Battery system state evaluation and early warning method and system based on multi-source information fusion

Abstract

The invention provides a battery system state evaluation and early warning method based on multi-source information fusion, and the method comprises the steps: synchronously collecting the multi-source state data of a battery system, which at least comprises a real-time synchronization parameter group for cooperatively evaluating the mechanical integrity and environment interaction risk, the real-time synchronization parameter group comprises vibration acceleration data, organic volatile matter concentration data, humidity data and corrosion rate data; processing the data of the real-time synchronization parameter group, extracting combined feature parameters having an association relationship with each other to form a feature vector, the combined feature parameters at least comprising: a structure risk feature, a sealing risk feature, and a connection risk feature; and inputting the feature vectors into a preset multi-parameter fusion analysis model to analyze a linkage mode among the combined feature parameters, and synchronously generating and outputting a safety risk report based on an analysis result of the linkage mode, the safety risk report at least comprising a mechanical integrity risk index and an environment interaction risk index.

Description

Technical Field

[0001] This application relates to the field of battery management system and battery state monitoring technology, specifically to a method and system for battery system state assessment and early warning based on multi-source information fusion. Background Technology

[0002] With the rapid development of the new energy industry, electrochemical energy storage systems, represented by lithium-ion batteries, have been widely used in electric vehicles, large-scale energy storage, and other fields. Ensuring the safe and reliable operation of battery systems is one of the challenges facing the industry. Currently, the battery management systems commonly used in the industry are mainly based on monitoring the basic electrical and thermodynamic parameters of the battery, such as voltage, current, and temperature. They estimate the battery's state of charge and health status through preset algorithm models and trigger protection or alarms when parameters are abnormal.

[0003] However, the aforementioned technical solutions still have limitations. Battery systems face diverse and coupled risks in practical applications, including not only internal electrochemical abuse such as overcharging, over-discharging, and internal short circuits, but also broader mechanical abuse such as impacts and vibrations, sealing failures such as electrolyte leakage, and environmental interactions such as corrosion and water ingress. Existing management solutions lack sufficient perception of these external and structural risks, employing single monitoring dimensions and lacking effective correlation analysis and fusion between parameters. This leads to delayed early warnings for complex faults and an inability to achieve early, comprehensive safety status assessments. For example, the system may only issue an alarm in the later stages of irreversible phases such as the large-scale generation of thermal runaway gases or a sudden drop in battery voltage, missing the optimal opportunity for intervention in the early stages of structural damage, minor leaks, or corrosion. Summary of the Invention

[0004] This application provides a method and system for battery system status assessment and early warning based on multi-source information fusion, in order to solve the problems of single monitoring dimensions and lack of correlation fusion in battery system monitoring.

[0005] To achieve the above technical objectives, the embodiments of this application provide the following technical solutions: In a first aspect, one embodiment of this application provides a method for battery system state assessment and early warning based on multi-source information fusion, the method comprising: The system synchronously collects multi-source state data of the battery system, including at least a set of real-time synchronous parameters for collaboratively assessing the risks of mechanical integrity and environmental interaction. The set of real-time synchronous parameters includes: vibration acceleration data, volatile organic compound concentration data, humidity data, and corrosion rate data. The data of the real-time synchronization parameter group is processed to extract the combined feature parameters that are related to each other, forming a feature vector. The combined feature parameters include at least: structural risk features generated based on vibration acceleration time series data to characterize structural impact and fatigue accumulation; sealing risk features generated based on the synergistic change trend of volatile organic compound concentration and humidity to characterize electrolyte leakage; and connection risk features generated based on corrosion rate data to characterize the deterioration of electrical connection point conditions. The feature vector is input into a preset multi-parameter fusion analysis model to analyze the linkage mode between the combined feature parameters. Based on the analysis results of the linkage mode, a safety risk report is generated and output simultaneously. The safety risk report includes at least a mechanical integrity risk index and an environmental interaction risk index.

[0006] In some optional embodiments, the step of synchronously acquiring multi-source state data of the battery system includes, in addition to, the real-time synchronous parameter set acquired for assessing pressure relief and casing integrity, oxygen concentration data for assessing fire risk, and collision trigger signal for directly determining physical collision events.

[0007] The introduction of air pressure data enables the system to monitor changes in internal battery pack pressure, effectively distinguishing between normal battery breathing effects and malfunctions such as casing damage and pressure relief valve abnormalities. Oxygen concentration data provides key environmental parameters for assessing the potential intensity of fire occurrence and spread after battery thermal runaway. Collision trigger signals provide a definitive assessment of physical impact, complementing vibration acceleration data. The inclusion of these parameters makes the assessment of environmental interaction risks more comprehensive and accurate, covering the entire chain from casing sealing to the internal atmospheric environment, and providing a dual chain of evidence for mechanical impact.

[0008] In some optional embodiments, the step of synchronously acquiring multi-source state data of the battery system specifically includes: first aggregating the data collected by each sensor deployed at key locations of the battery pack or module to the cell monitoring controller of the corresponding battery box for preliminary processing, and then converging it to the multi-dimensional data fusion controller through wired or wireless communication.

[0009] A two-level distributed processing architecture of cell monitoring controller and multi-dimensional data fusion controller is adopted. The cell monitoring controller, as an edge node, completes local data preprocessing, which reduces the bandwidth occupation of the central network by the raw data and the instantaneous computing pressure on the multi-dimensional data fusion controller. Furthermore, the failure of a single cell monitoring controller does not affect the data acquisition of other units. Moreover, it facilitates the accurate time synchronization of sensors in each box and provides a modular solution for the expansion of the battery system scale.

[0010] In some optional embodiments, the extraction of related combined feature parameters to form a feature vector specifically includes: extracting feature values ​​of pressure change rate, specific gas concentration of volatile organic compounds, AC impedance phase angle, average expansion force, vibration dominant frequency energy, corrosion current, humidity value, oxygen concentration, and the collision trigger signal, and combining the feature values ​​to form the feature vector.

[0011] This approach concretizes the feature extraction process into a defined vector consisting of nine key feature values, achieving standardization and structuring of multi-source heterogeneous data. Each feature dimension represents the evolutionary information of a specific risk pattern; for example, the rate of change in air pressure represents the depressurization dynamics, the dominant vibration frequency energy represents the structural resonance state, and the corrosion current directly quantifies the corrosion rate.

[0012] In some optional embodiments, the preset multi-parameter fusion analysis model includes an electrolyte leakage judgment model; the electrolyte leakage judgment model is configured to execute the following judgment logic: if the following conditions are met simultaneously: the concentration feature value of volatile organic compounds in the feature vector exceeds a first preset threshold; the humidity feature value in the feature vector exceeds a second preset threshold; the change amplitude of the air pressure change feature value in the feature vector is lower than a third preset threshold; the AC impedance ohmic component feature value in the feature vector exceeds a fourth preset threshold; then a seal failure risk judgment result characterizing electrolyte leakage and contact with air is generated.

[0013] This embodiment enables early diagnosis of electrolyte leakage faults. The effectiveness relies on the fusion of multiple parameters using an AND logic: excessive volatile organic compound (VOC) concentration is direct evidence of leakage; abnormally high humidity is chemical evidence of hydrolysis between the leaking electrolyte and moisture in the air; unchanged air pressure eliminates interference factors such as environmental ventilation; and an increase in the ohmic component of the AC impedance is electrical evidence of interface deterioration and increased internal resistance due to electrolyte loss within the battery. These four conditions form a mutually supportive and logically rigorous chain of evidence, effectively distinguishing between genuine leaks and other situations such as sensor false alarms and environmental interference, thus improving the accuracy and reliability of the diagnosis.

[0014] In some optional embodiments, the method further includes a collision probability assessment step: based on vibration acceleration sampling data used to capture impact spectrum characteristics, a collision damage probability index is calculated and output in real time using a built-in impact response spectrum algorithm and a material damage model.

