Lithium ion battery thermal runaway multi-stage early warning method and system based on optical fiber sensing

Abstract

This invention belongs to the technical field of lithium-ion battery thermal runaway early warning. It proposes a multi-level early warning method and system for lithium-ion battery thermal runaway based on fiber optic sensing. The method uses fiber optic gratings to measure temperature and strain signals at different locations within the lithium-ion battery, achieving high spatial resolution acquisition of multi-dimensional physical quantities to characterize changes in battery thermal behavior. The constructed graph nodes include global and local nodes. Local nodes contain surface temperature and strain information, while global nodes contain current and voltage information. By modeling the temporal and spatial characteristics of the battery's measurement parameters, the spatiotemporal evolution of internal thermal diffusion is learned to obtain an estimated internal temperature of the lithium-ion battery. In the early warning determination stage, multi-dimensional decision quantities such as internal temperature, temperature rise rate, and strain information are introduced. Early warning decisions are made through a threshold ratio and a continuous time window mechanism, thereby improving the accuracy and reliability of battery thermal runaway early warning.

Description

Technical Field

[0001] This invention belongs to the technical field of early warning of thermal runaway in lithium-ion batteries, and particularly relates to a multi-level early warning method and system for thermal runaway in lithium-ion batteries based on fiber optic sensing. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Lithium-ion batteries have been widely used in various energy storage systems, especially in the field of new energy, due to their high energy density, long cycle life, and clean and environmentally friendly properties.

[0004] However, as a crucial component of new energy vehicles, battery thermal runaway has frequently caused safety issues. Battery thermal runaway refers to a chain reaction process triggered by a series of internal and external factors such as electrical abuse, thermal abuse, and mechanical abuse during battery operation. It generally manifests as an imbalance between heat generation and dissipation within the battery, leading to a continuous rise in temperature and ultimately resulting in battery structural damage, fire, or even explosion.

[0005] Currently, research on battery safety management mainly relies on monitoring battery electrical signals and surface temperature, determining whether thermal runaway has occurred based on whether the temperature signal exceeds a threshold. However, in modern multi-mode signal interaction scenarios, relying solely on electrical characteristics or surface temperature is insufficient to accurately capture the changing patterns of the battery's internal state. This is particularly true in battery thermal runaway early warning tasks, where the lag between surface temperature and internal temperature changes fails to meet the characteristic information requirements for early warning. By the time the surface temperature exceeds a preset threshold, the warning time may be insufficient. Furthermore, relying primarily on temperature results at a single point in time—determining a risk of thermal runaway when the temperature exceeds a preset threshold at a certain moment—ignores the dynamic evolution of the thermal runaway process over time. In practical applications, this can easily lead to false alarms for normal batteries due to sensor noise, changes in operating conditions, and other factors. Summary of the Invention

[0006] To overcome the shortcomings of the prior art, this invention provides a multi-level early warning method and system for thermal runaway of lithium-ion batteries based on fiber optic sensing. It achieves high spatial resolution acquisition of multi-dimensional physical quantities of the battery to characterize changes in battery thermal behavior, and designs early warning rules based on spatial probability and temporal continuity to improve the early warning lead time while reducing false alarms and missed alarms.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing, comprising: Temperature and strain signals at different locations of a lithium-ion battery are measured using fiber optic gratings, and voltage and current signals of the lithium-ion battery are also acquired. Different locations of the lithium-ion battery are modeled as local nodes, and the temperature and strain signals at different locations are used as the node features of the corresponding local nodes. The voltage and current signals of the lithium-ion battery are modeled as the node features of the global nodes to construct the battery physical map. For the battery physical diagrams corresponding to different time windows, the temporal and spatial characteristics of the battery measurement parameters are modeled using an internal temperature estimation model. The internal temperature of the lithium-ion battery is estimated by learning the spatiotemporal evolution law of internal heat diffusion. Based on the estimated internal temperature and temperature rise rate of the lithium-ion battery corresponding to the future time window, as well as the strain change rate of the current time window, a multi-level early warning system for thermal runaway of lithium-ion batteries is implemented.

