Special vehicle battery thermal management safety early warning system based on multi-source information fusion

Abstract

The invention discloses a special vehicle battery thermal management safety early warning system based on multi-source information fusion, and the system comprises a sensing collection module which is composed of a plurality of groups of sensing units, collects the temperature, strain and gas concentration data of a battery cell according to a preset interval, and eliminates invalid data after verification of a receiving end to generate valid data; the time-space fusion module builds a carrier transmission synchronization model to guarantee transmission, records sensing unit coordinates and cell numbers through a time-space association mapping table, classifies data and performs time-space alignment to form a fusion data set; the risk judgment module outputs a thermal runaway risk level through data comparison based on the thermal runaway evolution path library; and the grading execution module divides instruction grades according to risk levels, generates instructions containing key information corresponding to various thermal management operation combinations, and sends the instructions to the corresponding thermal management parts. The early warning accuracy and response timeliness of the thermal runaway of the battery of the special vehicle are effectively improved, and the operation safety of the battery is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery management, in particular to a special vehicle battery thermal management safety warning system based on multi-source information fusion. BACKGROUND

[0002] The working condition of a special vehicle is complex and changeable, and the battery as the core power source, its thermal runaway risk is directly related to the driving safety, and the battery thermal management safety warning system is the key to guarantee the reliable operation of the special vehicle. Such system needs to accurately monitor the battery state and timely warn the risk, and has an irreplaceable role in harsh application scenarios such as military and engineering of the special vehicle.

[0003] However, the traditional special vehicle battery thermal management warning system has obvious limitations. The data collection dimension is single, and only temperature parameters are monitored, and key failure precursor information such as strain and gas concentration is ignored, resulting in one-sided risk identification; the data processing lacks a space-time correlation mechanism, and the data synchronization of different sensing units is poor, which is easy to misjudge or miss; the risk judgment relies on fixed threshold, and is not combined with the thermal runaway evolution law, so the graded response is not enough; there is no dynamic feedback optimization mechanism, and the monitoring and control strategy cannot be adjusted according to the change of the battery state, which is difficult to adapt to the dynamic demand of the complex working condition of the special vehicle.

[0004] Therefore, it is necessary to design a special vehicle battery thermal management safety warning system based on multi-source information fusion to solve the problems of one-sided data collection, poor space-time synchronization and low risk judgment accuracy of the traditional system. SUMMARY

[0005] In view of this, the present application provides a special vehicle battery thermal management safety warning system based on multi-source information fusion, which aims to solve the problems of one-sided data collection, poor space-time synchronization and low risk judgment accuracy of the traditional system.

[0006] In one aspect, the present application provides a special vehicle battery thermal management safety warning system based on multi-source information fusion, comprising: A sensing and collecting module composed of a plurality of groups of sensing units, the sensing units are used to collect temperature, strain and gas concentration data in the battery cell at a preset interval, and send the collected data to a receiving end, the receiving end performs preliminary verification on the data and eliminates missing data and repeated data in the data, and generates valid data; A space-time fusion module for establishing a carrier transmission synchronization model, allocating independent transmission channels, setting transmission frequency and data frame format, generating millisecond-level time stamps and embedding a retransmission mechanism; the space-time fusion module is also used to build a space-time correlation mapping table, which records the three-dimensional coordinates and cell numbers of each group of sensing units, classifies valid data according to the cell numbers and aligns and correlates through the space-time field, forming a space-time fusion data set; a risk determination module configured to establish a thermal runaway evolution path library, wherein a plurality of parameter variation sequences, failure modes and staged feature data intervals of different battery cell thermal runaway tests are stored in the thermal runaway evolution path library, and a space-time fusion data set is matched with the feature data intervals by a data comparison algorithm to output a thermal runaway risk level; a hierarchical execution module configured to divide instructions into levels according to the thermal runaway risk level, each of the instruction levels corresponding to a combination of thermal management operations of cooling medium circulation start, heat dissipation channel switching, battery cell isolation device triggering and thermal management component power adjustment, and to generate instructions containing operation objects, start time and duration and send them to corresponding thermal management components.

[0007] Further, the system further comprises a closed-loop optimization module configured to collect real-time data of temperature, strain and gas concentration in the battery cell after the thermal management operation, and operating parameters of the thermal management components such as working state, power and running time, to arrange the data in a data frame format preset by the space-time fusion module and mark the operation execution time stamp; The arranged data is fed back to the space-time fusion module to adjust the transmission channel frequency and data retransmission time threshold in the carrier transmission synchronization model, and to update the correlation weight and data alignment accuracy parameters of adjacent sensing units in the space-time correlation mapping table for system dynamic optimization.

[0008] Further, the arrangement of the sensing units in the sensing and collecting module and the adjustment of the data collection interval include: The battery cell type division criteria are preset, and the battery cell types include ternary lithium battery, iron phosphate lithium battery and cobalt lithium battery; The first, second and third collection intervals are preset, and the first collection interval is smaller than the second collection interval, which is smaller than the third collection interval; The collection interval initial value is determined according to the actual type of the battery cell in the special vehicle battery pack: When the battery cell type is cobalt lithium battery, the first collection interval is selected as the initial collection interval; When the battery cell type is ternary lithium battery, the second collection interval is selected as the initial collection interval; When the battery cell type is iron phosphate lithium battery, the third collection interval is selected as the initial collection interval; The deviation rate of the temperature, strain and gas concentration data in the real-time monitoring effective data from the preset reference value is monitored, and the deviation rate first threshold and the deviation rate second threshold are set, and the deviation rate second threshold is greater than the deviation rate first threshold: When the deviation rate is less than or equal to the deviation rate first threshold, the current collection interval is maintained unchanged; When the deviation rate is greater than the deviation rate first threshold and less than or equal to the deviation rate second threshold, the collection interval is shortened by one half; When the deviation rate is greater than the second threshold of the deviation rate, the acquisition interval is shortened to the first acquisition interval, and the first acquisition interval is greater than or equal to the minimum preset number of seconds.

[0009] Furthermore, the determination of the three-dimensional coordinates and the initial setting of the association weights of the spatiotemporal correlation mapping table include: The installation coordinates of each sensing unit are obtained based on the 3D modeling data of the battery pack. The coordinate axes include the X-axis, Y-axis, and Z-axis, where the X-axis corresponds to the length direction of the battery pack, the Y-axis corresponds to the width direction of the battery pack, and the Z-axis corresponds to the height direction of the battery pack. A first association weight, a second association weight, a third association weight, a first spatial distance, and a second spatial distance are pre-defined, wherein the first association weight is greater than the second association weight, which is greater than the third association weight, and the first spatial distance is less than the second spatial distance; Initial association weights are set based on the spatial distance between the sensing units: When the spatial distance between two sensing units is less than or equal to the first spatial distance, the spatiotemporal fusion module configures the first association weight of the two sensing units. When the spatial distance between two sensing units is greater than the first spatial distance and less than or equal to the second spatial distance, the spatiotemporal fusion module configures a second association weight for the two sensing units. When the spatial distance between two sensing units is greater than or equal to the second spatial distance, the spatiotemporal fusion module configures a third association weight for the two sensing units.