[0015] This embodiment uses the shock response spectrum algorithm to analyze the frequency domain energy distribution of high sampling rate vibration data, and combines it with a material damage model for the specific structure of the battery pack to realize the probabilistic and quantitative assessment of the collision perception from whether a collision has occurred to how great the damage risk is.

[0016] In some optional embodiments, the analysis of the linkage mode between the combined feature parameters specifically includes: analyzing the correlation and evolution trend between the structural risk feature, the sealing risk feature, and the connection risk feature through the multi-parameter fusion analysis model; the synchronous generation and output of the safety risk report specifically includes: generating assessment results corresponding to the mechanical integrity state and environmental interaction risk state of the battery system, respectively, based on the analysis of the correlation and evolution trend.

[0017] In this embodiment, the multi-parameter fusion analysis model not only analyzes the absolute values ​​of features but also delves deeper into the dynamic relationships between cross-domain features. For example, it analyzes whether long-term vibration accelerates the corrosion of connectors or whether leakage alters the local environment. Based on this in-depth analysis of correlations and evolution trends, the final report clearly decomposes the overall risk situation into two independent assessment dimensions: mechanical integrity status and environmental interaction risk status. This allows for rapid identification of the root causes of risks, enabling more targeted countermeasures and improving the efficiency and effectiveness of safety management.

[0018] In some optional embodiments, after the step of synchronously generating and outputting the security risk report, a security action execution step is further included: based on the security risk report, triggering and executing corresponding proactive security measures, the proactive security measures including at least one of activating the ventilation system, isolating the faulty unit, and notifying the maintenance platform.

[0019] This embodiment endows the battery system with automated active protection capabilities, improving the timeliness and effectiveness of safety response. It initiates ventilation and dilution when a flammable gas risk is detected; controls a relay to achieve electrical isolation when a battery cell is determined to have a serious fault risk; and simultaneously reports alarm information and data to the maintenance platform, enabling automatic intervention in the early stages of an accident and effectively preventing escalation.

[0020] Secondly, one embodiment of this application also provides a battery system state assessment and early warning system based on multi-source information fusion, used to implement the method described above. The system includes: a sensor array, including a vibration acceleration sensor, an organic volatile matter concentration sensor, a humidity sensor, and a corrosion rate sensor, used to acquire data from the real-time synchronized parameter group; a data acquisition and processing unit, configured to synchronously acquire data from the sensor array and process the data to extract the combined feature parameters to form a feature vector; a fusion analysis and early warning unit, configured to receive the feature vector, analyze the linkage mode between the combined feature parameters through a preset multi-parameter fusion analysis model, and generate a safety risk report based on the analysis results; and a controller area network, used to connect the sensor array, the data acquisition and processing unit, and the fusion analysis and early warning unit to realize data transmission and command interaction.

[0021] This embodiment solidifies the entire process of the aforementioned method using specific hardware modules and software units. The sensor array enables in-situ acquisition of non-electrical parameters; the data acquisition and processing unit is used for synchronous information aggregation and feature extraction; the fusion analysis and early warning unit performs intelligent fusion analysis and decision-making; and the controller area network ensures reliable interaction of instructions and data between modules. This system constitutes a deployable integrated hardware and software solution that can be integrated into various battery application scenarios.

[0022] In some optional embodiments, the deployment of each sensor in the sensor array is specifically configured as follows: the vibration acceleration sensor is deployed on the key load-bearing structure of the battery pack housing and internal module support of the battery system; the volatile organic compound concentration sensor is deployed in the air circulation areas of the battery pack air chamber and battery compartment; the humidity sensor is deployed inside the battery module or near the connector of the high-voltage sampling harness; the corrosion rate sensor is attached to the surface of the high-voltage electrical connector in the battery system; and the system also includes an expansion force / strain sensor, which is attached to the side plate or end plate of the battery module, or clamped between adjacent cells.

[0023] This embodiment maximizes the monitoring efficiency of each sensor through targeted spatial layout. Vibration sensors are deployed on critical load-bearing structures to effectively capture the overall structural mechanical response; volatile organic compound (VOC) concentration sensors are deployed in gas chambers and flow channels to facilitate the capture of leaked gas accumulation and diffusion; humidity sensors are deployed inside moisture-prone modules and wiring harness connectors to detect moisture intrusion at the earliest possible time; corrosion sensors are mounted on the surfaces of high-voltage connectors to achieve in-situ, direct monitoring of the weakest points in electrical safety; and the addition and proper arrangement of expansion force sensors monitors changes in the mechanical state of the battery during charging, discharging, and aging. This deployment strategy, deeply coupled with the physical structure and failure mechanism of the battery system, ensures that the collected data is the most representative.

[0024] Thirdly, one embodiment of this application also provides a vehicle including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described above.

[0025] Fourthly, one embodiment of this application also provides a computer program product, the computer program product comprising a computer program stored in a computer-readable storage medium; a processor of a computer device reads the computer program from the computer-readable storage medium, and when the processor executes the computer program, it implements the steps of the above-described battery system state assessment and early warning method based on multi-source information fusion. Optionally, the computer program may be stored in the readable storage medium of the computer device or in the cloud; the processor of the computer device reads the computer program from the readable storage medium or in the cloud.

[0026] Fifthly, one embodiment of this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0027] As can be seen from the above technical solutions, the battery system state assessment and early warning system, vehicle, computer program product, and computer-readable storage medium based on multi-source information fusion provided in this application all correspondingly implement the battery system state assessment and early warning method based on multi-source information fusion described in the first aspect. This method constructs a mechanical-chemical-environmental risk collaborative perception and assessment system, breaking through the limitation of traditional battery management systems that only focus on electrical parameters. It simultaneously collects non-electrical parameters that directly reflect external risks, such as vibration, volatile organic compounds, humidity, and corrosion, thus expanding the monitoring dimensions from the data source. Secondly, the method does not simply list the raw data, but extracts combined features with clear physical meaning, such as structural risk features, sealing risk features, and connection risk features, realizing the transformation from single data to multi-source associated information and emphasizing the correlation between features. Finally, by mining the linkage patterns between these features through a preset fusion analysis model, the originally isolated risk signals are linked together, thereby enabling the identification of complex coupled faults. This achieves early, quantitative, and comprehensive assessment of the mechanical integrity and environmental interaction risks of the battery system, ultimately upgrading the early warning from a post-event alarm based on a single threshold to risk prediction based on multi-parameter correlation analysis. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0029] Figure 1 This is a flowchart illustrating a battery system state assessment and early warning method based on multi-source information fusion, provided as one embodiment of this application.

[0030] Figure 2 This is a schematic diagram of the architecture of a battery system state assessment and early warning system based on multi-source information fusion, provided as one embodiment of this application.

[0031] Figure 3 This is a schematic diagram of the architecture of a vehicle provided for one embodiment of this application. Detailed Implementation

[0032] Unless otherwise defined, the technical or scientific terms used in the embodiments of this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to avoid confusion of the constituent elements.

[0033] Unless the context otherwise requires, throughout this specification, "a plurality of" means "at least two," and "including" is interpreted as open-ended or encompassing, that is, "including, but not limited to." In the description of this specification, terms such as "one embodiment," "some embodiments," "exemplary embodiment," "example," "specific example," or "some examples" are intended to indicate that a particular feature, structure, material, or characteristic associated with that embodiment or example is included in at least one embodiment or example of this application. The illustrative representations of the above terms do not necessarily refer to the same embodiment or example.

[0034] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0035] Overview Currently, battery management systems (BMS) widely deployed in engineering practice almost entirely revolve around internal electrical and thermal parameters such as voltage, current, and temperature in their core monitoring and algorithm models, forming a narrow safety perspective centered on the internal state of the battery cell. For crucial system-level safety issues such as whether the battery pack casing has developed microcracks due to long-term vibration, whether high-voltage connectors are silently corroding in high-humidity environments, and whether there are minor electrolyte leaks, existing solutions either completely lack sensing methods or provide isolated, delayed, and unreliable sensing. This current situation of single monitoring dimensions and lack of correlation and integration between parameters directly leads to delayed fault warnings and makes it impossible to conduct early and comprehensive assessments of the health and safety status of the battery system.

