based on the severity of the early warning signal and the results of the dynamic threshold model, and outputting multi-level warning results. The aforementioned multi-source data also includes the internal pressure of the battery module, the ambient temperature of the battery pack, and vehicle status signals. The extracted features characterizing the battery status include calculating at least one of the following: voltage standard deviation, temperature difference, temperature rise rate, or voltage-temperature coupling characteristics. The aforementioned dynamic threshold model is trained based on historical health data using an isolated forest or support vector machine algorithm. The aforementioned comprehensive assessment of thermal runaway risk involves inputting the early warning signals and the results of the dynamic threshold model into a long short-term memory network or a gated recurrent unit model for calculation to obtain a risk score. The aforementioned multi-level warning results include a level one risk alert, a level two severe warning, and a level three emergency alarm, with the level three warning triggering active vehicle safety protection measures.

[0053] The dynamic thermal runaway early warning method and system based on multi-source data fusion provided in this application can achieve early warning tens of seconds or even minutes in advance by capturing weak precursor signals in the early stage of thermal runaway. By using multi-source data fusion and dynamic threshold adjustment technology, it can significantly reduce the false alarm and false alarm rates and improve the accuracy of early warning. It also enhances the adaptability of the early warning model to changes in environment and operating conditions, enabling it to achieve reliable adaptive early warning under different battery states and operating conditions.

[0054] Based on the same inventive concept, this application also provides a dynamic thermal runaway early warning device based on multi-source data fusion, which corresponds to the dynamic thermal runaway early warning method based on multi-source data fusion. Since the principle of the device in this application is similar to the dynamic thermal runaway early warning method based on multi-source data fusion described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0055] Reference Figure 5 As shown in the figure, a dynamic thermal runaway early warning system based on multi-source data fusion provided in this application includes: The acquisition module 501 is used to acquire multi-dimensional operating data of the power battery system in the vehicle and the operating status data of the vehicle. Extraction module 502 is used to extract multi-dimensional feature quantities characterizing battery health status and potential thermal runaway risk based on the multi-dimensional operating data and the operating status data; The judgment module 503 is used to judge whether the battery state deviates from the normal range based on the pre-built dynamic threshold model and the multi-dimensional feature quantity, and obtain the judgment result. Analysis module 504 is used to determine the target battery module that triggered the first warning if the determination result triggers the first warning, and analyze the thermal runaway precursor signal of the target battery module based on the multi-dimensional operating data, the operating status data and the multi-dimensional feature quantity to obtain the thermal runaway precursor identification result. The evaluation module 505 is used to output a comprehensive risk score and provide a second early warning based on the judgment result of the dynamic threshold model and the thermal runaway precursor identification result.

[0056] In one possible implementation, the collection of multi-dimensional operational data of the vehicle's power battery system and the vehicle's operational status data includes: The system collects raw multi-category data of each individual cell, the battery module composed of multiple cells, and the entire battery pack of the power battery system in the vehicle, as well as the vehicle's raw operating parameter data; the raw multi-category data includes: voltage and current signals, temperature signals, and pressure signals; The first abnormal data and the second abnormal data that do not conform to the preset standard rules are cleaned from the original multi-category data and the original operating parameter data to obtain the preprocessed multi-category data and operating parameter data. The preprocessed multi-category data and the operating parameter data are normalized to obtain the multi-dimensional operating data and the operating status data.

[0057] In one possible implementation, the step of extracting multi-dimensional features characterizing battery health status and potential thermal runaway risk based on the multi-dimensional operational data and the operational status data includes: The first feature quantity is extracted by separately analyzing the voltage and current signals, temperature signals, and pressure signals in the multi-dimensional operation data within a preset time period. By comprehensively analyzing the voltage and current signals, temperature signals, and pressure signals in the multi-dimensional operational data within a preset time period, the second feature quantity is extracted. The multi-dimensional feature quantity is generated based on the first feature quantity and the second feature quantity; the multi-dimensional feature quantity includes at least: features reflecting voltage consistency, features characterizing temperature distribution and uniformity, features reflecting temperature rise rate, and features revealing electro-thermal coupling relationship.