[0010] Furthermore, the data comparison algorithm of the risk assessment module specifically includes: Based on the phased feature data intervals in the thermal runaway evolution path library, the upper and lower limits of the intervals for temperature, strain, and gas concentration are extracted, and the similarity between each parameter in the spatiotemporal fusion dataset and the corresponding interval is calculated. A first similarity threshold, a second similarity threshold, a third similarity threshold, and thresholds for the rate of change of temperature, strain, and gas concentration in the spatiotemporal fusion dataset are preset, and the first similarity threshold is less than the second similarity threshold and less than the third similarity threshold. The similarity is compared with each threshold: when the similarity is greater than or equal to the third similarity threshold, the thermal runaway risk level corresponding to the matching feature data interval of the spatiotemporal fusion dataset is directly determined; When the similarity is greater than or equal to the second similarity threshold and less than the third similarity threshold, the rate of change of temperature and gas concentration in the spatiotemporal fusion dataset is extracted. If the rate of change is greater than or equal to the preset rate of change threshold, the risk level corresponding to that interval is increased by one level; if the rate of change is less than the preset rate of change threshold, the risk level corresponding to that interval is maintained. When the similarity is greater than or equal to the first similarity threshold and less than the second similarity threshold, the risk level is raised by one level if the adjacent unit data matches a feature range greater than the risk level, based on the associated data of adjacent sensing units in the spatiotemporal correlation mapping table; otherwise, the original risk level is maintained. When the similarity is less than the first similarity threshold, it is determined to be a no-match risk level, a low-risk initial level is output, and it is marked as needing continuous monitoring.

[0011] Furthermore, the updating of the thermal runaway evolution path library includes: Pre-set path library update conditions: the amount of new thermal runaway test data is greater than or equal to 50 sets, a new failure mode appears in the same type of cell, and the existing feature data interval matching accuracy is less than 85%. When any update condition is met, collect the multi-parameter change sequence, failure mode and corresponding stage data of the newly added thermal runaway test; Set a data validity threshold to remove abnormal data and retain valid new data; According to the rules for classifying thermal runaway evolution stages, the newly added data are classified into the corresponding stages, and the upper and lower limits of the characteristic data interval are updated. The updated path library is tested using a validation dataset. If the matching accuracy is greater than or equal to 90%, the updated results are saved; if the matching accuracy is less than 90%, the system reverts to the previous version and re-filters valid data.

[0012] Furthermore, the instruction execution feedback verification of the hierarchical execution module includes: Pre-set the following indicators for judging the execution effect of the command: temperature drop rate, strain recovery rate, and gas concentration decay rate; Set a first judgment threshold and a second judgment threshold for each indicator, and the first judgment threshold shall be less than the second judgment threshold; After a preset time has elapsed since the instruction was executed, real-time data of the corresponding battery cell is collected, and the actual values ​​of each judgment indicator are calculated. The actual values ​​are compared with the thresholds: when the actual values ​​of all indicators are greater than or equal to the second judgment threshold, the judgment instruction is valid and the current instruction is maintained until the risk level decreases. When the actual value of at least one indicator is greater than or equal to the first judgment threshold and less than the second judgment threshold, the judgment instruction execution part is valid, and the power or duration of the thermal management operation is adjusted. When the actual values ​​of all indicators are less than the first judgment threshold, the judgment instruction is invalid, the current instruction level is upgraded by one level, and the instruction is regenerated and sent.

[0013] Furthermore, the dynamic adjustment of the deviation rate threshold includes: Collect valid data and corresponding deviation rates from the past 6 months, as well as actual thermal runaway risk records, and establish a deviation rate-risk correlation database. Calculate the risk identification accuracy under different deviation rate thresholds and set the target accuracy. If the recognition accuracy corresponding to the current deviation rate first threshold is less than the target accuracy, and the proportion of missed detection risk is greater than or equal to 10%, then the first threshold will be lowered by 10%. If the proportion of misjudgment risk corresponding to the current deviation rate second threshold is greater than or equal to 15%, then the second threshold will be increased by 15%. After adjustment, the recognition accuracy is recalculated. If the accuracy is greater than or equal to the target accuracy, the adjusted threshold is saved. If the accuracy is less than the target accuracy, the adjustment is repeated until the accuracy is greater than or equal to the target accuracy and the adjusted threshold is within the preset threshold range.

[0014] Furthermore, the dynamic updating of the association weights in the spatiotemporal association mapping table includes: Collect fault propagation records of adjacent sensor units in historical fault cases, and statistically analyze the matching accuracy of association weights under different spatial distances; Set a weighted matching accuracy threshold greater than or equal to 88%. When the weighted matching accuracy corresponding to a certain spatial distance is less than the weighted matching accuracy threshold: If the spatial distance is less than or equal to the first spatial distance, the first association weight will be increased by the first weight coefficient. If the spatial distance is greater than the first spatial distance and less than or equal to the second spatial distance, the second association weight will be increased by the second weight coefficient. When the spatial distance is greater than the second spatial distance, the third association weight will be increased by the third weight coefficient. After the update, the matching accuracy is recalculated. If the weighted matching accuracy is greater than or equal to 88%, the adjusted weights are saved. If the weighted matching accuracy is still less than 88%, the adjustment is repeated, and the total weights remain unchanged.

[0015] Furthermore, the modification of the thermal runaway risk level includes: After obtaining the thermal management operation data fed back by the closed-loop optimization module, and combining it with the spatiotemporal fusion dataset, the matching similarity with the thermal runaway evolution path library is recalculated. Set a risk level correction threshold. When the difference between the recalculated similarity and the initial similarity is greater than or equal to the risk level correction threshold: If the recalculated similarity increases, the original risk level will be raised by one level; if the recalculated similarity decreases, the original risk level will be lowered by one level. If the difference between the recalculated similarity and the initial similarity is less than the risk level correction threshold, the original risk level is maintained. The revised risk level is synchronously fed back to the hierarchical execution module to adjust the level of subsequent instructions.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The sensor acquisition module synchronously collects three key data types: cell temperature, strain, and gas concentration, covering multiple precursor signals such as temperature anomalies, structural deformation, and gas release during thermal runaway. Combined with the receiver data verification and invalid data rejection mechanism, it not only makes up for the limitations of single temperature monitoring, but also ensures the integrity and reliability of the collected data, providing comprehensive data support for risk assessment.

[0017] 2. The spatiotemporal fusion module ensures the stability and timeliness of data transmission through independent transmission channel allocation, millisecond-level timestamp embedding, and retransmission mechanism. At the same time, it relies on the spatiotemporal association mapping table to achieve precise binding of the three-dimensional coordinates of the sensing unit, the cell number, and the data. By aligning the spatiotemporal fields, it completes the fusion of multi-source data, effectively solving the problems of misjudgment and omission caused by data dispersion and insufficient synchronization, and improving the spatiotemporal consistency of the dataset.

[0018] 3. The risk assessment module constructs a thermal runaway evolution path library containing different battery cells and different failure modes. Combined with phased feature data intervals and multi-parameter comparison algorithms, it upgrades from single threshold judgment to "multi-parameter-phase-full-scenario" matching. It can accurately identify the thermal runaway development stage, output the risk level that fits the reality, greatly reduce the probability of misjudgment and missed judgment, and improve the accuracy of early warning.

[0019] 4. The graded execution module matches differentiated operation combinations such as cooling medium circulation, heat dissipation channel switching, and cell isolation according to risk level, and generates precise instructions containing the operation object, start time, and duration. This avoids the "one-size-fits-all" extensive control and can dynamically adjust the thermal management strategy according to the complex working conditions of special vehicles, so as to achieve rapid and accurate risk handling and comprehensively ensure battery operation safety and vehicle reliable operation. Attached Figure Description

[0020] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A first functional block diagram of a special vehicle battery thermal management safety early warning system based on multi-source information fusion provided in an embodiment of the present invention; Figure 2 The second functional block diagram of the special vehicle battery thermal management safety early warning system based on multi-source information fusion provided in the embodiments of the present invention; Figure 3 This is a flowchart illustrating the workflow of a special vehicle battery thermal management safety early warning system based on multi-source information fusion, as provided in an embodiment of the present invention. Detailed Implementation

[0021] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey its scope to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features described herein can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] Reference Figure 1 In some embodiments of this application, a special vehicle battery thermal management safety early warning system based on multi-source information fusion includes: a sensor acquisition module, a spatiotemporal fusion module, a risk determination module, and a graded execution module.