[0036] The inventive concept of this application begins with a re-examination and definition of the aforementioned technical challenges. We first extend the safety boundary of the battery system from the inside of the cell to the mechanical integrity of the entire battery system and its interaction with the surrounding environment. Based on this, the core concept of this invention is to construct a non-electrical parameter sensing system directly addressing mechanical and environmental risks. Through information fusion technology, these parameters are no longer isolated alarm points, but rather interconnected and mutually corroborating, jointly depicting the risk characteristics of the overall system safety posture. Specifically, the concept is divided into three layers: First, systematic synchronous sensing: synchronously collecting a carefully selected set of non-electrical parameters that directly characterize the target risk, such as vibration, volatile organic compounds and humidity, and corrosion. Second, correlation feature purification: extracting combined features with clear physical meaning and potential correlations from the raw data stream, such as structural risk features and sealing risk features. Third, linked intelligent diagnosis: using a fusion analysis model to explore the inherent linkage patterns and evolution trends between these cross-domain features, ultimately outputting quantitative assessments of the mechanical integrity risk index and the environmental interaction risk index. The essence of this concept is to achieve a leap from monitoring the internal state of the battery to assessing the comprehensive safety status of the battery system throughout its entire life cycle, thereby transforming passive alarms into proactive risk prediction.

[0037] Exemplary methods One embodiment of this application provides a battery system state assessment and early warning method based on multi-source information fusion, see [link to relevant documentation]. Figure 1 , Figure 1 The present application provides a flowchart illustrating a battery system state assessment and early warning method based on multi-source information fusion, wherein the method includes: Step S110: Synchronously acquire multi-source state data of the battery system, including at least a set of real-time synchronous parameters for collaboratively assessing the risks of mechanical integrity and environmental interaction.

[0038] This step forms the foundation for building the sensing network. The system activates and coordinates a cluster of sensors deployed at key locations within the battery system to acquire time-synchronized data. The data collected is a set of real-time synchronization parameters specifically defined for assessing the risks of mechanical integrity and environmental interaction, and includes at least the following data: Vibration acceleration data comes from high dynamic range accelerometers deployed on the main load-bearing and force transmission paths, such as the battery pack crossbeams, side walls, and internal module supports. It is used to assess structural damage, namely, to capture transient impact events and monitor the continuous road vibration spectrum.

[0039] Volatile organic compound (VOC) concentration data and humidity data are used to jointly assess seal failure. VOC concentration data is acquired using a metal-oxide-semiconductor (MOS) sensor optimized for electrolyte solvents, deployed in the top gas chamber within the battery pack—the area where leaked vapors most easily accumulate. Humidity data is acquired using a high-precision humidity sensor, preferably deployed inside the module or in localized microenvironments prone to leakage or condensation, such as sampling harness connectors. The joint assessment of VOC concentration and humidity data is necessary because, in electrolyte leakage scenarios, solvent evaporation (increased VOC concentration) and the decomposition of electrolytic salts upon contact with water (increased humidity) often occur simultaneously; joint analysis improves diagnostic specificity.

[0040] Corrosion rate data serves as a direct indicator for assessing the reliability of electrical connections. This data is derived from miniaturized corrosion rate sensors attached to the surfaces of critical electrical connectors such as high-voltage copper busbars and connecting bolts. These sensors are typically based on electrochemical impedance spectroscopy or thin-film resistance variation principles, enabling real-time, in-situ measurement of corrosion current or resistance change rates on metal surfaces. This directly reflects the corrosion state of the connection points, providing early warnings of overheating or even short-circuit risks caused by increased contact resistance.

[0041] The sampling clocks of all sensors are synchronized by the main controller to ensure accurate time-series correlation analysis, which is a prerequisite for realizing linkage pattern mining.

[0042] Step S120: Process the data of the real-time synchronization parameter group, extract the combined feature parameters that are related to each other, and form a feature vector.

[0043] This step involves information purification and feature engineering of the raw data collected in step S110, extracting correlated combined feature parameters from the time-series data stream. These combined feature parameters include at least: structural risk features generated from vibration acceleration time-series data to characterize structural impact and fatigue accumulation; sealing risk features generated from the coordinated changing trends of volatile organic compound concentration and humidity to characterize electrolyte leakage; and connection risk features generated from corrosion rate data to characterize the deterioration of electrical connection points.

[0044] The structural risk characteristics are generated based on vibration acceleration time-series data using digital signal processing algorithms, jointly characterizing structural impact and fatigue accumulation. For example, a fast Fourier transform is performed on continuous vibration signals to extract the dominant frequency energy reflecting the natural frequency vibration energy of the battery pack structure; time-domain integration and envelope analysis are performed on impact event signals to extract impact intensity and impact duration; and vibration load spectra are extracted through statistical analysis of long-term vibration data for fatigue damage estimation.

[0045] The sealing risk features are generated based on the coordinated changing trends of volatile organic compound (VOC) concentration and humidity, and are used to characterize the possibility and dynamic process of electrolyte leakage. Feature extraction focuses not only on the absolute concentration values ​​of VOC and humidity, but also on their dynamic correlation. For example, the instantaneous rate of change and trend slope of VOC concentration are calculated; the covariance or correlation coefficient between humidity and VOC concentration within the same time window is calculated; and a joint time-series change model of VOC concentration and humidity is constructed.

[0046] The connection risk feature is generated based on corrosion rate data and directly characterizes the severity and rate of deterioration of electrical connection points. This feature directly uses the corrosion current value output by the sensor, or calculates its average corrosion rate and corrosion acceleration trend.

[0047] These feature parameters, extracted from different physical domains but all pointing to external risks to the system, are combined to form a multi-dimensional feature vector. This operation means mapping multi-source heterogeneous information to a unified structured feature space, providing standardized and information-intensive input for the next step of the fusion analysis model.

[0048] Step S130: Input the feature vector into the preset multi-parameter fusion analysis model to analyze the linkage mode between the combined feature parameters. Based on the analysis results of the linkage mode, generate and output a security risk report simultaneously.

[0049] The feature vector obtained in step S120 is input into a preset multi-parameter fusion analysis model. The main task of the multi-parameter fusion analysis model is to go beyond independent threshold judgment of individual features and instead focus on analyzing the linkage patterns between the combined feature parameters. For example, the multi-parameter fusion analysis model might analyze whether the corrosion rate of the corresponding connection risk feature abnormally accelerates in the following days after a severe impact on the structural risk feature, which may indicate that the impact caused the sealing strip to fail and moisture intrusion accelerated corrosion. Alternatively, when the concentration of volatile organic compounds and humidity of the corresponding sealing risk feature increase synergistically, whether the structural risk feature remains calm, this helps to rule out leakage caused by collision and confirm spontaneous sealing failure.

[0050] Based on the analysis results of the aforementioned linkage mode, the multi-parameter fusion analysis model synchronously generates and outputs a safety risk report, which includes at least the mechanical integrity risk index and the environmental interaction risk index.

[0051] The Mechanical Integrity Risk Index is a quantitative output of the fusion analysis model that comprehensively assesses structural risk characteristics and their interrelationships with other characteristics. It is used to quantify the level of structural damage and fatigue risk. The higher the Mechanical Integrity Risk Index value, the greater the risk of structural loosening, crack propagation, or overall mechanical failure of the battery pack.

[0052] The Environmental Interaction Risk Index is a quantitative output of the model after comprehensively assessing the characteristics of sealing risks, connection risks, and possible interrelationships of environmental parameters. It is used to quantify the level of sealing failure and corrosion risk. The higher the Environmental Interaction Risk Index value, the greater the risk of electrolyte leakage, intrusion of external corrosive media, or adverse changes in the environment inside the battery compartment.