[0008] Secondly, the present invention provides a multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing, comprising: The fiber optic sensing module is configured to measure temperature and strain signals at different locations of the lithium-ion battery via a fiber optic grating, and to acquire voltage and current signals of the lithium-ion battery. The graph construction module is configured to: model different locations of the lithium-ion battery as local nodes, use the temperature and strain signals at different locations as node features of the corresponding local nodes, and model the voltage and current signals of the lithium-ion battery as node features of the global nodes to construct a battery physical graph. The internal temperature estimation module is configured to: model the temporal and spatial characteristics of the battery measurement parameters using the internal temperature estimation model for the battery physical diagrams corresponding to different time windows, and realize the internal temperature estimation of the lithium-ion battery by learning the spatiotemporal evolution law of internal heat diffusion. The multi-level early warning module is configured to provide multi-level early warning of thermal runaway of lithium-ion batteries based on the estimated internal temperature and temperature rise rate of the lithium-ion battery corresponding to the future time window, as well as the strain change rate of the current time window.

[0009] Thirdly, the present invention provides an electronic device including a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the method described in the first aspect.

[0010] Fourthly, the present invention provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the method described in the first aspect.

[0011] Fifthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the method described in the first aspect.

[0012] The above one or more technical solutions have the following beneficial effects: This invention measures temperature and strain signals at different locations in a lithium-ion battery using fiber optic gratings, achieving high spatial resolution acquisition of multidimensional physical quantities of the battery to characterize changes in its thermal behavior. The constructed graph nodes include global and local nodes; local nodes contain surface temperature and strain information, while global nodes contain current and voltage information. By modeling the temporal and spatial characteristics of the battery's measurement parameters, the spatiotemporal evolution of internal thermal diffusion is learned to obtain an estimated internal temperature of the lithium-ion battery. In the early warning judgment stage, multidimensional decision quantities such as internal temperature, temperature rise rate, and strain information are introduced. Early warning decisions are made through a threshold ratio and a continuous time window mechanism, changing from "point" triggering to "surface" judgment. This solves the problems of high false alarm rate and insufficient early warning time caused by relying on surface temperature signals at a single time point for early warning judgment, thereby improving the accuracy and reliability of battery thermal runaway early warning.

[0013] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0014] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0015] Figure 1 This is a block diagram of a multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing in an embodiment of the present invention. Figure 2 This is a schematic diagram of the fiber optic sensing arrangement in an embodiment of the present invention; Figure 3 This is a block diagram of a multi-level early warning system for thermal runaway of lithium-ion batteries based on fiber optic sensing, as described in an embodiment of the present invention. Detailed Implementation

[0016] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0017] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0018] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0019] Example 1 like Figures 1-2 As shown, this embodiment discloses a multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing, including: Temperature and strain signals at different locations of a lithium-ion battery are measured using fiber optic gratings, and voltage and current signals of the lithium-ion battery are also acquired. Different locations of the lithium-ion battery are modeled as local nodes, and the temperature and strain signals at different locations are used as the node features of the corresponding local nodes. The voltage and current signals of the lithium-ion battery are modeled as the node features of the global nodes to construct the battery physical map. For the battery physical diagrams corresponding to different time windows, the temporal and spatial characteristics of the battery measurement parameters are modeled using an internal temperature estimation model. The internal temperature of the lithium-ion battery is estimated by learning the spatiotemporal evolution law of internal heat diffusion. Based on the estimated internal temperature and temperature rise rate of the lithium-ion battery corresponding to the future time window, as well as the strain change rate of the current time window, a multi-level early warning system for thermal runaway of lithium-ion batteries is implemented.

[0020] In this embodiment, to address the problem that traditional methods rely on a single monitoring signal and cannot comprehensively characterize the evolution of battery thermal runaway, fiber optic sensing technology is introduced to achieve high spatial resolution acquisition of multi-dimensional physical quantities of the battery, which is used to characterize changes in battery thermal behavior. To address the issues of false alarms, missed alarms, and limited warning time caused by single-time-point determination methods, a warning rule based on spatial probability and temporal continuity is designed, changing "point" triggering to "surface" decision-making, increasing the warning lead time while reducing false alarms and missed alarms.