[0023] Specifically, the sensing and acquisition module consists of several sets of sensing units. The sensing units are used to collect data on temperature, strain and gas concentration inside the battery cell at preset intervals, and send the collected data to the receiving end. The receiving end performs preliminary verification on the data, removes missing and duplicate data, and generates valid data. The spatiotemporal fusion module is used to establish a carrier transmission synchronization model, allocate independent transmission channels, set transmission frequencies and data frame formats, generate millisecond-level timestamps and embed retransmission mechanisms; the spatiotemporal fusion module is also used to construct a spatiotemporal association mapping table, which records the three-dimensional coordinates and cell number of each group of sensing units, classifies valid data according to cell number and aligns and associates them through spatiotemporal fields to form a spatiotemporal fusion dataset. The risk assessment module is used to establish a thermal runaway evolution path library. The thermal runaway evolution path library stores multi-parameter change sequences, failure modes and stage characteristic data intervals of thermal runaway tests of different cells. The module matches the spatiotemporal fusion dataset with the characteristic data intervals through a data comparison algorithm and outputs the thermal runaway risk level. The hierarchical execution module is used to classify instruction levels according to the thermal runaway risk level. Each instruction level corresponds to a thermal management operation combination such as cooling medium circulation start-up, heat dissipation channel switching, cell isolation device triggering, and thermal management component power adjustment. It generates instructions containing the operation object, start time, and duration and sends them to the corresponding thermal management component.

[0024] The above embodiments synchronously collect three key data types—cell temperature, strain, and gas concentration—through a sensing acquisition module. This covers multi-dimensional precursor signals during thermal runaway, such as temperature anomalies, structural deformation, and gas release. Combined with receiver data verification and invalid data removal mechanisms, this approach not only overcomes the limitations of single-temperature monitoring but also ensures the integrity and reliability of the collected data, providing comprehensive data support for risk assessment. The spatiotemporal fusion module ensures the stability and timeliness of data transmission through independent transmission channel allocation, millisecond-level timestamp embedding, and retransmission mechanisms. Simultaneously, it relies on a spatiotemporal correlation mapping table to accurately bind the three-dimensional coordinates of the sensing unit, cell number, and data. By aligning spatiotemporal fields, it completes multi-source data fusion, effectively solving the problems of misjudgment and missed judgment caused by data dispersion and insufficient synchronization, and improving the spatiotemporal consistency of the dataset. The risk assessment module constructs a thermal runaway evolution path library containing different cells and failure modes. Combined with phased feature data intervals and multi-parameter comparison algorithms, it upgrades from single-threshold judgment to "multi-parameter-phase-full-scenario" matching. This enables accurate identification of thermal runaway development stages, outputting realistic risk levels, significantly reducing the probability of false positives and false negatives, and improving early warning accuracy. The graded execution module matches differentiated operation combinations such as cooling medium circulation, heat dissipation channel switching, and cell isolation according to risk levels, generating precise instructions containing the operation object, start time, and duration. This avoids a "one-size-fits-all" approach to control and dynamically adjusts thermal management strategies based on the complex operating conditions of special vehicles, achieving rapid and accurate risk handling and comprehensively ensuring battery operation safety and vehicle reliability.

[0025] Reference Figure 2 In some embodiments of this application, it further includes: a closed-loop optimization module, used to collect real-time data of temperature, strain, and gas concentration in the cell after thermal management operation, as well as operating parameters such as working status, power, and running time of the thermal management component; organize the data according to the data frame format preset by the spatiotemporal fusion module, and mark the operation execution timestamp; The processed data is fed back to the spatiotemporal fusion module, which adjusts the transmission channel frequency and data retransmission time threshold in the carrier transmission synchronization model. At the same time, it updates the association weights of adjacent sensing units and the data alignment accuracy parameters in the spatiotemporal association mapping table to perform dynamic system optimization.

[0026] Specifically, after the thermal management operation is executed, the closed-loop optimization module continuously collects real-time data on the temperature, strain, and gas concentration inside the battery cell at short intervals of 5-10 seconds. Simultaneously, it collects operating parameters such as the working status (running / stopping), real-time power, cumulative runtime, and fault codes of thermal management components like the cooling pump, heat dissipation valve, and isolation device. During the data collection process, a data validity verification mechanism is embedded to remove abnormal parameters that exceed the normal operating range of the battery cell, ensuring the validity of the feedback data. In the data processing stage, the module strictly follows the preset format of the spatiotemporal fusion module, classifying data according to a hierarchical structure of "cell number - component number - operation timestamp - parameter type - value." The operation timestamp is accurate to the millisecond level and consistent with the time base of the spatiotemporal fusion dataset. In the parameter optimization stage, the transmission channel frequency is dynamically adapted according to the packet loss rate (increased by 5%-10% when the packet loss rate is >3%). When the transmission delay is greater than 20ms, a low-interference channel is reallocated. The data retransmission time threshold is adjusted according to the reception success rate (the reception success rate is reduced from the default 50ms to 30ms when the reception success rate is less than 95%). The association weight of adjacent sensor units in the spatiotemporal correlation mapping table is corrected by combining historical fault propagation data (the deviation rate is increased by 0.05~0.1 when the synchronization change is greater than 10 times). The data alignment accuracy parameter is optimized according to the timestamp synchronization error (the time compensation coefficient is adjusted when the synchronization error is greater than 5ms to ensure the spatiotemporal matching accuracy is ±2ms). During the optimization process, an effect evaluation threshold is set. When the transmission stability (packet loss rate <1%, delay <10ms) and data alignment accuracy (≥98%) both meet the standards, the current parameters are saved. If they do not meet the standards, the "collection-organization-adjustment" process is repeated until the parameters are adapted to the current battery operating status and working conditions.

[0027] The above embodiments collect real-time data on cell temperature, strain, gas concentration, and parameters such as working status, power, and runtime of thermal management components after thermal management operations. These data are then organized according to a preset data frame format, marked with an operation execution timestamp, and fed back to the spatiotemporal fusion module. The transmission channel frequency, data retransmission time threshold, and adjacent sensor unit association weights and data alignment accuracy parameters in the spatiotemporal association mapping table are dynamically adjusted. This effectively solves the problem of poor adaptability caused by the lack of dynamic feedback optimization and fixed parameters in traditional systems, continuously improving data transmission stability and spatiotemporal fusion accuracy. It provides dynamic adaptation guarantee for accurate risk assessment and precise execution of thermal management operations, further enhancing the system's adaptability to complex working conditions of special vehicles.

[0028] Specifically, the settings of the sensing units and the adjustment of the data acquisition interval in the sensing acquisition module include: A pre-defined standard for classifying battery cell types is established, which includes ternary lithium batteries, lithium iron phosphate batteries, and lithium cobalt oxide batteries. A first acquisition interval, a second acquisition interval, and a third acquisition interval are preset, wherein the first acquisition interval is less than the second acquisition interval and the third acquisition interval is less than the third acquisition interval; The initial value of the data collection interval is determined based on the actual type of the battery cells in the battery pack of the special vehicle. When the cell type is lithium cobalt oxide battery, the first sampling interval is selected as the initial sampling interval; When the cell type is a ternary lithium battery, the second sampling interval is selected as the initial sampling interval. When the cell type is lithium iron phosphate battery, the third sampling interval is selected as the initial sampling interval; Real-time monitoring of the deviation rate between valid data (temperature, strain, gas concentration) and preset benchmark values; setting a first threshold and a second threshold for the deviation rate, with the second threshold being greater than the first threshold: When the deviation rate is less than or equal to the first threshold of the deviation rate, the current collection interval remains unchanged; When the deviation rate is greater than the first threshold and less than or equal to the second threshold, the sampling interval will be shortened by half. When the deviation rate is greater than the second threshold of the deviation rate, the acquisition interval is shortened to the first acquisition interval, and the first acquisition interval is greater than or equal to the minimum preset number of seconds.