[0053] This embodiment addresses the issues of limited monitoring dimensions and lack of integrated correlation in battery system monitoring by executing the aforementioned steps. By defining and synchronously acquiring real-time parameter groups, the monitoring dimensions are expanded from a single electrical domain to mechanical, chemical, and environmental domains. By extracting and combining feature parameters, an information leap and correlation preprocessing from low-level sensor signals to high-level risk semantics is achieved, providing a foundation for intelligent analysis. By analyzing linkage patterns and generating index reports on mechanical integrity risk and environmental interaction risk, a transformation from parameter exceedance alarms to multi-parameter coupling relationship diagnosis is realized. The final output is no longer an isolated alarm, but rather a mechanical integrity risk index and an environmental interaction risk index, enabling battery safety management to move from focusing on cell health to assessing system safety, achieving early, comprehensive evaluation and warning.

[0054] In some optional embodiments, the step of synchronously acquiring multi-source state data of the battery system includes, in addition to, the real-time synchronous parameter set acquired for assessing pressure relief and casing integrity, oxygen concentration data for assessing fire risk, and collision trigger signal for directly determining physical collision events.

[0055] Specifically, the system continuously monitors the absolute pressure or differential pressure changes of the internal air chambers of the battery pack relative to the external environment through high-precision micro-differential pressure sensors deployed at key locations inside the battery pack. The pressure data used to assess pressure relief and casing integrity has dual diagnostic value: firstly, it can sensitively capture the periodic breathing effect caused by the minute changes in electrode volume due to lithium-ion insertion / extraction during normal charging and discharging, with pressure changes typically within the normal range; secondly, any abnormal rapid rise or sustained slow decline in pressure deviating from this normal pattern will be identified by the system as a risk signal to casing integrity. By analyzing the trends, rates of change, and steady-state values ​​of the pressure data, the system can effectively distinguish between normal physiological phenomena and early signs of failure.

[0056] The system acquires real-time data on the volumetric oxygen concentration in the cabin atmosphere using electrochemical or optical oxygen sensors deployed in the air domain of the battery compartment. This data is defined as a core parameter for assessing fire risk. When the battery system experiences thermal runaway, and flammable gases such as hydrogen, carbon monoxide, and electrolyte vapors leak into the cabin, the occurrence and severity of a fire are closely related to the oxygen concentration within the cabin. Monitoring dynamic changes in oxygen concentration, such as sudden drops due to combustion or recovery due to ventilation, provides environmental input for assessing the development stage of thermal runaway events, potential fire intensity, and optimal fire suppression strategies, thus achieving a leap from detecting thermal runaway to assessing its consequences and risks.

[0057] To provide an immediate and reliable response to sudden mechanical abuse incidents, impact-sensitive strips or pressure-sensitive films with special mechanical properties are placed on the outside of the battery box or on critical protective structures of the chassis. Under normal conditions, this device exhibits a high resistance state, for example, greater than 1000 ohms; when the battery system is subjected to compression or impact exceeding the design threshold, the device undergoes irreversible physical deformation, causing its resistance value to drop sharply, for example, less than 10 ohms. The resulting abrupt change in electrical signal is defined as the impact trigger signal used to directly determine physical impact events.

[0058] In this embodiment, a three-dimensional environmental risk assessment is achieved. Both air pressure and oxygen concentration data extend the assessment perspective from the inside of the battery pack to the sealed casing state and the atmospheric environment within the compartment. Air pressure data allows the system to assess the physical sealing health of the casing, while oxygen data quantifies the environment's potential to contribute to fire risk, creating a complete closed loop for environmental risk assessment from the inside out. Furthermore, collision trigger signals and vibration acceleration data form a double safeguard for sensing mechanical impacts. Vibration data analysis focuses on the quantitative analysis of impact events and damage probability prediction, while collision trigger signals provide simple and direct qualitative confirmation. When both are triggered simultaneously, it can be confirmed as a high-confidence collision event; if there is only abnormal vibration without a collision signal, further analysis combining other data is required; if only a collision signal is triggered with slight vibration, it suggests possible localized slow compression. This redundant design reduces false alarms and false negatives.

[0059] Slow gas leakage can be one of the earliest signs of shell corrosion or seal aging, and its correlation with corrosion rate data can provide early warning of the coupled risk of structural corrosion leading to seal failure. Changes in baseline oxygen concentration can also reflect the status of the cabin ventilation system. These parameters collectively enhance the system's ability to detect early, slowly changing faults, making the comprehensive assessment more holistic and the conclusions more reliable.

[0060] In some optional embodiments, the step of synchronously acquiring multi-source state data of the battery system specifically includes: first aggregating the data collected by each sensor deployed at key locations of the battery pack or module to the cell monitoring controller of the corresponding battery box for preliminary processing, and then converging it to the multi-dimensional data fusion controller through wired or wireless communication.

[0061] In this embodiment, the step of synchronously acquiring multi-source state data of the battery system specifically includes a distributed data acquisition and aggregation hardware architecture, and the specific process is as follows: Sensor deployment and local data acquisition: Various types of sensors, such as vibration, gas, humidity, corrosion, air pressure, and strain sensors, are deployed in key locations of the battery pack, modules, and cabin. These sensors convert physical and chemical signals into raw electrical signals in real time.

[0062] Initial data aggregation and processing at the cell monitoring controller: Each physically independent battery box or pack is equipped with a cell monitoring controller. All sensor data deployed within this battery box or pack (including collision strip signals) is first aggregated to the cell monitoring controller of that single pack. As an edge computing node, the cell monitoring controller's initial processing functions include at least the following: It amplifies and filters analog signals and performs high-precision analog-to-digital conversion.

[0063] The battery cell monitoring controller can immediately identify a collision event locally when the resistance of the collision strip changes abruptly.

[0064] The digital signals are initially filtered and calibrated, and data from different sensors within the same time window are stamped with a unified timestamp and encapsulated into standard data frames.

[0065] The structured data, after initial processing by the cell monitoring controller, is aggregated to the central multi-dimensional data fusion controller via wired or wireless communication. Typically, a high-reliability controller area network (LAN) bus or local interconnection network (NAT) bus is used. The LAN bus features priority arbitration and strong anti-interference capabilities, making it suitable for reliable transmission of critical safety data in the complex electromagnetic environment of a vehicle. Wireless methods can employ short-range wireless communication technologies such as Bluetooth Low Energy (BLE) or ZigBee.

[0066] Edge - side pre - processing is performed through the battery cell monitoring controller, dispersing the computational burden of a large amount of raw data. This enables the central multi - dimensional data fusion controller to focus on feature fusion and risk analysis algorithms without having to process a huge amount of raw signals, thus improving the overall response speed and real - time performance of the system. The battery cell monitoring controller stamps accurate time stamps on local data. The multi - dimensional data fusion controller aligns and fuses the data uploaded from each battery box according to this time stamp, ensuring the temporal synchronization of multi - source data from different parts of the battery system, which is a prerequisite for subsequent linkage mode analysis. The distributed architecture avoids the risk of the entire system crashing due to a single - point failure. The failure of a single battery cell monitoring controller or its sensor network is usually isolated within the battery box. At the same time, this modular design makes the system easy to expand, providing a clear engineering path for the application of large - scale battery systems (such as energy storage power stations). The data processed by the battery cell monitoring controller is already refined feature information or compressed data packets, rather than raw waveform data, which reduces the bandwidth pressure on the backbone communication network and avoids data congestion.

[0067] In some optional embodiments, extracting combined feature parameters that are related to each other to form a feature vector specifically includes: extracting the air pressure change rate, the concentration of specific organic volatile compounds, the phase angle of the AC impedance, the average expansion force, the main frequency energy of vibration, the corrosion current, the humidity value, the oxygen concentration, and the eigenvalue of the collision trigger signal, and jointly forming the feature vector with these eigenvalues.