[0021] The following is a detailed description of the multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing proposed in this embodiment: To address the issues of traditional temperature and pressure sensors being susceptible to electromagnetic interference and limited to single-point measurements, this study utilizes the advantages of fiber optic gratings (FBGs) for pure optical signal transmission, which is unaffected by electromagnetic fields, supports multi-point measurements, and offers high sensitivity. Two optical fibers, each with eight gratings, are respectively attached to one front and one side of the pouch lithium battery used in the experiment. The optical fibers are made of high-temperature resistant materials to ensure effective monitoring under high-temperature conditions during battery thermal runaway.

[0022] Fiber Bragg gratings are positioned along the central axis of the front and sides of the battery. Based on previous experiments and literature review, the central axis region of a pouch cell is typically the location of internal heat accumulation and diffusion. Eight gratings are used to form a centimeter-level multi-point sensing field to characterize the battery's thermal features. Due to the high contact resistance near the tabs, which easily generates initial temperature rise points, one grating is placed close to the battery tab. This mounting scheme fully considers the initial temperature rise location and heat source accumulation location during battery operation. The front and side mounting design can simultaneously reflect the heat transfer characteristics along the battery's length and thickness directions, achieving high spatial resolution temperature monitoring. This information allows for more accurate estimation of internal temperature changes, improving the sensitivity and reliability of thermal runaway early warning, while reducing monitoring of ineffective areas and improving signal utilization efficiency.

[0023] To obtain the internal temperature distribution of the battery, a batch of pouch cells were selected and flexible micro-optical fibers were embedded in the gaps between the internal electrodes during manufacturing. This data was used as label data for subsequent training of the internal temperature prediction model. This batch of battery data was used only for training the internal temperature prediction model. After the model was trained, only external monitoring data was needed to provide early warning of battery thermal runaway. All fiber optic ports were connected to the fiber optic demodulator, and this process should minimize fiber bending and avoid fiber entanglement.

[0024] Overcharge thermal runaway experiments were conducted on the tested soft-pack batteries to induce thermal runaway behavior. The battery parameters such as current and voltage were recorded simultaneously throughout the process, with a sampling frequency of 1 Hz.

[0025] Based on the arrangement of the fiber optic sensing module described above, the fiber optic grating in this embodiment is based on the Bragg reflection principle.

[0026] The wavelength shift of each grating reflection peak is detected by measuring the reflection spectrum in real time using an optical fiber demodulator. Different optical signals are converted into changes in battery temperature and strain.

[0027] Specifically, the wavelength change of each grating can be expressed as:

[0028] in, This indicates a change in the grating wavelength. Indicates the temperature sensitivity coefficient. Indicates the strain sensitivity coefficient. and These represent temperature change and strain change, respectively.

[0029] In this embodiment, the battery's current, voltage, surface temperature, and strain signals are simultaneously acquired during the overcharge thermal runaway experiment. The surface temperature and strain are obtained by demodulation using a Bragg fiber grating sensing system, with a sampling frequency of 1 Hz. To ensure the integrity and consistency of the data, the experimentally acquired signal data undergoes systematic preprocessing, including data storage, time synchronization, and missing value imputation.

[0030] Experimental data were recorded in real time by a demodulator and testing system. The optical signals were demodulated into surface temperature and strain information, which were then stored as a time series along with current and voltage. Each fiber grating simultaneously acquired the temperature and strain signals of the battery surface; one fiber contained eight gratings, allowing for the acquisition of eight sets of measurement data. Therefore, eight sets of temperature and strain signals were obtained for each battery from both the front and side views. For the training dataset, internal battery temperature information was additionally acquired for label values ​​in the subsequent internal temperature prediction model.

[0031] Data loss may occur during signal transmission and sampling. For data loss at certain time points, the missing data rate is first calculated based on the sampling frequency. If there is no missing data, no processing is performed on the time series; if the missing rate is less than 2%, linear interpolation is used to supplement the missing data along the time dimension.

[0032] This embodiment uses a graph convolutional long short-term memory (GCN-LSTM) network designed based on fiber optic sensing signals and electrical signals as the internal temperature estimation model. It employs a sequence-to-sequence multi-step prediction framework, using current, voltage, strain, and surface temperature within a sliding time window as inputs. The GCN-LSTM learns the spatial and temporal interaction between physical characteristics and internal temperature, predicting the future trend of battery internal temperature changes, thus achieving reliable estimation of the battery's internal temperature. The GCN-LSTM mainly consists of two parts: a graph convolutional neural network and a long short-term memory network.