[0029] Specifically, the settings of the sensing units and the adjustment of the data acquisition interval in the sensing acquisition module are as follows: The classification standards for ternary lithium batteries, lithium iron phosphate batteries, and lithium cobalt oxide batteries are pre-set based on the chemical characteristics of the battery cells. Simultaneously, a first acquisition interval (0.5~1 second), a second acquisition interval (2~3 seconds), and a third acquisition interval (5~8 seconds) are preset, ensuring that the first acquisition interval is less than the second acquisition interval, which is less than the third acquisition interval. The minimum preset number of seconds is set to 0.5 seconds. The preset reference value is determined based on the factory rated parameters and safe operating thresholds of various battery cells. The deviation rate is calculated as (measured value - preset reference value) / preset reference value × 100%. A first threshold (5%~8%) and a second threshold (5%~8%) for the deviation rate are also set. The value is (10%~15%), and the second threshold of the deviation rate is greater than the first threshold of the deviation rate. In practical applications, the initial value of the collection interval is first determined according to the actual type of the cells in the battery pack of the special vehicle. That is, the first collection interval corresponds to the lithium cobalt oxide battery, the second collection interval corresponds to the ternary lithium battery, and the third collection interval corresponds to the lithium iron phosphate battery. Then, the deviation rate of the temperature, strain, and gas concentration data in the effective data and the preset benchmark value is monitored in real time. The collection interval is adjusted according to the range of the deviation rate. When the deviation rate is less than or equal to the first threshold, the current interval is maintained. When it is greater than the first threshold and less than or equal to the second threshold, the collection interval is shortened by half. When it is greater than the second threshold, it is shortened to the first collection interval to ensure that the collection interval is not less than the minimum preset number of seconds.

[0030] The above embodiments, by pre-setting differentiated initial acquisition intervals according to the characteristics of ternary lithium batteries, lithium iron phosphate batteries, and lithium cobalt oxide batteries, adapt to the different safety risks of different types of battery cells. At the same time, the acquisition interval is dynamically adjusted based on the deviation rate between temperature, strain, gas concentration data and preset benchmark values. When the deviation is small, the interval is maintained to avoid data redundancy, and when the deviation is large, the interval is shortened to ensure that no precursor signals of thermal runaway are missed. This effectively solves the problems of fixed acquisition intervals and mismatch with battery cell types and changes in operating status in traditional systems, improves the relevance, timeliness and effectiveness of data acquisition, and lays a reliable data foundation for subsequent spatiotemporal fusion and accurate risk assessment.

[0031] Specifically, the determination of the three-dimensional coordinates and the initial setting of the association weights in the spatiotemporal correlation mapping table include: The installation coordinates of each sensing unit are obtained based on the 3D modeling data of the battery pack. The coordinate axes include the X-axis, Y-axis, and Z-axis, where the X-axis corresponds to the length direction of the battery pack, the Y-axis corresponds to the width direction of the battery pack, and the Z-axis corresponds to the height direction of the battery pack. A first association weight, a second association weight, a third association weight, a first spatial distance, and a second spatial distance are pre-defined, wherein the first association weight is greater than the second association weight, which is greater than the third association weight, and the first spatial distance is less than the second spatial distance; Initial association weights are set based on the spatial distance between the sensing units: When the spatial distance between two sensing units is less than or equal to the first spatial distance, the spatiotemporal fusion module configures the first association weight of the two sensing units. When the spatial distance between two sensing units is greater than the first spatial distance and less than or equal to the second spatial distance, the spatiotemporal fusion module configures the second association weight of the two sensing units. When the spatial distance between two sensing units is greater than or equal to the second spatial distance, the spatiotemporal fusion module configures the third association weight of the two sensing units.

[0032] Specifically, based on the 3D modeling data of the battery pack (modeled using SolidWorks software with a modeling accuracy of ±0.1mm), the installation coordinates of each group of sensing units are obtained. The coordinate axes include the X-axis, Y-axis, and Z-axis, where the X-axis corresponds to the length direction of the battery pack, the Y-axis corresponds to the width direction of the battery pack, and the Z-axis corresponds to the height direction of the battery pack. The coordinate unit is uniformly millimeters (mm). The first association weight is preset to 0.8, the second association weight to 0.5, and the third association weight to 0.2. The first spatial distance is 50mm, and the second spatial distance is 100mm, and the conditions are met that the first association weight is greater than the second association weight, which is greater than the third association weight, and the first spatial distance is less than the second spatial distance. The 3D coordinate straight-line distance between any two sensing units is calculated using the Euclidean distance formula, and the initial association weight is set according to this spatial distance—when the spatial distance is less than or equal to 50mm, the spatiotemporal fusion module configures the first association weight of the two sensing units to be 0.8; when the spatial distance is greater than 50mm but less than or equal to 100mm, the second association weight is configured to be 0.5; and when the spatial distance is greater than or equal to 100mm, the third association weight is configured to be 0.2.

[0033] The above embodiments accurately obtain the installation coordinates of the sensing units along the X, Y, and Z axes (corresponding to the length, width, and height directions of the battery pack, respectively) based on the 3D modeling data of the battery pack, achieving precise binding between the data and the physical location of the battery cells. At the same time, the association weights are set according to spatial distance differences (high weight for close distances and low weight for distant distances), which solves the problem of lack of spatial association logic and low fusion accuracy of multi-source data in traditional systems. This strengthens the correlation and spatiotemporal matching degree of adjacent sensing unit data, improves the accuracy of spatiotemporal fusion datasets, and provides reliable support for the accurate location and judgment of thermal runaway risk.

[0034] Specifically, the data comparison algorithm of the risk assessment module includes: Based on the phased feature data intervals in the thermal runaway evolution path library, the upper and lower limits of the intervals for temperature, strain, and gas concentration are extracted, and the similarity between each parameter in the spatiotemporal fusion dataset and the corresponding interval is calculated. A first similarity threshold, a second similarity threshold, a third similarity threshold, and thresholds for the rate of change of temperature, strain, and gas concentration in the spatiotemporal fusion dataset are preset, and the first similarity threshold is less than the second similarity threshold and less than the third similarity threshold. The similarity is compared with each threshold: when the similarity is greater than or equal to the third similarity threshold, the thermal runaway risk level corresponding to the matching feature data interval of the spatiotemporal fusion dataset is directly determined; When the similarity is greater than or equal to the second similarity threshold and less than the third similarity threshold, the rate of change of temperature and gas concentration in the spatiotemporal fusion dataset is extracted. If the rate of change is greater than or equal to the preset rate of change threshold, the risk level corresponding to that interval is increased by one level; if the rate of change is less than the preset rate of change threshold, the risk level corresponding to that interval is maintained. When the similarity is greater than or equal to the first similarity threshold and less than the second similarity threshold, the risk level is raised by one level if the adjacent unit data matches a feature range greater than the risk level, based on the associated data of adjacent sensing units in the spatiotemporal correlation mapping table; otherwise, the original risk level is maintained. When the similarity is less than the first similarity threshold, it is determined to be a no-match risk level, a low-risk initial level is output, and it is marked as needing continuous monitoring.