[0068] In this embodiment, extracting combined feature parameters that are related to each other to form a feature vector specifically includes a standardized feature engineering process, aiming to transform multi - source heterogeneous raw state data into an information - intensive and structurally unified mathematical expression. The specific extraction and formation methods are as follows: Perform real - time calculations on the time - series air pressure data collected by the air pressure sensor to obtain the air pressure change rate, usually obtained through differential or derivative operations within a unit time. The air pressure change rate is used to characterize the dynamic change rate of the internal pressure of the battery pack and is a key indicator for distinguishing normal slow breathing from abnormal rapid pressure relief or leakage.

[0069] Directly read or calibrate the signal from the organic volatile compound concentration sensor to obtain the concentration of organic volatile compounds for the electrolyte solvent. The concentration of organic volatile compounds is the most direct chemical evidence of electrolyte leakage or decomposition.

[0070] Measure the AC impedance of the battery cell through a dedicated impedance measurement circuit or through the equalization line of the slave unit of the battery management system or by injecting a small - amplitude multi - frequency AC excitation signal additionally. Extract the phase - angle information at a specific characteristic frequency from the measured impedance spectrum. The phase angle of the AC impedance is very sensitive to the state changes at the internal electrode / electrolyte interface of the battery.

[0071] The average value of the expansion force is obtained by averaging the data collected by strain gauges or pressure sensor arrays deployed on the side plates, end plates, or between the cells of the module within a time window. The average expansion force quantifies the overall mechanical expansion degree of the battery module and is related to the internal side reactions and aging state of the battery.

[0072] Frequency domain analysis is performed on the time-series signals acquired by the vibration acceleration sensor to obtain the dominant vibration frequency energy. The structural condition is monitored by analyzing this dominant vibration frequency energy. The dominant vibration frequency energy is obtained by calculating the integral of the vibration energy of the battery pack or module within the frequency band near its primary natural frequency. It is used to characterize the intensity of structural resonance and reflects the tightness of mechanical connections and the overall health of the structure.

[0073] The corrosion current is obtained by directly reading the output of a corrosion rate sensor based on electrochemical principles. The corrosion rate is then determined by inverting the corrosion current. The corrosion current value is directly proportional to the electrochemical reaction rate of metal corrosion and is a direct physical quantity for quantifying the degree of corrosion at electrical connection points.

[0074] The system directly reads calibrated data from humidity sensors deployed inside the module or at the wiring harness connectors, i.e., the ambient relative humidity or absolute humidity value. Humidity is a key parameter for assessing ambient moisture intrusion and electrolyte hydrolysis reactions.

[0075] The oxygen concentration is obtained by directly reading the calibrated data (oxygen volume fraction) from the oxygen concentration sensor deployed in the battery compartment air. Oxygen concentration is used to assess fire risk and environmental oxidizing properties.

[0076] The step-type trigger signal from devices such as the collision strip is characterized to obtain the characteristic value of the collision trigger signal. After the collision strip is squeezed, the resistance decreases from 1000 ohms to below 10 ohms. This change in resistance or the resulting voltage change (i.e., the collision ΔV) can be used as a characteristic value to characterize the confirmation and rough intensity of the collision event.

[0077] Within each sampling or analysis cycle, the data processing unit synchronously performs the extraction operations of the aforementioned nine types of feature values, and then combines these nine values ​​into an ordered, fixed-dimensional array, namely the feature vector. For example, a feature vector can be represented as: [ΔP / Δt, C VOC , θ EIS , F swell E vib , I corr , RH, C O2 , ΔV collision ].

[0078] In summary, raw sensor data from completely different physical domains, such as mechanical vibration, gas chemistry, electrochemical impedance spectroscopy, and pressure strain, are uniformly transformed into pure numerical features with clearly defined physical units. The selected nine types of features, after targeted signal processing, can characterize specific risk patterns. For example, the rate of change of air pressure reveals dynamic faults better than the absolute value of air pressure; the dominant frequency energy of vibration reflects changes in the inherent properties of the structure better than the original acceleration waveform. Placing these nine features in the same vector space allows the fusion analysis model to directly calculate the correlation, covariance, or nonlinear dependence between them, thus providing mathematically operable input objects for linkage mode analysis. These nine-dimensional feature vectors essentially map the comprehensive safety state of the battery system to a feature space that can be efficiently processed by machine learning or intelligent algorithms.

[0079] In some optional embodiments, the preset multi-parameter fusion analysis model includes an electrolyte leakage judgment model; the electrolyte leakage judgment model is configured to execute the following judgment logic: If the following conditions are met simultaneously: The concentration of volatile organic compounds in the feature vector exceeds a first preset threshold. The humidity feature value in the feature vector exceeds a second preset threshold. The magnitude of the change in the air pressure feature value in the feature vector is lower than the third preset threshold. The eigenvalue of the AC impedance ohmic component in the eigenvector exceeds a fourth preset threshold. This generates a risk assessment result for seal failure, indicating that the electrolyte has leaked and has come into contact with air.

[0080] In this embodiment, the preset multi-parameter fusion analysis model includes an executable expert rule sub-model—an electrolyte leakage judgment model. The electrolyte leakage judgment model takes four key feature values ​​from the aforementioned nine-dimensional feature vector as direct input and executes a set of multi-condition joint judgment logic. The input and judgment logic of the electrolyte leakage judgment model are detailed below: The electrolyte leakage detection model is configured to continuously monitor the following four dimensions of the feature vector: Characteristic value of volatile organic compound concentration C VOC This corresponds to the concentration of volatile organic compounds in the feature vector.

[0081] The humidity feature value H corresponds to the humidity value in the feature vector.

[0082] characteristic value of air pressure change ΔP Rate This corresponds to the rate of change of air pressure in the eigenvector.

[0083] Eigenvalue R of the ohmic component of AC impedance OhmIt can be separated from the online measured AC impedance spectrum data, reflecting the ohmic internal resistance of the battery.

[0084] The electrolyte leakage detection model follows an "AND" logic, meaning all conditions must be met simultaneously. The model will trigger a final decision only if all four of the following preset conditions are met: Condition 1: C VOC >θ1 (First preset threshold). This condition corresponds to the judgment of a sharp increase in the concentration of volatile organic compounds (VOCs). The θ1 threshold is determined based on the normal background VOC concentration and the identifiable concentration jump in the early stages of a leak. This is direct evidence of the evaporation of organic solvents in the electrolyte.

[0085] Condition 2: H > θ2 (Second preset threshold). This condition corresponds to the judgment of a significant increase in humidity. The θ2 threshold needs to take into account the baseline ambient humidity and the additional humidity increase caused by the hydrolysis reaction that may occur after the electrolyte leak. Some electrolyte will hydrolyze when it comes into contact with moisture in the air, producing substances such as hydrofluoric acid, which leads to an abnormal increase in local ambient humidity and is accompanying chemical evidence of leakage.

[0086] Condition 3: |ΔP Rate |<θ3 (Third preset threshold). This condition is used to determine if there is no significant change in air pressure. θ3 is a small absolute value used to define the range of no significant change. This condition is used to exclude abnormal readings of volatile organic compounds and humidity caused by battery pack breathing effects, start-up and shutdown of the environmental ventilation system, or fluctuations in external atmospheric pressure. It is a key exclusionary condition to improve diagnostic specificity.

[0087] Condition 4: R Ohm >θ4 (Fourth preset threshold). This condition corresponds to the judgment of an increase in the ohmic component of the battery impedance. The θ4 threshold is set based on the baseline ohmic internal resistance under healthy battery conditions. Electrolyte leakage will cause a decrease in the amount or concentration of electrolyte inside the battery, resulting in deterioration of the electrode-electrolyte interface, thereby causing a significant increase in ohmic internal resistance. This is direct evidence of an abnormal internal state of the battery itself.