[0033] In battery thermal runaway experiments, fiber Bragg grating sensors are distributed at multiple points on the battery surface to simultaneously acquire temperature and strain signals. The temperature and strain at different locations are influenced by factors such as the internal thermal diffusion path and local electrochemical reactions within the battery. Therefore, the data collected from different sensing points reflects the spatial relationships between features, exhibiting clear spatial location relationships and thermal conduction correlations. Traditional convolutional neural networks and recurrent neural networks cannot learn the spatial relationships between data. However, graph convolutional neural networks (GCNs) perform feature propagation and fusion on non-Euclidean structured data, modeling the relationships between nodes through adjacency matrices, enabling efficient learning of spatial dependencies. To obtain the spatial information implicit in the data, GCNs are used to capture the spatial dependencies of experimental data based on physical location.

[0034] In GCN, data is represented in the form of graphs, and a graph is represented as... ,in Represents a set of nodes. Represents the set of edges. Node features are denoted as a matrix. ,in For the number of nodes, Each node represents a feature dimension. The connection relationships between nodes are represented by an adjacency matrix. This indicates that if the node and If they are interconnected, then Otherwise, it is 0.

[0035] The graph nodes constructed in this embodiment include global nodes and local nodes. Each fiber optic sensor measurement point is defined as a local node, referred to as a sensing node. Voltage and current are global features, therefore electrical signals are set as separate global nodes. Global nodes are connected to all sensing nodes. Node features represent the local time series of that feature within the current time period. Sensing nodes contain surface temperature and strain information, while global nodes contain current and voltage information.

[0036] Specifically, firstly, a sliding time window is used to divide the preprocessed time series data into several time segments, each segment containing multidimensional features over a period of time. Then, the K-Nearest Neighbors (KNN) algorithm is used to measure the feature similarity between sensing nodes based on Euclidean distance. With nodes The feature vectors are respectively , Then the distance between the two is defined as:

[0037] Sort each sensor node by its Euclidean distance from other sensor nodes in ascending order. When the node... Belongs to node When the K nearest neighbors are found, connect them in the graph. and Simultaneously, the global node is connected to all sensor nodes to fuse global electrical signals with local fiber optic sensing information. The connection relationships of all nodes are represented by an adjacency matrix. It means that, among them:

[0038] The KNN algorithm is used to obtain a set V of 17 nodes related to the physical parameters of the battery, and an adjacency matrix A (17×17) reflecting their spatial relationships.

[0039] Then, each node is self-connected:

[0040] Where A represents the graph structure after all nodes are connected, which is mathematically represented by an adjacency matrix; express The identity matrix; This indicates that each node is connected to itself in addition to its neighbors. This is represented by setting the main diagonal of the adjacency matrix to 1. Mathematically, this is A plus the identity matrix. The subscript N represents the number of nodes, 17. Node self-connection helps preserve the characteristics of each node during information updates, while also ensuring computational stability during matrix normalization.

[0041] Perform symmetric normalization:

[0042] in, Represents a diagonal matrix, with diagonal elements. Define as a node The degree, that is .

[0043] Finally, GCN performs convolution operations on the constructed battery physical graph structure to extract and learn the spatial relationships between features. The calculation formula for each GCN layer is as follows:

[0044] In the above formula, Indicates the current number The node feature matrix of the layer; This represents the learnable weights of this layer; The normalized graph adjacency matrix controls the direction and intensity of information diffusion. This is a nonlinear activation function that introduces nonlinear characteristics into the model, thereby improving its nonlinear mapping capability. In this example, only the node features are dynamically updated as the time window slides, thus enabling the modeling of battery thermal behaviors such as thermal diffusion.

[0045] Through the above operations, the Graph Convolutional Neural Network (GCN) constructs a graph adjacency matrix based on spatial distance and physical similarity for battery features. This explicitly integrates the battery's geometry, thermal coupling characteristics, and the distribution of measurement points into the model structure, giving the model a certain degree of physical interpretability. Simultaneously, each sensing node fuses information from neighboring nodes and global nodes during feature updates. After multi-layer GCN computation, the output features of each node contain both global and local information.