[0035] Specifically, based on the phased feature data intervals in the thermal runaway evolution path library, the upper and lower limits of temperature, strain, and gas concentration intervals are extracted. A cosine similarity algorithm is used to calculate the similarity between each parameter in the spatiotemporal fusion dataset and its corresponding interval (similarity values ​​range from 0 to 1). Pre-set first similarity thresholds of 0.6, second similarity thresholds of 0.75, and third similarity thresholds of 0.9, ensuring that the first similarity threshold is less than the second similarity threshold, which is less than the third similarity threshold. Simultaneously, set thresholds for temperature change rate of 5℃ / min, strain change rate of 0.002 / min, and gas concentration (taking CO2 as an example) change rate of 0.5% / min. During comparison, if the similarity is greater than or equal to 0... 9. Directly determine the thermal runaway risk level corresponding to the matching feature data interval in the spatiotemporal fusion dataset; if the similarity is greater than or equal to 0.75 and less than 0.9, extract the real-time change rate of temperature and gas concentration in the spatiotemporal fusion dataset. If the change rate is greater than or equal to the corresponding preset threshold, the risk level is increased by one level; otherwise, the original level is maintained. If the similarity is greater than or equal to 0.6 and less than 0.75, combine the associated data of adjacent sensing units in the spatiotemporal association mapping table. If the adjacent unit data matches the feature interval higher than the current risk level, the risk level is increased by one level; otherwise, the original level is maintained. If the similarity is less than 0.6, it is determined that there is no matching risk level, outputs a low-risk initial level and marks it as needing continuous monitoring (the monitoring frequency is executed according to the shortest interval of the sensing acquisition module).

[0036] The above embodiments, by calculating multi-parameter similarity and combining multi-gradient similarity thresholds, parameter change rates, and associated data of adjacent sensing units for hierarchical comparison and dynamic adjustment of risk levels, solve the problems of one-sided, low-accuracy, and easy misjudgment and omission in traditional fixed threshold judgments. This enables phased and refined identification of thermal runaway risks, improves the accuracy, pertinence, and timeliness of risk judgment, and provides a reliable basis for the precise execution of subsequent graded thermal management operations.

[0037] Specifically, updates to the thermal runaway evolution path library include: Pre-set path library update conditions: the amount of new thermal runaway test data is greater than or equal to 50 sets, a new failure mode appears in the same type of cell, and the existing feature data interval matching accuracy is less than 85%. When any update condition is met, collect the multi-parameter change sequence, failure mode and corresponding stage data of the newly added thermal runaway test; Set a data validity threshold to remove abnormal data and retain valid new data; According to the rules for classifying thermal runaway evolution stages, the newly added data are classified into the corresponding stages, and the upper and lower limits of the characteristic data interval are updated. The updated path library is tested using a validation dataset. If the matching accuracy is greater than or equal to 90%, the updated results are saved; if the matching accuracy is less than 90%, the system reverts to the previous version and re-filters valid data.

[0038] Specifically, the path library update conditions are pre-set (checked at least once per quarter): ≥50 sets of thermal runaway test data under different ambient temperatures (-30℃~60℃) and different charge / discharge rates (0.5C~3C); new failure modes appearing in the same type of battery cell; and existing characteristic data range matching accuracy <85%. When any update condition is met, the data on temperature, strain, gas concentration, multi-parameter change sequences, failure modes, and corresponding stages of the newly added thermal runaway tests are collected. The data validity threshold is set to data deviation ≤±3%, including data exceeding -40℃~150℃, strain exceeding 0~0.01, and gas... Abnormal data with CO / CO2 concentrations exceeding 0%–20% were removed, and valid new data were retained. Valid new data were categorized into corresponding stages based on the evolutionary stages of preheating, heating up, thermal runaway, and cooling. The upper and lower limits of the feature data range were updated using a weighted average method. A sample representing 30% of the new valid data was selected as a validation dataset to test the updated path library. If the matching accuracy was ≥90%, the update results were saved. If the matching accuracy was <90%, the system reverted to the previous version, and valid data was re-screened by re-verifying the accuracy of the experimental equipment and supplementing with experimental data under the same conditions.

[0039] The above embodiments, by clearly defining three types of update triggering conditions—new experimental data volume, novel failure modes, and substandard matching accuracy—ensure the timely inclusion of multi-parameter data on thermal runaway and novel failure modes under different scenarios. Abnormal data is eliminated through data validity screening, and the updated path library matching accuracy is ensured through validation datasets. This solves the problems of outdated data, incomplete scenario coverage, and feature intervals that do not align with reality in traditional fixed path libraries. It continuously optimizes the completeness and accuracy of the path library, providing up-to-date and reliable data support for the dynamic and accurate assessment of thermal runaway risks.

[0040] Specifically, the instruction execution feedback verification of the hierarchical execution module includes: Pre-set the following indicators for judging the execution effect of the command: temperature drop rate, strain recovery rate, and gas concentration decay rate; Set a first judgment threshold and a second judgment threshold for each indicator, and the first judgment threshold shall be less than the second judgment threshold; After a preset time has elapsed since the instruction was executed, real-time data of the corresponding battery cell is collected, and the actual values ​​of each judgment indicator are calculated. The actual values ​​are compared with the thresholds: when the actual values ​​of all indicators are greater than or equal to the second judgment threshold, the judgment instruction is valid and the current instruction is maintained until the risk level decreases. When the actual value of at least one indicator is greater than or equal to the first judgment threshold and less than the second judgment threshold, the judgment instruction execution part is valid, and the power or duration of the thermal management operation is adjusted. When the actual values ​​of all indicators are less than the first judgment threshold, the judgment instruction is invalid, the current instruction level is upgraded by one level, and the instruction is regenerated and sent.

[0041] Specifically, pre-set criteria for judging the execution effect of the command include: temperature drop rate, strain recovery rate, and gas concentration decay rate. The first threshold for temperature drop rate is 2℃ / min, and the second threshold is 3℃ / min; the first threshold for strain recovery rate is 30%, and the second threshold is 50%; and the first threshold for gas concentration decay rate is 20%, and the second threshold is 40%. Each criterion satisfies the condition that the first threshold is less than the second threshold. The preset execution time for the command is set to 30 seconds. After this time, the sensor acquisition module collects real-time data on the temperature, strain, and gas concentration of the corresponding battery cell. These data are then calculated as (temperature before command execution - real-time temperature) / preset time and (real-time strain - strain before command execution) / strain before command execution, respectively. The actual values ​​of each judgment indicator are calculated using the formulas: ×100% and (Gas concentration before command execution - Real-time gas concentration) / Gas concentration before command execution ×100%. During comparison, if the actual values ​​of all indicators are ≥ the second judgment threshold, the command execution is deemed valid, and the current command is maintained until the thermal runaway risk level decreases. If the actual value of at least one indicator is ≥ the first judgment threshold and < the second judgment threshold, the command execution is deemed partially valid, and the power of the thermal management operation is increased by 20%-30% or the duration is extended by 50%-100%. If the actual values ​​of all indicators are < the first judgment threshold, the command execution is deemed invalid, and the current command level is upgraded by one level according to the preset command level system. A new command containing the adapted operation object, start time, and duration is generated and sent to the corresponding thermal management component.

[0042] The above embodiments, by pre-setting three judgment indicators—temperature drop rate, strain recovery rate, and gas concentration decay rate—and dual-gradient thresholds, collect data in real time after a preset time for command execution, calculate actual values, and compare them. The commands are dynamically maintained, operating parameters are adjusted, or command levels are increased according to three results: effective, partially effective, and ineffective. This solves the problems of lack of effect feedback, blind adjustment, and insufficient targeting in the execution of traditional thermal management commands. It ensures that thermal management operations are accurately matched with risk disposal needs, improves the effectiveness and timeliness of command execution, quickly curbs the spread of thermal runaway risk, and further protects the operational safety of special vehicle batteries.

[0043] Specifically, the dynamic adjustment of the deviation rate threshold includes: Collect valid data and corresponding deviation rates from the past 6 months, as well as actual thermal runaway risk records, and establish a deviation rate-risk correlation database. Calculate the risk identification accuracy under different deviation rate thresholds and set the target accuracy. If the recognition accuracy corresponding to the current deviation rate first threshold is less than the target accuracy, and the proportion of missed detection risk is greater than or equal to 10%, then the first threshold will be lowered by 10%. If the proportion of misjudgment risk corresponding to the current deviation rate second threshold is greater than or equal to 15%, then the second threshold will be increased by 15%. After adjustment, the recognition accuracy is recalculated. If the accuracy is greater than or equal to the target accuracy, the adjusted threshold is saved. If the accuracy is less than the target accuracy, the adjustment is repeated until the accuracy is greater than or equal to the target accuracy and the adjusted threshold is within the preset threshold range.