[0088] The execution flow of the electrolyte leakage assessment model is as follows: The system continuously reads sensor data and checks the four conditions mentioned above sequentially or in parallel. If all four conditions are confirmed to be met simultaneously within the same analysis time window, the electrolyte leakage assessment model immediately generates a seal failure risk assessment result characterizing electrolyte leakage and exposure to air. This assessment not only includes the factual judgment of leakage but also further infers the stage of exposure to air, indicating more urgent chemical reaction risks, such as violent hydrolysis or potential combustion. Subsequently, this assessment result is integrated into the environmental interaction risk index with high weight and triggers high-level warnings and safety protocols.

[0089] In summary, by employing a strict AND operation of four conditions, the electrolyte leak detection model constructs a multi-layered chain of evidence. False alarms from a single sensor or environmental interference typically cannot simultaneously satisfy all four conditions, especially the two supporting conditions—unchanged gas pressure and increased impedance—which stem from different principles. This solves the problem of high false alarm rates in single-gas sensor warning systems. The electrolyte leak detection model integrates direct evidence of the leak event (volatile organic compounds), chemical reaction evidence (humidity), environmental exclusion evidence (gas pressure), and evidence of the leak's physical state (impedance), achieving cross-domain information verification. This allows the system to make a high-confidence judgment early in the leak's occurrence, when indirect evidence is still weak but a combined pattern has already emerged. The electrolyte leak detection model outputs not only an alarm but also a diagnostic conclusion including the nature and stage of the leak. This provides guidance for implementing differentiated safety measures and enhances the overall approach's ability to manage core safety risks.

[0090] In some optional embodiments, the method further includes a collision probability assessment step: based on vibration acceleration sampling data used to capture impact spectrum characteristics, a collision damage probability index is calculated and output in real time using a built-in impact response spectrum algorithm and a material damage model.

[0091] This embodiment includes a collision probability assessment step that runs parallel to the main process, enabling the probabilistic and quantitative assessment of potential latent structural damage after a collision. The specific implementation of this step is as follows: The input for this step is vibration acceleration sampling data used to capture the spectral characteristics of the impact. The vibration sensor channel used for this analysis needs to be configured with a sufficiently high sampling rate, for example, no less than 10 kHz, to ensure that the high-frequency components contained in the transient impact event can be captured without distortion, providing a basis for spectral analysis. When the amplitude of the vibration acceleration exceeds a low initial threshold, the system determines that a potential impact event has occurred and triggers this evaluation process.

[0092] The system retrieves high-sampling-rate vibration acceleration time-history data within a time window before and after the event, and inputs it into the built-in Shock Response Spectrum (SRS) algorithm. The Shock Response Spectrum is an analytical method that converts time-domain shock signals to the frequency domain. It describes the maximum response, such as acceleration, velocity, or displacement, of a single-degree-of-freedom system under a shock at its series of natural frequencies. Calculating the Shock Response Spectrum essentially quantifies the potential excitation intensity of the shock on structural components with different resonant frequencies.

[0093] After obtaining the impact response spectrum, the system compares and calculates it against a pre-defined material damage model. The material damage model is based on the mechanical properties of the specific battery pack structure and includes: Structural model: Simplified mechanical models of key components such as battery housing, module bracket, and cell fixing structure, and their natural frequencies.

[0094] Material properties: Fatigue characteristic curves (SN curves) or impact damage criteria of the materials used in the above components.

[0095] Damage accumulation algorithms: such as Miner's linear cumulative damage rule.

[0096] The material damage model calculates the theoretical damage caused to each component by analyzing the response amplitude of the impact response spectrum at the natural frequency of each structural component and combining it with the fatigue strength of the material.

[0097] By combining the theoretical damage assessment results of all key components and using a predetermined synthesis or mapping algorithm, such as taking the weighted maximum value of the damage of each component or performing system-level reliability calculations, a normalized collision damage probability index is calculated and output in real time. This index is a value between 0 and 1, such as the collision probability. The higher the value, the greater the statistical probability that the impact will cause measurable or latent damage (such as microcracks, plastic deformation, or loosening of connectors) to the battery system structure.

[0098] In summary, this collision probability assessment step transforms the time-domain waveform of a single impact into a spectrum of potential destructive forces on structures at different frequencies through impact response spectra and material damage models. This allows for the calculation of a quantified collision probability, achieving a refined and graded assessment conclusion. The collision probability index output in this step is a probabilistic prediction of this latent damage risk, enabling the system to provide early warnings of potential long-term consequences at the moment a collision event occurs. The output collision probability can be directly used to guide maintenance, achieving optimized resource allocation and rapid response to high-risk events. Furthermore, this probabilistic assessment complements the collision trigger signal; the collision strip provides physical contact evidence, while this step provides damage analysis. The combination of these two ensures the real-time nature and reliability of collision perception, together forming a collision safety perception system.

[0099] In some optional embodiments, the analysis of the linkage mode between the combined feature parameters specifically includes: analyzing the correlation and evolution trend between the structural risk feature, the sealing risk feature, and the connection risk feature through the multi-parameter fusion analysis model; the synchronous generation and output of the safety risk report specifically includes: generating assessment results corresponding to the mechanical integrity state and environmental interaction risk state of the battery system, respectively, based on the analysis of the correlation and evolution trend.

[0100] In this embodiment, the process of analyzing the linkage pattern between the combined feature parameters and simultaneously generating a security risk report is specifically defined as an intelligent analysis process that deeply explores the inherent causal relationship of cross-domain features and generates targeted assessment conclusions accordingly.

[0101] The detailed analysis process of the linkage mode is as follows: The multi-parameter fusion analysis model deeply analyzes the dynamic and nonlinear correlations and evolution trends among structural risk characteristics, sealing risk characteristics, and connection risk characteristics extracted from different physicochemical processes. This analysis goes beyond threshold judgments of individual characteristics, aiming to reveal potential pathways for cross-domain risk transmission and coupling. Specific analytical content includes: the multi-parameter fusion analysis model quantifies the synchronicity, lag, or causality among these characteristics using statistical or machine learning methods. For example: analyzing whether long-term high-level fluctuations in structural risk characteristics have a significant positive correlation with the growth rate of connection risk characteristics; analyzing whether the temporal changes in volatile organic compound concentration and humidity within sealing risk characteristics are highly coordinated, and whether their coordinated change pattern conforms to the electrolyte leakage judgment model, thereby confirming the leakage diagnosis; analyzing whether sealing risk characteristics remain within a subsequent time window after a sudden peak in structural risk characteristics. Anomalies are detected to determine whether the collision caused immediate damage to the sealing structure. The model focuses not only on the instantaneous values ​​of features, but also on their direction of change, acceleration, and pattern over a period of time. For example, it determines whether the connection risk feature shows a stable linear upward trend or an inflection point of accelerated increase, with the latter indicating a higher risk level. It analyzes whether the spectral distribution of the structural risk feature drifts slowly over time, which may indicate a decrease in structural connection stiffness and a change in natural frequency, an early sign of structural fatigue. It monitors whether the upward trend of volatile organic compound concentration in the sealing risk feature is accompanied by a slow downward trend of oxygen concentration after reaching a certain plateau, which may suggest that the leaked material is undergoing slow oxidation and consumption.

[0102] The specific process for generating a security risk report is as follows: Based on the above in-depth analysis of the correlation and evolution trend, the synchronous generation and output of the security risk report is specifically executed as a process of attribution decomposition and comprehensive judgment: The model will primarily be based on the analysis of structural risk characteristics, with a focus on their correlation with connection risk characteristics. For example, when the model identifies significant and continuous high-frequency vibration energy, coupled with a correlated acceleration in corrosion rate, even if the absolute values ​​of both do not trigger an electrical alarm, the mechanical integrity status may be comprehensively determined to be of moderate risk, indicating that the risk mainly stems from the coupling of loose connections and corrosion that may be caused by continuous vibration. This assessment result is a holistic evaluation of structural health and connection reliability.