[0046] Next, the output of each time window processed by GCN is flattened and used as the input to the LSTM at the corresponding time step. The LSTM calculation process is as follows:

[0047] in, , , These represent the forget gate, input gate, and output gate, respectively. In a hidden state, it can remember the temporal relationships between data; This indicates the hidden state of the input at the current moment; Indicates spatial feature input; The output is determined by both the output gate and the hidden state. , , , , , , As weight, , Indicates bias. Using the Sigmoid activation function, the LSTM determines how much information to retain. Its unique gating mechanism effectively learns the temporal dependence of the battery's physical spatial features. The LSTM output represents the dynamic pattern of the battery's thermal runaway process over time. Finally, the LSTM output is fed into a fully connected layer (FC), which maps the high-dimensional spatiotemporal features learned by the GCN-LSTM to the predicted internal temperature for the corresponding future time window. The overall process can be simplified as follows:

[0048] This embodiment fully considers the spatial distribution characteristics of the sensors in the experiment and designs a GCN-LSTM model based on graph convolution and long short-term memory. By modeling the temporal and spatial characteristics of battery measurement parameters, it learns the spatiotemporal evolution law of battery thermal diffusion, thereby improving the accuracy of battery internal temperature estimation.

[0049] This embodiment uses a battery dataset with internal fiber optic sensing to train an internal temperature estimation model. The dataset is divided into a training set and a validation set. By comparing the internal temperature prediction results of the validation set with their corresponding label data, the number of layers in each module, as well as parameters such as learning rate, number of iterations, K-neighbor range, and time window, are adjusted.

[0050] For the predicted internal temperature results for each time window, the temperature rise rate at each time point within that window is calculated in real time. It serves as an auxiliary indicator for early warning and judgment of thermal runaway.

[0051] Multi-level early warning is the core component of this embodiment and is crucial for achieving early warning of battery thermal runaway. This module comprehensively utilizes predicted internal temperature, temperature rise rate, and current strain information as the basis for thermal runaway early warning decisions. The battery thermal runaway early warning method proposed in this embodiment mainly relies on two states: the current state and the future state.

[0052] The internal temperature and rate of temperature rise of a battery are defined as the future state and are an essential reflection of the battery's thermal behavior. When a battery experiences thermal runaway, it manifests as an internal temperature imbalance. Therefore, using the internal temperature and rate of temperature rise within a future time window as a warning basis is crucial for achieving early warning. Furthermore, battery thermal reactions follow physical laws such as the law of conservation of energy and the thermal diffusion equation, and battery temperature exhibits a traceable evolutionary trend, making the prediction of the battery's internal temperature physically reasonable.

[0053] Strain signals are the immediate responses of battery chemical reactions and internal structures. Significant strain changes occur when side reactions, gas generation, or structural deformation occur within the battery; however, weak reactions may not produce strain. Therefore, strain is not a continuously evolving state, but rather an observed variable triggered by multiple events, lacking a stable temporal evolution pattern. This embodiment does not predict future strain signals but defines the strain signal within the current time window as the current state, serving as an auxiliary early warning feature reflecting changes in the battery's internal structure and gas generation behavior. This feature, together with the future state, constitutes a multi-level early warning decision-making basis.

[0054] Throughout the multi-level early warning decision-making process, the predicted internal temperature and its rate of temperature rise serve as the primary benchmark for thermal runaway early warning, used to determine the level of thermal runaway and its evolution stage. Strain signals, observed in real-time within the current time window, participate in the judgment as auxiliary criteria reflecting changes in the battery's internal structure and gas generation behavior. They complement the temperature information of the future state, effectively improving the accuracy and reliability of early warning of battery thermal runaway.

[0055] Specifically, this embodiment abandons the thermal runaway early warning method that relies on a single point in time for judgment. In response to the problem of high false alarm rate of traditional single-point judgment, it proposes a battery thermal runaway early warning method based on multi-level judgment. It expands the single feature at a single moment into a multi-variable spatial probability, uses the proportion of samples exceeding the threshold and continuous time window features as the decision basis, and combines the temperature rise rate and strain information to achieve auxiliary early warning.