[0044] Specifically, effective data on cell temperature, strain, and gas concentration were collected daily from 8:00 AM to 10:00 PM for one hour and from 10:00 PM to 8:00 AM for two hours each day for the past six months, along with corresponding deviation rates and actual thermal runaway risk records marked as no risk, low risk, medium risk, and high risk. A deviation rate-risk correlation database was established. The risk identification accuracy was calculated at different deviation rate thresholds using (number of correctly identified risks / total number of risk records) × 100%, with a target accuracy of 92%. The preset range for the first threshold was defined as 3%-8%, and the preset range for the second threshold as 8%-15%. The percentage of missed risks was calculated as (number of unidentified actual risks / total number of actual risks). The percentage of false positives is calculated as (number of risk-free records misjudged as risk / total number of risk-free records) × 100%. If the recognition accuracy corresponding to the first threshold of the current deviation rate is <92% and the percentage of false negatives is ≥10%, then the first threshold is lowered by 10%. If the percentage of false positives corresponding to the second threshold of the current deviation rate is ≥15%, then the second threshold is raised by 15%. After adjustment, the recognition accuracy is recalculated. If the accuracy is ≥92% and the adjusted threshold is within the corresponding preset range, then the adjusted threshold is saved. If the accuracy is <92% or the threshold exceeds the preset range, then the above adjustment process is repeated until the accuracy meets the standard and the threshold is compliant.

[0045] The above embodiments establish a deviation rate-risk correlation database by collecting valid data from the past six months and actual thermal runaway risk records. Guided by the target accuracy, the first threshold is lowered for cases where the proportion of missed risks exceeds the standard, and the second threshold is raised for cases where the proportion of misjudged risks exceeds the standard. This solves the problems of low risk identification accuracy and prominent missed and misjudged cases caused by traditional fixed deviation rate thresholds. It continuously optimizes the threshold adaptability, improves the accuracy of risk identification, and provides reasonable threshold support for the scientific adjustment of the data collection interval and the reliable determination of subsequent thermal runaway risks. Specifically, the dynamic updating of association weights in the spatiotemporal association mapping table includes: Collect fault propagation records of adjacent sensor units in historical fault cases, and statistically analyze the matching accuracy of association weights under different spatial distances; Set a weighted matching accuracy threshold greater than or equal to 88%. When the weighted matching accuracy corresponding to a certain spatial distance is less than the weighted matching accuracy threshold: If the spatial distance is less than or equal to the first spatial distance, the first association weight will be increased by the first weight coefficient. If the spatial distance is greater than the first spatial distance and less than or equal to the second spatial distance, the second association weight will be increased by the second weight coefficient. When the spatial distance is greater than the second spatial distance, the third association weight will be increased by the third weight coefficient. After the update, the matching accuracy is recalculated. If the weighted matching accuracy is greater than or equal to 88%, the adjusted weights are saved. If the weighted matching accuracy is still less than 88%, the adjustment is repeated, and the total weights remain unchanged.

[0046] Specifically, historical fault cases related to thermal runaway of special vehicle batteries were collected over the past 12 months (no fewer than 100 cases in total). Fault propagation records of adjacent sensing units were extracted. The matching accuracy of association weights at different spatial distances (50mm, 50-100mm, and ≥100mm, the formula (number of times the association weight matches the actual fault propagation / total number of statistical times at the corresponding spatial distance) × 100% was used to calculate the accuracy. The weight matching accuracy threshold was set to ≥88%. The initial first association weight was 0.8, the second association weight was 0.5, and the third association weight was 0.2 (the sum was fixed at 1.5). The coefficients for the first weight were 0.1, the second weight was 0.08, and the third weight was 0.05. After adjustment, the first association weight was limited to no more than 0.95 and the second association weight was limited to no more than 0. 6. The third association weight shall not exceed 0.35. When the matching accuracy of the association weight corresponding to a certain spatial distance is <88%, if the spatial distance is ≤50mm, the first association weight shall be increased by 0.1 (first weight coefficient); if 50mm < spatial distance ≤100mm, the second association weight shall be increased by 0.08 (second weight coefficient); if the spatial distance is >100mm, the third association weight shall be increased by 0.05 (third weight coefficient). During the adjustment process, the total weight shall be kept constant at 1.5 by making single minor adjustments to other weights (with an amplitude not exceeding 0.03). After the update, the matching accuracy of the association weight under the corresponding spatial distance shall be recalculated. If the accuracy is ≥88% and the adjusted weight is within the limit range, the adjusted weight shall be saved. If it is still <88% or the weight exceeds the limit range, the above adjustment process shall be repeated until the matching accuracy reaches the standard and the weight is compliant.

[0047] The above embodiments, by collecting fault propagation records of adjacent sensor units in historical fault cases, statistically analyzing the correlation weight matching accuracy according to different spatial distances, and precisely adjusting the weights that do not reach the preset threshold of 88% while keeping the total unchanged, solve the problems of mismatch between traditional fixed correlation weights and actual fault propagation patterns and insufficient data correlation reliability. This continuously optimizes weight adaptability, improves the accuracy and rationality of spatiotemporal correlation of multi-source data, and provides more reliable correlation logic support for the accurate construction of spatiotemporal fusion datasets and the accurate location and judgment of thermal runaway risks.

[0048] Specifically, the revision of the thermal runaway risk level includes: After obtaining the thermal management operation data fed back by the closed-loop optimization module, and combining it with the spatiotemporal fusion dataset, the matching similarity with the thermal runaway evolution path library is recalculated. Set a risk level correction threshold. When the difference between the recalculated similarity and the initial similarity is greater than or equal to the risk level correction threshold: If the recalculated similarity increases, the original risk level will be raised by one level; if the recalculated similarity decreases, the original risk level will be lowered by one level. If the difference between the recalculated similarity and the initial similarity is less than the risk level correction threshold, the original risk level is maintained. The revised risk level is synchronously fed back to the hierarchical execution module to adjust the level of subsequent instructions.

[0049] Specifically, one minute after the thermal management operation is executed, real-time data on cell temperature, strain, gas concentration, and thermal management component operating parameters are obtained from the closed-loop optimization module. Combined with the original spatiotemporal fusion dataset, a cosine similarity algorithm (within the range of 0 to 1) is used to recalculate the matching similarity with the corresponding stage feature data intervals in the thermal runaway evolution path library. A risk level correction threshold of 0.1 is pre-set, clearly defining the thermal runaway risk level as low, medium, high, and extremely high. The corrected risk level must not be lower than low risk or higher than extremely high risk. When the difference between the recalculated similarity and the initial similarity is ≥0.1, if the recalculated similarity increases, the original risk level is increased by one level; if the recalculated similarity decreases, the original risk level is decreased by one level. When the difference between the recalculated similarity and the initial similarity is <0.1, the original risk level remains unchanged. The corrected risk level is synchronously fed back to the hierarchical execution module in real time, which adjusts the subsequent thermal management command level according to the corrected level to ensure that the command accurately matches the current risk state.

[0050] The above embodiments, by acquiring the data after thermal management operations fed back by the closed-loop optimization module, and recalculating the matching similarity with the thermal runaway evolution path library in combination with the original spatiotemporal fusion dataset, dynamically adjust, lower or maintain the original risk level according to the preset correction threshold and synchronize it to the graded execution module, solve the problem that the traditional risk level is fixed once determined and cannot adapt to the actual changes in risk after thermal management operations. This improves the real-time performance and accuracy of risk level determination, ensures that subsequent thermal management commands are accurately matched with the actual risk state of the battery cell, and more effectively curbs the development of thermal runaway risk.