[0058] In one possible implementation, the multidimensional characteristic quantity includes at least one of the following: voltage standard deviation, maximum voltage difference, highest temperature, lowest temperature, temperature difference, temperature gradient, temperature rise rate, or voltage-temperature coupling characteristic; the temperature rise rate includes the temperature change rate of a single cell in the power battery system.

[0059] In one possible implementation, the step of determining whether the battery state deviates from the normal range based on a pre-built dynamic threshold model and the multi-dimensional feature quantities, and obtaining a determination result, includes: Based on historical health data, a dynamic threshold model is constructed by learning the data distribution pattern of the power battery system under normal operating conditions through machine learning algorithms. The multi-dimensional operating data and the multi-dimensional feature values ​​are input into the dynamic threshold model, and the dynamic threshold model is used to determine whether it deviates from the normal range. If so, the first warning result will be triggered.

[0060] In one possible implementation, the precursory signals of thermal runaway include a small voltage drop, an abnormal temperature rise, and weak oscillations: wherein, The micro voltage drop is defined as follows: when the target cell is in a static or charging phase, the voltage of the target cell undergoes an irreversible negative drift within a preset amplitude range for a period of at least time threshold T1, and the drift exceeds K times (K≥3) the average voltage change of other cells in the same module. The abnormal temperature rise is defined as follows: under the condition that the external heat source and the charging and discharging current are greater than the preset threshold, the temperature rise rate of the target battery cell exceeds the preset dynamic threshold and the duration of this state is greater than T2. ​​At the same time, the temperature difference between the cell and the reference point in the module continues to increase and exceeds the ΔT threshold. The weak oscillation is defined as follows: after detrending the voltage or temperature signal of the target unit, the signal power or amplitude is detected to be higher than the statistical baseline under normal conditions within a preset frequency band by short-time Fourier transform or wavelet transform, and the power enhancement phenomenon in this frequency band continues to exceed T3.

[0061] In one possible implementation, the step of outputting a comprehensive risk score and issuing a second early warning based on the determination result of the dynamic threshold model and the thermal runaway precursor identification result includes: Pre-trained risk assessment models; The judgment result of the dynamic threshold model is weighted and fused with the thermal runaway precursor identification result, and then input into the risk assessment model to trigger a second warning. The risk assessment model combines historical data, current operating conditions, and the results of thermal runaway precursor identification to output a comprehensive risk score. Based on the comprehensive risk score, a warning level is determined and the user is notified.

[0062] This application provides a dynamic thermal runaway early warning system based on multi-source data fusion, comprising: collecting multi-dimensional operating data of the power battery system in a vehicle and operating status data of the vehicle; extracting multi-dimensional feature quantities characterizing the battery health status and potential thermal runaway risk based on the multi-dimensional operating data and operating status data, and judging whether the multi-dimensional operating data and multi-dimensional features deviate from the normal range based on a pre-constructed dynamic threshold model to obtain a judgment result; if the judgment result triggers a first warning, identifying the target battery module that triggered the first warning, and analyzing the thermal runaway precursor signal of the target battery module based on the multi-dimensional operating data, operating status data and multi-dimensional features to obtain a thermal runaway precursor identification result; and outputting a comprehensive risk score and issuing a second warning based on the judgment result of the dynamic threshold model and the thermal runaway precursor identification result. This application can achieve early warning tens of seconds or even minutes in advance by capturing weak precursor signals in the early stage of thermal runaway. By using multi-source data fusion and dynamic threshold adjustment technology, it can significantly reduce the false alarm and false alarm rates and improve the accuracy of the warning. It also enhances the adaptability of the warning model to changes in environment and operating conditions, enabling it to achieve reliable adaptive warning under different battery states and operating conditions.

[0063] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some communication interfaces; the indirect coupling or communication connection of devices or modules can be electrical, mechanical, or other forms.