[0056] The alarm for the corresponding level of the multi-level early warning strategy will be triggered when the following conditions are met: 1. When for K1 consecutive windows, the temperature p1% is greater than the temperature threshold T1 and the temperature rise rate reaches the temperature rise threshold R1, or the proportion of measuring points whose strain change rate exceeds the threshold S1 within the current time window reaches a certain threshold. Furthermore, when the rate of temperature rise reaches the temperature rise threshold R1, a level one alarm is triggered, indicating that the battery has begun to heat up abnormally and there are initial structural changes, which may pose a potential risk of thermal runaway and require continuous monitoring.

[0057] 2. When for K2 consecutive windows, the temperature p2% is greater than the temperature threshold T1 and the temperature rise rate reaches the temperature rise threshold R2, and the proportion of measuring points whose strain change rate exceeds the threshold S1 within the current time window reaches a certain threshold. If a level 2 alarm is triggered, it indicates that the battery is experiencing internal thermal diffusion, accompanied by structural expansion or gas generation. In this case, it is necessary to reduce the battery load and have the BMS investigate the cause of the fault.

[0058] 3. When for K3 consecutive windows, the temperature at p2% is greater than the temperature threshold T2 and the temperature rise rate reaches the temperature rise threshold R3, and the proportion of measuring points whose strain change rate exceeds the threshold S2 within the current time window reaches a certain threshold. The system triggers a Level 3 alarm, indicating that the battery is undergoing significant structural changes while its temperature continues to rise. At this point, the battery thermal runaway has entered a critical stage, requiring the activation of safety protection and other preventative measures.

[0059] In this multi-level early warning strategy, the threshold was determined through thermal runaway experimental results, battery characteristic analysis, and extensive literature review. The threshold condition satisfies K1. <K2<K3,p1<p2,T1<T2,R1<R2<R3,S1<S2, < Different thresholds are used to distinguish different stages in the thermal runaway evolution process of a battery.

[0060] In the early stages, localized changes in the battery structure and an initial temperature rise occur, but the duration is short. At this time, a lower temperature threshold T1, strain threshold S1, and a smaller duration window K1 are used for judgment, enabling a rapid response to potential anomalies. After entering the thermal diffusion stage, the temperature rise range and strain intensity expand, while the duration lengthens. Therefore, a higher temperature probability P2 and strain ratio are used. The determination is made using a longer duration window K2; in the critical stage of thermal runaway, the temperature continues to rise, the rate of temperature rise continues to increase, and the battery strain changes significantly, using a threshold combination K3, R3, T2, S2. Enables real-time alarms.

[0061] Based on the predicted internal temperature and the strain changes over the current time, various decision-making indicators are calculated in real time, and the statistical results of continuous time windows are recorded. When each indicator is detected to have reached the preset conditions, the corresponding result and warning level are output. This embodiment abandons the traditional thermal runaway early warning method that relies on a single time point and a single variable for judgment. The designed multi-level early warning strategy for batteries fully considers the physical evolution law of battery thermal runaway. It uses fiber optic sensing technology to collect multi-dimensional signals during the battery thermal runaway process. In the early warning judgment stage, it introduces multi-dimensional decision quantities such as internal temperature, temperature rise rate, and strain information, changing "point" triggering to "surface" judgment. Combining the future state and the current state, it achieves step-by-step identification from local anomalies to critical risks through multi-temporal state analysis of the predicted temperature exceeding the threshold ratio, the evolution characteristics of continuous time windows, and the current strain anomaly ratio. While ensuring the early warning, it effectively avoids the problem of false alarms caused by a single anomaly point and the problem of missed alarms triggered only under extreme conditions.

[0062] Test results show that the early multi-level warning method for lithium batteries designed based on this embodiment can achieve stable and accurate risk assessment before the arrival of each key stage in the physical evolution of battery thermal runaway, with an average warning lead time of 17 minutes.