[0051] Reference Figure 3In a specific embodiment of this application, the above steps are implemented as follows: Initial deployment is completed during the battery pack assembly stage of special vehicles (military off-road vehicles, engineering rescue vehicles, etc.). Technicians use SolidWorks software (modeling accuracy ±0.1mm) to create a 3D model of the battery pack. According to the assembly type of ternary lithium, lithium iron phosphate, and lithium cobalt oxide batteries, sensing units are installed at key positions in the X-axis (length), Y-axis (width), and Z-axis (height) directions. The spacing between sensing units for lithium cobalt oxide batteries is ≤50mm, and for lithium iron phosphate batteries, it is relaxed to 80~100mm. After system startup, it enters normal operation mode: the sensing acquisition module collects data at differentiated initial intervals (0.5~1 second for lithium cobalt oxide, 2~3 seconds for ternary lithium, and 5~8 seconds for lithium iron phosphate). The data is then processed by the receiving end to remove missing and duplicate values ​​to generate valid data. The spatial fusion module is allocated a 2.4GHz independent channel for transmission, encapsulated in the format of "cell number-component number-timestamp-parameter type-value". Millisecond-level timestamps and a default 50ms retransmission threshold ensure data timeliness and integrity. The spatiotemporal fusion module calculates the spatial distance of the sensing units using Euclidean distance, divides the intervals into 50mm and 100mm ranges, and configures association weights of 0.8, 0.5, and 0.2. It constructs a spatiotemporal association mapping table to complete the binding of data with physical locations and spatiotemporal alignment. The resulting spatiotemporal fusion dataset is compared with the thermal runaway evolution path library (including data from -30℃ to 60℃ and 0.5C to 3C operating conditions) by the risk judgment module using the cosine similarity algorithm. Under normal conditions, a similarity of less than 0.6 indicates low risk. The hierarchical execution module maintains the basic cooling cycle, and the closed-loop optimization module collects status data every 5 to 10 seconds to calibrate the transmission parameters. When the vehicle is traveling in high-temperature deserts, frigid snowfields, or off-road conditions, and the battery cell parameters fluctuate, the system enters a dynamic response mode: the sensor acquisition module calculates the deviation rate as (measured value - reference value) / reference value × 100%. When the deviation reaches the first threshold of 5%~8%, the acquisition interval is halved; when it exceeds the second threshold of 10%~15%, it is reduced to 0.5~1 second. If the spatiotemporal fusion module detects a packet loss rate >3%, it increases the channel frequency by 5%~10%; if the delay >20ms, it switches to a low-interference channel; and if the synchronization error >5ms, it adjusts the compensation coefficient to ensure an accuracy of ±2m. If the similarity of the risk assessment module is in the range of 0.75 to 0.9 and the temperature change rate is ≥5℃ / min, or the similarity is 0.6 to 0.75 and the data of adjacent units matches the higher risk range, the risk level will be increased. The graded execution module will start the cooling cycle (30kW) and heat dissipation channel switching for medium risk. After 30 seconds, it will be checked according to the temperature drop rate (≥2℃ / min), strain recovery rate (≥30%), and gas concentration decay rate (≥20%). If some are effective, the power will be increased by 20%. If not, the command level will be increased to trigger cell isolation.After fault handling, the system enters the closed-loop optimization phase: The closed-loop optimization module collects post-operation data and feeds it back to the spatiotemporal fusion module. If the reception success rate is <95%, the retransmission threshold is reduced to 30ms. If the data deviation rate between adjacent units is <5% and there are ≥10 synchronous changes, the correlation weight is increased by 0.05~0.1. When the deviation rate threshold is dynamically adjusted, if the first threshold has a missed detection rate ≥10% and an accuracy rate <92%, it is decreased by 10%; if the second threshold has a false detection rate ≥15%, it is increased by 15%. The spatiotemporal correlation mapping table is updated quarterly based on ≥100 fault cases from the past 12 months. If the matching accuracy rate is less than 88%, the weight is adjusted according to spatial distance. The system accurately adjusts the weight and maintains the total weight at 1.5. One minute after the thermal management operation, the risk level correction is initiated. When the similarity difference from the initial value is ≥0.1, the level is adjusted according to the rising and falling trend and synchronized to the hierarchical execution module. The path library is checked and updated every quarter. When ≥50 new sets of data are added or new failure modes appear, abnormal data with a deviation >±3% are removed. The feature interval is updated by classifying the preheating, heating, thermal runaway and cooling stages. After verification by 30% of the samples, the data is saved. In long-term use, the system will also adapt to the aging characteristics of the battery cells, summarize the fleet data to optimize the path library, and continuously improve the accuracy of early warning and the efficiency of handling in different scenarios.

[0052] The above scenarios are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0053] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0054] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0055] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0056] These computer program instructions can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A special vehicle battery thermal management safety warning system based on multi-source information fusion, characterized in that, The application relates to a battery thermal management system, which comprises the following parts: a sensing and collecting module, which is composed of a plurality of groups of sensing units, the sensing units are used for collecting temperature, strain and gas concentration data in the battery at preset intervals, and the collected data are sent to a receiving end, the receiving end performs preliminary checking on the data and eliminates missing data and repeated data in the data, and valid data are generated; a space-time fusion module, which is used for establishing a carrier transmission synchronization model, allocating independent transmission channels, setting transmission frequency and data frame format, generating a millisecond-level timestamp and embedding a retransmission mechanism; the space-time fusion module is also used for constructing a space-time correlation mapping table, which records the three-dimensional coordinates of each group of the sensing units and the battery cell number, classifies the valid data according to the battery cell number and aligns and correlates the data through a space-time field, and forms a space-time fusion data set; a risk judgment module, which is used for establishing a thermal runaway evolution path library, the thermal runaway evolution path library stores the multi-parameter variation sequence, failure mode and staged characteristic data interval of different battery thermal runaway tests, matches the space-time fusion data set and the characteristic data interval through a data comparison algorithm, and outputs a thermal runaway risk level; a hierarchical execution module, which is used for dividing instruction levels according to the thermal runaway risk level, each instruction level corresponds to a combination of thermal management operations of cooling medium circulation start, heat dissipation channel switching, battery isolation device triggering and thermal management component power adjustment, generates an instruction containing an operation object, a starting time and a duration and sends the instruction to the corresponding thermal management component.

2. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 1, characterized in that, The application further comprises: a closed-loop optimization module, which is used for collecting real-time data of temperature, strain and gas concentration in the battery after the thermal management operation, and collecting operation parameters of the thermal management component such as working state, power and running time; the data are arranged according to the data frame format preset by the space-time fusion module, and an operation execution timestamp is marked; the arranged data are fed back to the space-time fusion module, the transmission channel frequency and data retransmission time threshold in the carrier transmission synchronization model are adjusted, the correlation weight of adjacent sensing units and the data alignment accuracy parameters in the space-time correlation mapping table are updated, and system dynamic optimization is performed.

3. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 1, characterized in that, The setting of the sensing units in the sensing and collecting module and the adjustment of the data collection interval comprise the following steps: pre-set battery type division standards, the battery types include ternary lithium battery, iron phosphate lithium battery and cobalt lithium battery; pre-set a first collection interval, a second collection interval and a third collection interval, and the first collection interval is smaller than the second collection interval, and the second collection interval is smaller than the third collection interval; determine the initial value of the collection interval according to the actual type of the battery in the battery pack of the special vehicle: when the battery type is the cobalt lithium battery, the first collection interval is selected as the initial collection interval; when the battery type is the ternary lithium battery, the second collection interval is selected as the initial collection interval; when the battery type is the iron phosphate lithium battery, the third collection interval is selected as the initial collection interval; monitor the deviation rate of the temperature, strain and gas concentration data in the valid data from the preset reference value in real time, set a first threshold value of the deviation rate and a second threshold value of the deviation rate, and the second threshold value of the deviation rate is greater than the first threshold value of the deviation rate: when the deviation rate is less than or equal to the first threshold value of the deviation rate, the current collection interval is maintained unchanged; When the deviation rate is greater than the first deviation rate threshold and less than or equal to the second deviation rate threshold, the collection interval is shortened by one-half; When the deviation rate is greater than the second deviation rate threshold, the collection interval is shortened to the first collection interval, and the first collection interval is greater than or equal to the minimum preset seconds.

4. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 1, characterized in that, The three-dimensional coordinate determination and initial setting of the correlation weight of the spatiotemporal correlation mapping table include: Based on the three-dimensional modeling data of the battery pack, the installation coordinates of each group of sensing units are obtained, and the coordinate axes include X-axis, Y-axis and Z-axis, wherein the X-axis corresponds to the length direction of the battery pack, the Y-axis corresponds to the width direction of the battery pack, and the Z-axis corresponds to the height direction of the battery pack; The first correlation weight, the second correlation weight, the third correlation weight and the first spatial distance and the second spatial distance are preset, and the first correlation weight is greater than the second correlation weight, and the second correlation weight is greater than the third correlation weight, and the first spatial distance is less than the second spatial distance; According to the spatial distance of the sensing unit, the initial correlation weight is set: When the spatial distance of two sensing units is less than or equal to the first spatial distance, the spatiotemporal fusion module configures the first correlation weight of the two sensing units; When the spatial distance of two sensing units is greater than the first spatial distance and less than or equal to the second spatial distance, the spatiotemporal fusion module configures the second correlation weight of the two sensing units; When the spatial distance of two sensing units is greater than or equal to the second spatial distance, the spatiotemporal fusion module configures the third correlation weight of the two sensing units.

5. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 1, characterized in that, The data comparison algorithm of the risk determination module specifically includes: Based on the interval of the staged feature data in the thermal runaway evolution path library, the upper limit and the lower limit of the interval of temperature, strain and gas concentration are extracted, and the similarity of each parameter in the spatiotemporal fusion data set and the corresponding interval is calculated; The first similarity threshold, the second similarity threshold, the third similarity threshold and the change rate threshold of temperature, strain and gas concentration in the spatiotemporal fusion data set are preset, and the first similarity threshold is less than the second similarity threshold, and the second similarity threshold is less than the third similarity threshold; The similarity is compared with each threshold: when the similarity is greater than or equal to the third similarity threshold, the thermal runaway risk level corresponding to the matching feature data interval of the spatiotemporal fusion data set is directly determined; When the similarity is greater than or equal to the second similarity threshold and less than the third similarity threshold, the change rate of temperature and gas concentration in the spatiotemporal fusion data set is extracted, if the change rate is greater than or equal to the preset change rate threshold, the risk level corresponding to the interval is adjusted by one level; if the change rate is less than the preset change rate threshold, the risk level corresponding to the interval is maintained; When the similarity is greater than or equal to the first similarity threshold and less than the second similarity threshold, the correlation data of adjacent sensing units in the spatiotemporal correlation mapping table is combined, if the adjacent unit data matches more than the feature interval of the risk level, the current risk level is adjusted by one level; if the adjacent unit data matches less than or equal to the feature interval of the risk level, the original level is maintained; When the similarity is less than the first similarity threshold, it is determined that there is no matching risk level, and a low risk initial level is output, and it is marked that it needs to be continuously monitored.

6. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 1, characterized in that, The update of the thermal runaway evolution path library includes: Pre-set path library update condition: the number of newly added thermal runaway test data is greater than or equal to 50 groups, the same type of battery cell appears a new failure mode, and the matching accuracy of the existing characteristic data interval is less than 85%; When any of the update conditions is met, collect the multi-parameter change sequence, failure mode and corresponding stage data of the newly added thermal runaway test; Set a data validity threshold, eliminate abnormal data and retain valid new data; According to the thermal runaway evolution stage division rule, the valid new data is classified into the corresponding stage, and the upper and lower limit values of the characteristic data interval are updated; The updated path library is tested by using the verification data set, if the matching accuracy is greater than or equal to 90%, the updated result is saved; if the matching accuracy is less than 90%, the previous version is returned, and the valid data is reselected.

7. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 1, characterized in that, The instruction execution feedback verification of the hierarchical execution module includes: Pre-set instruction execution effect determination index: temperature drop rate, strain recovery rate, gas concentration decay rate; Set the first determination threshold and the second determination threshold of each index, and the first determination threshold is less than the second determination threshold; After the preset duration of instruction execution, the real-time data of the corresponding battery cell is collected, and the actual value of each determination index is calculated; Compare the actual value with the threshold value: when the actual value of all indexes is greater than or equal to the second determination threshold, it is determined that the instruction execution is effective, and the current instruction is maintained until the risk level decreases; When at least one index actual value is greater than or equal to the first determination threshold and less than the second determination threshold, it is determined that the instruction execution is partially effective, and the power or duration of the thermal management operation is adjusted; When the actual value of all indexes is less than the first determination threshold, it is determined that the instruction execution is invalid, the current instruction level is adjusted by one level, and the instruction is regenerated and sent.

8. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 3, characterized in that, The dynamic adjustment of the deviation rate threshold includes: Collect the effective data and corresponding deviation rate and actual thermal runaway risk record in the past 6 months, and establish a deviation rate-risk association database; Calculate the risk identification accuracy rate under different deviation rate thresholds, and set the target accuracy rate; If the identification accuracy rate corresponding to the first threshold of the current deviation rate is less than the target accuracy rate, and the missed risk proportion is greater than or equal to 10%, the first threshold is lowered by 10%; If the misjudgment risk proportion corresponding to the second threshold of the current deviation rate is greater than or equal to 15%, the second threshold is increased by 15%; After adjustment, the identification accuracy rate is recalculated, if the accuracy rate is greater than or equal to the target accuracy rate, the adjusted threshold is saved, if the accuracy rate is less than the target accuracy rate, the adjustment is repeated until the accuracy rate is greater than or equal to the target accuracy rate, and the adjusted threshold is within the preset threshold range.

9. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 4, characterized in that, The dynamic update of the association weight in the space-time association mapping table includes: Collect the failure propagation records of adjacent sensing units in historical failure cases, and calculate the matching accuracy rate of the association weight under different spatial distances; Set the weight matching accuracy threshold to be greater than or equal to 88%, when the association weight matching accuracy corresponding to a certain spatial distance is less than the weight matching accuracy threshold: If the spatial distance is less than or equal to the first spatial distance, the first association weight is increased by the first weight coefficient; If the spatial distance is greater than the first spatial distance and less than or equal to the second spatial distance, the second correlation weight is adjusted by a second weight coefficient; When the spatial distance is greater than the second spatial distance, the third correlation weight is adjusted by a third weight coefficient; After updating, the matching accuracy is re-counted, if the weight matching accuracy is greater than or equal to 88%, the adjusted weight is saved, if the weight matching accuracy is less than 88%, the adjustment is repeated, and the total sum of the adjusted weight remains unchanged.

10. The multi-source information fusion based special vehicle battery thermal management safety warning system according to claim 2, characterized in that, The modification of the thermal runaway risk level includes: Obtain the thermal management operation data after feedback from the closed-loop optimization module, combine the spatio-temporal fusion data set, and re-calculate the matching similarity with the thermal runaway evolution path library; Set a risk level modification threshold, when the difference between the re-calculated similarity and the initial similarity is greater than or equal to the risk level modification threshold: If the re-calculated similarity increases, the original risk level is adjusted by one level; if the re-calculated similarity decreases, the original risk level is adjusted by one level; When the difference between the re-calculated similarity and the initial similarity is less than the risk level modification threshold, the original risk level is maintained; The modified risk level is fed back to the hierarchical execution module, and the subsequent instruction level is adjusted.

Images