[0064] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0065] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0066] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0067] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A dynamic thermal runaway early warning method based on multi-source data fusion, characterized in that, The method includes: Collect multi-dimensional operational data of the power battery system in the vehicle and operational status data of the vehicle; Based on the multi-dimensional operating data and the operating status data, multi-dimensional feature quantities characterizing the battery health status and potential thermal runaway risk are extracted. Based on the pre-built dynamic threshold model and the multi-dimensional feature quantities, it is determined whether the battery status deviates from the normal range, and the judgment result is obtained. If the determination result triggers the first warning, the target battery module that triggered the first warning is identified, and the thermal runaway precursor signal of the target battery module is analyzed based on the multi-dimensional operating data, the operating status data and the multi-dimensional feature quantity to obtain the thermal runaway precursor identification result; Based on the judgment results of the dynamic threshold model and the identification results of the thermal runaway precursors, a comprehensive risk score is output and a second warning is issued.

2. The dynamic thermal runaway early warning method based on multi-source data fusion according to claim 1, characterized in that, The multi-dimensional operational data of the power battery system in the vehicle and the operational status data of the vehicle are collected, including: The system collects raw multi-category data of each individual cell, the battery module composed of multiple cells, and the entire battery pack of the power battery system in the vehicle, as well as the vehicle's raw operating parameter data; the raw multi-category data includes: voltage and current signals, temperature signals, and pressure signals; The first abnormal data and the second abnormal data that do not conform to the preset standard rules are cleaned from the original multi-category data and the original operating parameter data to obtain the preprocessed multi-category data and operating parameter data. The preprocessed multi-category data and the operating parameter data are normalized to obtain the multi-dimensional operating data and the operating status data.

3. The dynamic thermal runaway early warning method based on multi-source data fusion according to claim 2, characterized in that, The extraction of multi-dimensional features characterizing battery health status and potential thermal runaway risk based on the multi-dimensional operational data and operational status data includes: The first feature quantity is extracted by separately analyzing the voltage and current signals, temperature signals, and pressure signals in the multi-dimensional operation data within a preset time period. By comprehensively analyzing the voltage and current signals, temperature signals, and pressure signals in the multi-dimensional operational data within a preset time period, the second feature quantity is extracted. The multi-dimensional feature quantity is generated based on the first feature quantity and the second feature quantity; the multi-dimensional feature quantity includes at least: features reflecting voltage consistency, features characterizing temperature distribution and uniformity, features reflecting temperature rise rate, and features revealing electro-thermal coupling relationship.

4. The dynamic thermal runaway early warning method based on multi-source data fusion according to claim 3, characterized in that, The multidimensional characteristic quantities include at least one of the following: voltage standard deviation, maximum voltage difference, highest temperature, lowest temperature, temperature difference, temperature gradient, temperature rise rate, or voltage-temperature coupling characteristics; the temperature rise rate includes the temperature change rate of individual cells in the power battery system.

5. The dynamic thermal runaway early warning method based on multi-source data fusion according to claim 1, characterized in that, The determination of whether the battery state deviates from the normal range based on the pre-built dynamic threshold model and the multi-dimensional feature quantities yields the following results: Based on historical health data, a dynamic threshold model is constructed by learning the data distribution pattern of the power battery system under normal operating conditions through machine learning algorithms. The multi-dimensional operating data and the multi-dimensional feature values ​​are input into the dynamic threshold model, and the dynamic threshold model is used to determine whether it deviates from the normal range. If so, the first warning result will be triggered.