[0063] One example is that the Level 1 alarm responds 952 seconds before the lithium-ion battery self-heats, the Level 2 alarm responds 1183 seconds before pressure relief, and the Level 3 alarm responds 997 seconds before the thermal runaway critical time (the moment when the self-heating rate exceeds 60 °C / min in the experiment). Compared with traditional thermal runaway early warning methods that rely solely on a single signal threshold, the method proposed in this invention significantly improves both the accuracy and stability of the early warning.

[0064] Meanwhile, the spatial probability-based judgment rule, while maintaining a high lead time for early warning, did not result in significant or frequent false alarms.

[0065] In this embodiment, during the early warning determination stage, by combining the future state and the current state, and through multi-temporal state analysis of the predicted temperature exceeding the threshold ratio, the evolution characteristics of the continuous time window, and the current strain anomaly ratio, the step-by-step identification from local anomalies to critical risks is achieved. This ensures the early warning capability while effectively avoiding false alarms caused by a single anomaly point and missed alarms triggered only under extreme conditions.

[0066] Example 2 like Figure 3 As shown, the purpose of this embodiment is to provide a multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing, including: The fiber optic sensing module is configured to measure temperature and strain signals at different locations of the lithium-ion battery via a fiber optic grating, and to acquire voltage and current signals of the lithium-ion battery. The graph construction module is configured to: model different locations of the lithium-ion battery as local nodes, use the temperature and strain signals at different locations as node features of the corresponding local nodes, and model the voltage and current signals of the lithium-ion battery as node features of the global nodes to construct a battery physical graph. The internal temperature estimation module is configured to: model the temporal and spatial characteristics of the battery measurement parameters using the internal temperature estimation model for the battery physical diagrams corresponding to different time windows, and realize the internal temperature estimation of the lithium-ion battery by learning the spatiotemporal evolution law of internal heat diffusion. The multi-level early warning module is configured to provide multi-level early warning of thermal runaway of lithium-ion batteries based on the estimated internal temperature and temperature rise rate of the lithium-ion battery corresponding to the future time window, as well as the strain change rate of the current time window.

[0067] In further embodiments, the following is also provided: An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When executed by the processor, the computer instructions perform the method described in Embodiment 1. For brevity, further details are omitted here.

[0068] It should be understood that in this embodiment, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0069] Memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of memory may also include non-volatile random access memory. For example, memory may also store information about the device type.

[0070] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, perform the method described in Embodiment 1.

[0071] The method in Embodiment 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The software modules can reside in readily available storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. 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. To avoid repetition, a detailed description is not provided here.

[0072] A computer program product includes a computer program that, when executed by a processor, implements the method described in Embodiment 1.

[0073] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which execute in a device on a target real or virtual processor to perform the processes / methods described above. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided among program modules as needed. The machine-executable instructions for the program modules can execute within a local or distributed device. In a distributed device, the program modules can reside in both local and remote storage media.

[0074] The computer program code used to implement the methods of the present invention may be written in one or more programming languages. This computer program code may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the computer or other programmable data processing device, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a computer, partially on a computer, as a stand-alone software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0075] In the context of this invention, computer program code or related data may be carried by any suitable carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals may include electrical, optical, radio, sound, or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0076] Those skilled in the art will recognize that the units and algorithm steps described in conjunction with the embodiments herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0077] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing, characterized in that, Including: Measuring the temperature signals and strain signals at different positions of the lithium-ion battery through fiber Bragg gratings, and acquiring the voltage signal and current signal of the lithium-ion battery; Modeling different positions of the lithium-ion battery as local nodes, taking the temperature signals and strain signals at different positions as the node features of the corresponding local nodes, modeling the voltage signal and current signal of the lithium-ion battery as the node features of the global node, and constructing a battery physical graph; For the battery physical graphs corresponding to different time windows, using an internal temperature estimation model to model the temporal and spatial characteristics of the battery measurement parameters, realizing the internal temperature estimation of the lithium-ion battery by learning the spatio-temporal evolution law of internal heat diffusion in the battery, and calculating the temperature rise rate; Based on the estimated internal temperature and temperature rise rate of the lithium-ion battery corresponding to the future time window, and the strain change rate of the current time window, realizing multi-level early warning of thermal runaway of the lithium-ion battery.