[0103] The model will primarily be based on the analysis of sealing risk characteristics and will comprehensively examine the evolution trends of connection risk characteristics and environmental parameters. For example, when the concentration of volatile organic compounds and humidity increase in tandem but the air pressure remains stable and the impedance increases, the model will not only determine the leak, but will also combine whether the corrosion rate is synchronously abnormal and whether the oxygen concentration changes to assess the comprehensive impact level of the leak on environmental safety, and output the conclusion that the environmental interaction risk status is high risk.

[0104] In summary, by clearly analyzing the correlations among the three core risk characteristics, the safety risk report can include explanatory notes, enhancing the credibility of the assessment results and its guiding value for maintenance actions. Proactively analyzing the relationships and evolution trends between characteristics allows for the identification and early warning of complex failure modes. The resulting mechanical integrity status assessment report and environmental interaction risk status assessment report provide precise guidance for taking countermeasures. If the risk is concentrated on the mechanical side, the maintenance focus is on structural tightening and inspection; if the risk is concentrated on the environmental interaction side, the focus is on sealing checks and environmental control system inspections. This targeted risk decomposition improves resource allocation efficiency and risk response efficiency.

[0105] In some optional embodiments, after the step of synchronously generating and outputting the security risk report, a security action execution step is further included: based on the security risk report, triggering and executing corresponding proactive security measures, the proactive security measures including at least one of activating the ventilation system, isolating the faulty unit, and notifying the maintenance platform.

[0106] The safety action execution steps transform the cognitive results generated by intelligent analysis into direct intervention and control of entities. The specific implementation logic and measures of this step are as follows: The system receives and parses the safety risk report generated by the fusion analysis and early warning unit. The mechanical integrity risk index, environmental interaction risk index, and specific diagnostic conclusions in the safety risk report, such as electrolyte leakage determination and high probability of collision damage, together constitute the input for the execution decision. Based on the preset response strategy, the system automatically triggers and executes corresponding proactive safety measures, specifically including one or more combinations of the following: When the environmental interaction risk index enters the high-risk range, or when the model directly determines that an electrolyte leak has occurred and has come into contact with air, the ventilation system is activated. At this time, the system sends a command to the battery compartment environmental management system to immediately activate the explosion-proof forced ventilation device, rapidly diluting the flammable volatile organic compounds leaked from the battery pack and the toxic gases that may be generated by electrolyte hydrolysis, reducing the gas concentration in the compartment to below the safe limit, eliminating the risk of fire and explosion, and improving environmental safety.

[0107] When a report clearly indicates a high risk in a battery box, battery cluster, or module, the faulty unit is isolated. At this point, the system, in conjunction with the high-voltage power distribution management interface of the battery management system, sends a command to disconnect the high-voltage contactor or relay of the faulty unit. This physically disconnects the faulty unit from the main electrical circuit, preventing the fault from spreading to the entire battery system.

[0108] The maintenance platform is simultaneously notified whenever any level of warning or proactive measure is triggered. At this time, the system uploads a complete safety risk report, relevant raw data snapshots, timestamps, and recommended measures in real time to the cloud-based operations and maintenance platform or the vehicle manufacturer's service backend via a wireless network through the in-vehicle telematics processor or gateway. Simultaneously, alert information is pushed to the driver through the in-vehicle human-machine interface, or work order notifications are sent to fleet administrators and maintenance personnel via a mobile app. This ensures seamless flow of risk information between the vehicle, cloud, and personnel.

[0109] In summary, by automatically executing initial response measures through pre-programmed logic, the system interrupts the initial stage of the incident chain. This forms a complete safety closed loop of perception, analysis, decision-making, and execution, enhancing the active safety level and autonomy of the battery system. It provides tiered and precise active protection; the execution of measures is based on tiered responses according to safety risk reports, ensuring safety while avoiding unnecessary system interruptions or resource consumption. Through the notification and maintenance platform, all risk events and system responses are fully recorded and uploaded to the cloud. This accumulates data assets for post-incident root cause analysis, strategy optimization, and predictive maintenance, enabling the safety system to continuously learn and improve.

[0110] Exemplary System In one exemplary embodiment of this application, a battery system state assessment and early warning system 200 based on multi-source information fusion is also provided, see [link to relevant documentation]. Figure 2 , Figure 2 This application provides an architecture diagram of a battery system state assessment and early warning system based on multi-source information fusion as one embodiment of the present application. The battery system state assessment and early warning system 200 based on multi-source information fusion is a tightly integrated hardware and software entity, including a sensor array 210, a data acquisition and processing unit 220, a fusion analysis and early warning unit 230, and a controller area network 240.

[0111] The sensor array 210 is the physical basis for building a fusion sensing network of electrical, mechanical, chemical and environmental domains. It includes vibration acceleration sensors, volatile organic compound concentration sensors, humidity sensors, corrosion rate sensors and expansion force / strain sensors.

[0112] Vibration acceleration sensors are specifically deployed on key load-bearing and force-transmitting structures such as the longitudinal beams, transverse beams, and internal module supports of the battery pack housing to ensure that vibration excitation and collision impact from the road surface can be effectively captured on the overall structure.

[0113] The volatile organic compound (VOC) concentration sensor is specifically deployed in the air chamber at the top of the battery pack and along the airflow path within the battery compartment to monitor the concentration of diffused gas.

[0114] Humidity sensors are specifically deployed in the gaps inside the battery module or near the connector insulation sleeve of the high-voltage sampling harness. These locations are critical points where moisture is most likely to penetrate, condense, and pose a direct threat to electrical safety.

[0115] The corrosion rate sensor is specifically mounted on the surface of high-voltage electrical connectors within the battery system, such as the connection surface of copper-aluminum composite busbars and bolt fastening points. This in-situ installation method can directly sense the corrosion state of the connection interface.

[0116] There are two optimized deployment methods for expansion force / strain sensors: one is to attach them to the outer surface of the side or end plate of the battery module to monitor the overall expansion pressure of the module; the other is to clamp them between adjacent cells to directly measure the expansion force between the cells. Both methods provide direct data for assessing the degree of battery bulging and internal side reaction pressure.

[0117] In addition, the sensor array 210 can further integrate a pressure sensor, an oxygen concentration sensor, and a collision signal sensor (such as a collision rubber strip interface) to form a fully functional multidimensional sensing network.

[0118] The data acquisition and processing unit 220 consists of two levels of hardware: a cell monitoring controller and a multi-dimensional data fusion controller. Each battery box is equipped with one cell monitoring controller, which synchronously acquires all signals from the sensor array 210 within its box, performs analog filtering, digital filtering, baseline calibration, and preliminary data encapsulation, and adds precise timestamps. The multi-dimensional data fusion controller receives preprocessed data packets from each cell monitoring controller via a network such as CAN. Its built-in processor runs feature extraction algorithms, performing tasks such as the aforementioned nine-dimensional feature vector calculation and the aforementioned impact response spectrum analysis.

[0119] The fusion analysis and early warning unit 230 can function as a standalone computing module or be integrated into the high-performance chip of the data acquisition and processing unit 220. The fusion analysis and early warning unit 230 stores and runs a preset multi-parameter fusion analysis model, which receives feature vectors from the data acquisition and processing unit 220 and performs the following intelligent tasks: Analyze the complex linkage patterns among the parameters in the feature vector, and perform the correlation and trend analysis as described above; Based on the analysis results, a safety risk report is generated that includes a mechanical integrity risk index and an environmental interaction risk index. According to the preset strategy, a local early warning signal is triggered directly, or a control command is sent to the actuator.

[0120] The controller area network 240 is used to connect the sensor array 210, the data acquisition and processing unit 220 and the fusion analysis and early warning unit 230, and ensures the uplink transmission of sensor data, the downlink distribution of control commands and the status interaction between the units.

[0121] In the battery system state assessment and early warning system 200 based on multi-source information fusion provided in this embodiment, the targeted deployment of the sensor array 210 ensures the high quality and representativeness of the collected signals from the source. The data acquisition and processing unit 220 achieves global information fusion through central processing. The standardized safety risk report output by the fusion analysis and early warning unit 230 enables this system to function as an independent safety monitoring unit, or to seamlessly integrate the assessment results into the vehicle control system or cloud big data platform, demonstrating excellent modularity and integration.