6. The dynamic thermal runaway early warning method based on multi-source data fusion according to claim 1, characterized in that, The precursory signals of thermal runaway include micro-voltage drops, abnormal temperature rises, and weak oscillations: among which... The micro voltage drop is defined as follows: when the target cell is in a static or charging phase, the voltage of the target cell undergoes an irreversible negative drift within a preset amplitude range for a period of at least time threshold T1, and the drift exceeds K times (K≥3) the average voltage change of other cells in the same module. The abnormal temperature rise is defined as follows: under the condition that the external heat source and the charging and discharging current are greater than the preset threshold, the temperature rise rate of the target battery cell exceeds the preset dynamic threshold and the duration of this state is greater than T2. ​​At the same time, the temperature difference between the cell and the reference point in the module continues to increase and exceeds the ΔT threshold. The weak oscillation is defined as follows: after detrending the voltage or temperature signal of the target unit, the signal power or amplitude is detected to be higher than the statistical baseline under normal conditions within a preset frequency band by short-time Fourier transform or wavelet transform, and the power enhancement phenomenon in this frequency band continues to exceed T3.

7. The dynamic thermal runaway early warning method based on multi-source data fusion according to claim 1, characterized in that, The step of outputting a comprehensive risk score and issuing a second early warning based on the judgment result of the dynamic threshold model and the thermal runaway precursor identification result includes: Pre-trained risk assessment models; The judgment result of the dynamic threshold model is weighted and fused with the thermal runaway precursor identification result, and then input into the risk assessment model to trigger a second warning. The risk assessment model combines historical data, current operating conditions, and the results of thermal runaway precursor identification to output a comprehensive risk score. Based on the comprehensive risk score, a warning level is determined and the user is notified.

8. The dynamic thermal runaway early warning method based on multi-source data fusion according to claim 7, characterized in that, The warning levels include Level 1, Level 2, and Level 3. The step of determining the warning level and notifying the user based on the comprehensive risk score includes: If the comprehensive risk score exceeds the first threshold, a level one warning is triggered and a first prompt message is generated to alert the user that the battery has a potential risk and recommend that it be repaired as soon as possible. If the comprehensive risk score exceeds the second threshold, a second warning is triggered to generate a second prompt message, which will remind the user that the risk of battery thermal runaway is high and that the user should stop the vehicle safely immediately. If the comprehensive risk score exceeds the third threshold, a level 3 early warning is triggered and execution measures are generated; wherein, the execution measures include: actively disconnecting the high-voltage circuit, starting the thermal management system to cool down, sending rescue information to the cloud, and issuing an audible and visual alarm; the third threshold is greater than the second threshold, and the second threshold is greater than the first threshold.

9. A dynamic thermal runaway early warning system based on multi-source data fusion, characterized in that, include; The data acquisition module is used to collect multi-dimensional operating data of the power battery system in the vehicle and the operating status data of the vehicle. The extraction module is used to extract multi-dimensional feature quantities that characterize the battery health status and potential thermal runaway risk based on the multi-dimensional operating data and the operating status data. The judgment module is used to determine whether the battery state deviates from the normal range based on the pre-built dynamic threshold model and the multi-dimensional feature quantity, and to obtain the judgment result; The analysis module is used to determine the target battery module that triggered the first warning if the judgment result triggers the first warning, and to analyze the thermal runaway precursor signal of the target battery module based on the multi-dimensional operating data, the operating status data and the multi-dimensional feature quantity, so as to obtain the thermal runaway precursor identification result. The evaluation module is used to output a comprehensive risk score and issue a second early warning based on the judgment results of the dynamic threshold model and the thermal runaway precursor identification results.

10. The dynamic thermal runaway early warning system based on multi-source data fusion according to claim 9, characterized in that, The multi-dimensional operational data of the power battery system in the vehicle and the operational status data of the vehicle are collected, including: The system collects raw multi-category data of each individual cell, the battery module composed of multiple cells, and the entire battery pack of the power battery system in the vehicle, as well as the vehicle's raw operating parameter data; the raw multi-category data includes: voltage and current signals, temperature signals, and pressure signals; The first abnormal data and the second abnormal data that do not conform to the preset standard rules are cleaned from the original multi-category data and the original operating parameter data to obtain the preprocessed multi-category data and operating parameter data. The preprocessed multi-category data and the operating parameter data are normalized to obtain the multi-dimensional operating data and the operating status data.

Images