2. The multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing as described in claim 1, characterized in that, For the battery physical graphs corresponding to different time windows, using an internal temperature estimation model to model the temporal and spatial characteristics of the battery measurement parameters, realizing the internal temperature estimation of the lithium-ion battery by learning the spatio-temporal evolution law of internal heat diffusion in the battery, specifically: Sorting the Euclidean distances between each local node and other local nodes from small to large, and establishing a connection between the first local node and the second local node when the second local node is in the set of K nearest neighbor nodes of the first local node, and each local node is connected to the global node to fuse the global electrical signal and local fiber optic sensing information; Representing the connection relationship of each node through an adjacency matrix, and calculating the normalized graph adjacency matrix; Extracting and learning the spatial relationship between features based on the graph adjacency matrix using a graph convolutional neural network; Taking the output of each time window processed by the graph convolutional neural network as the input of the long short-term memory network with the corresponding time step length, and using a fully connected layer to map the learned high-dimensional spatio-temporal features to the internal temperature prediction result corresponding to the future time window.

3. The multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing as described in claim 1, characterized in that, Based on the estimated internal temperature and temperature rise rate of the lithium-ion battery corresponding to the future time window, and the strain change rate of the current time window, realizing multi-level early warning of thermal runaway of the lithium-ion battery, specifically: When the temperature of p1% is greater than the first temperature threshold and the temperature rise rate reaches the first temperature rise threshold for K1 consecutive time windows, triggering a first-level alarm; When the temperature of p2% is greater than the first temperature threshold and the temperature rise rate reaches the second temperature rise threshold for K2 consecutive windows, and the proportion of measurement points where the strain change rate exceeds the first strain change rate threshold within the current time window reaches the second set ratio, triggering a second-level alarm; When the temperature of p2% is greater than the second temperature threshold and the temperature rise rate reaches the third temperature rise threshold for K3 consecutive windows, and the proportion of measurement points where the strain change rate exceeds the second strain change rate threshold within the current time window reaches the second set ratio, triggering a third-level alarm; where K1 < K2 < K3, p1% < p2%, the first temperature threshold < the second temperature threshold, the first temperature rise threshold < the second temperature rise threshold < the third temperature rise threshold, and the first strain change rate threshold < the second strain change rate threshold.

4. The multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing as described in claim 1, characterized in that, It also includes: If the proportion of measuring points whose strain rate of change exceeds the first strain rate of change threshold within the current time window reaches the first set proportion, and the temperature rise rate reaches the first temperature rise threshold, then a level one alarm is triggered; wherein, the first set proportion < the second set proportion.

5. The multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing as described in claim 1, characterized in that, The internal temperature of a lithium-ion battery is obtained by embedding flexible micro-optical fibers between the electrodes inside the battery, and this data is used as label data for training the internal temperature estimation model.

6. The multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing as described in claim 1, characterized in that, Fiber gratings were placed at the central axis positions on the front and side of the lithium-ion battery to measure temperature and strain signals at different locations on the battery.

7. A multi-level early warning method for thermal runaway of lithium-ion batteries based on fiber optic sensing, characterized in that, include: The fiber optic sensing module is configured to measure temperature and strain signals at different locations of the lithium-ion battery via a fiber optic grating, and to acquire voltage and current signals of the lithium-ion battery. The graph construction module is configured to: model different locations of the lithium-ion battery as local nodes, use the temperature and strain signals at different locations as node features of the corresponding local nodes, and model the voltage and current signals of the lithium-ion battery as node features of the global nodes to construct a battery physical graph. The internal temperature estimation module is configured to: model the temporal and spatial characteristics of the battery measurement parameters using the internal temperature estimation model for the battery physical diagrams corresponding to different time windows; estimate the internal temperature of the lithium-ion battery by learning the spatiotemporal evolution law of internal heat diffusion; and calculate the temperature rise rate. The multi-level early warning module is configured to provide multi-level early warning of thermal runaway of lithium-ion batteries based on the estimated internal temperature and temperature rise rate of the lithium-ion battery corresponding to the future time window, as well as the strain change rate of the current time window.

8. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, perform the method described in any one of claims 1-6.

10. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the method described in any one of claims 1-6.

Images