[0122] The 200 battery system status assessment and early warning system based on multi-source information fusion is independent of traditional battery management systems, or serves as their safety coprocessor. It provides a ready-to-use, proactive system-level safety situation awareness solution for various battery application scenarios such as electric vehicles and energy storage power stations.

[0123] Exemplary control terminal In one exemplary embodiment of this application, a vehicle is also provided, see [link to example]. Figure 3 , Figure 3 This application provides a schematic diagram of a vehicle architecture, the vehicle including: a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the steps in the battery system state assessment and early warning method based on multi-source information fusion according to various embodiments of this application as described in the above embodiments.

[0124] The vehicle includes a processor, memory, network interface, and input devices connected via a system bus. The vehicle's processor provides computing and control capabilities. The vehicle's memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The vehicle's network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it follows the steps of the battery system state assessment and early warning method based on multi-source information fusion according to various embodiments of this application, as described in the above embodiments.

[0125] The processor may include the main processor, as well as baseband chips, modems, etc.

[0126] It is understood that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly implemented by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0127] It is understood that the memory in the embodiments of this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Specifically, non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM). It should be noted that the memory in the devices and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0128] Input devices may include devices that receive data and information input by the user, such as keyboards, mice, cameras, scanners, light pens, voice input devices, touch screens, pedometers, or gravity sensors.

[0129] Output devices may include devices that allow information to be output to the user, such as displays, printers, speakers, etc.

[0130] The communication interface may include any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, Radio Access Network (RAN), Wireless Local Area Network (WLAN), etc.

[0131] The vehicle may also include a display component and a voice component. The display component may be an LCD screen or an e-ink screen. The vehicle's input device may be a touch layer covering the display component, or a button, trackball, or touchpad set on the vehicle body, or an external keyboard, touchpad, or mouse, etc.

[0132] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the vehicle to which the present application is applied. A specific vehicle may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0133] Exemplary computer program products and storage media In addition to the methods and devices described above, the battery system state assessment and early warning method based on multi-source information fusion provided in the embodiments of this application can also be a computer program product, which includes computer program instructions. When the computer program instructions are run by a processor, the processor performs the steps in the battery system state assessment and early warning method based on multi-source information fusion according to various embodiments of this application as described in the "Exemplary Methods" section above.

[0134] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0135] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this application. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0136] Furthermore, embodiments of this application also provide a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor of the steps in the battery system state assessment and early warning method based on multi-source information fusion according to various embodiments of this application as described in the "Exemplary Methods" section above.

[0137] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0138] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0139] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the solutions provided in the embodiments of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for battery system state assessment and early warning based on multi-source information fusion, characterized in that, The method includes: The system synchronously collects multi-source state data of the battery system, including at least a set of real-time synchronous parameters for collaboratively assessing the risks of mechanical integrity and environmental interaction. The set of real-time synchronous parameters includes: vibration acceleration data, volatile organic compound concentration data, humidity data, and corrosion rate data. The data of the real-time synchronization parameter group is processed to extract the combined feature parameters that are related to each other, forming a feature vector. The combined feature parameters include at least: structural risk features generated based on vibration acceleration time series data to characterize structural impact and fatigue accumulation; sealing risk features generated based on the synergistic change trend of volatile organic compound concentration and humidity to characterize electrolyte leakage; and connection risk features generated based on corrosion rate data to characterize the deterioration of electrical connection point conditions. The feature vector is input into a preset multi-parameter fusion analysis model to analyze the linkage mode between the combined feature parameters. Based on the analysis results of the linkage mode, a safety risk report is generated and output simultaneously. The safety risk report includes at least a mechanical integrity risk index and an environmental interaction risk index.

2. The method according to claim 1, characterized in that, In the step of synchronously acquiring multi-source state data of the battery system, the acquired real-time synchronous parameter set also includes air pressure data for assessing depressurization and shell integrity, oxygen concentration data for assessing fire risk, and collision trigger signal for directly determining physical collision events.

3. The method according to claim 1 or 2, characterized in that, The steps for synchronously acquiring multi-source state data of the battery system specifically include: Data collected by various sensors deployed at key locations in the battery pack or module is first aggregated to the cell monitoring controller in the corresponding battery box for preliminary processing, and then converged to the multi-dimensional data fusion controller via wired or wireless communication.

4. The method according to claim 2, characterized in that, The extraction of related combined feature parameters to form a feature vector specifically includes: extracting the feature values ​​of pressure change rate, specific gas concentration of volatile organic compounds, AC impedance phase angle, average expansion force, vibration dominant frequency energy, corrosion current, humidity value, oxygen concentration, and the collision trigger signal, and combining the feature values ​​to form the feature vector.

5. The method according to claim 4, characterized in that, The preset multi-parameter fusion analysis model includes an electrolyte leakage judgment model; the electrolyte leakage judgment model is configured to execute the following judgment logic: If the following conditions are met simultaneously: The concentration of volatile organic compounds in the feature vector exceeds a first preset threshold. The humidity feature value in the feature vector exceeds a second preset threshold. The magnitude of the change in the air pressure feature value in the feature vector is lower than the third preset threshold. The eigenvalue of the AC impedance ohmic component in the eigenvector exceeds a fourth preset threshold. This generates a risk assessment result for seal failure, indicating that the electrolyte has leaked and has come into contact with air.

6. The method according to claim 1 or 2, characterized in that, The method also includes a collision probability assessment step: based on vibration acceleration sampling data used to capture impact spectrum characteristics, the collision damage probability index is calculated and output in real time using a built-in impact response spectrum algorithm and material damage model.

7. The method according to claim 1, characterized in that, The analysis of the linkage pattern between the combined feature parameters specifically includes: The multi-parameter fusion analysis model is used to analyze the correlation and evolution trend among the structural risk characteristics, the sealing risk characteristics, and the connection risk characteristics. The synchronous generation and output of the safety risk report specifically includes: generating assessment results corresponding to the mechanical integrity status and environmental interaction risk status of the battery system, respectively, based on the analysis of the correlation and evolution trend.

8. The method according to claim 1, characterized in that, Following the step of synchronously generating and outputting a security risk report, the method further includes a security action execution step: based on the security risk report, triggering and executing corresponding proactive security measures, which include at least one of activating the ventilation system, isolating faulty units, and notifying the maintenance platform.

9. A battery system state assessment and early warning system based on multi-source information fusion, characterized in that, The system for implementing the method as described in any one of claims 1 to 8 comprises: The sensor array, including a vibration acceleration sensor, an organic volatile concentration sensor, a humidity sensor, and a corrosion rate sensor, is used to acquire data from the real-time synchronization parameter group; The data acquisition and processing unit is configured to synchronously acquire data from the sensor array and process the data to extract the combined feature parameters to form a feature vector; The fusion analysis and early warning unit is configured to receive the feature vector, analyze the linkage pattern between the combined feature parameters through a preset multi-parameter fusion analysis model, and generate a security risk report based on the analysis results. In addition, a controller area network is used to connect the sensor array, the data acquisition and processing unit, and the fusion analysis and early warning unit to realize data transmission and command interaction.

10. The system according to claim 9, characterized in that, The deployment method of each sensor in the sensor array is specifically configured as follows: The vibration acceleration sensor is deployed on the key load-bearing structure of the battery pack housing and internal module support of the battery system. The volatile organic compound concentration sensor is deployed in the air chamber of the battery pack and in the air circulation area of ​​the battery compartment. The humidity sensor is deployed inside the battery module or near the connector of the high-voltage sampling harness; The corrosion rate sensor is attached to the surface of the high-voltage electrical connector within the battery system. Furthermore, the system also includes an expansion force / strain sensor, which is attached to the side plate or end plate of the battery module, or clamped between adjacent cells.

Images