Method for predicting thermal runaway of energy storage lithium battery based on neural network reduced-order model

Abstract

The invention provides an energy storage lithium battery thermal runaway prediction method based on a neural network reduced-order model, and relates to the technical field of energy storage, and the method comprises the specific steps: S1, constructing a thermal runaway simulation model; s2, constructing an original data matrix; s3, model order reduction processing; s4, calculating thermal runaway stage time; according to the method, a complete lithium battery thermal runaway simulation model is formed by constructing a lithium battery electrochemical-heat dissipation simulation basic model containing electrochemistry and heat dissipation correlation and fusing a thermal runaway mechanism model, related items of thermal runaway reaction heat are accurately represented, and the calculation result is accurate in data processing. According to the method, eigenvalue decomposition is adopted to carry out dimension reduction on an original data matrix, regression is carried out in combination with a convolutional neural network, accurate mapping of working conditions and physical field results is established, and the model construction and data processing mode not only improves the calculation efficiency, but also ensures the accuracy of a prediction result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy storage, in particular to a thermal runaway prediction method for energy storage lithium batteries based on a neural network reduced-order model. BACKGROUND

[0002] With the wide application of high-energy-density energy storage lithium batteries in energy storage power stations, new energy grid-connected fields, etc., the high-density arrangement of battery modules and the coupling with complex working conditions lead to a sharp rise in thermal runaway risk. Energy storage lithium batteries are prone to thermal runaway under conditions such as internal short circuit, overcharge or overheating. Once it happens, it can cause multi-module cascade combustion within a few minutes, release a large amount of toxic and flammable gas, and be accompanied by explosive pressure surge, which seriously threatens personnel safety and equipment integrity. Therefore, the thermal runaway early warning of energy storage lithium batteries has important research value and practical significance in the field of energy storage equipment safety protection. At present, the thermal runaway prediction technology of energy storage lithium batteries mainly focuses on the passive protection mechanism based on threshold alarm, but this method has the problem of response lag and is difficult to effectively block the irreversible thermal runaway process. With the increasing requirements for the safety and reliability of energy storage systems, developing mechanism-based thermal runaway prediction technology and pre-warning time window have become a research hotspot in this field.

[0003] Traditional thermal runaway prediction technology for energy storage lithium batteries mainly relies on threshold alarm mechanism, i.e. triggering alarm when the battery temperature, voltage or current exceeds the preset threshold. However, this method has significant limitations. First, the threshold alarm mechanism belongs to passive protection, which can only issue an alarm after the thermal runaway occurs or enters the irreversible stage, and cannot predict in advance and take preventive measures; second, although the existing high-precision CFD simulation technology can simulate the temperature field evolution in the thermal runaway process by building an electrochemical-thermal-flow coupled model, it faces the problem of long simulation calculation time for multi-physical field simulation of full-size energy storage systems, which cannot support real-time warning requirements; in addition, the traditional method lacks dynamic prediction ability for the development process of thermal runaway, which makes it difficult to accurately determine the stages and remaining time of thermal runaway, thus limiting the effectiveness and timeliness of prevention and control measures. Therefore, the traditional technology has obvious shortcomings in the field of thermal runaway prediction of energy storage lithium batteries, and innovative technology is needed to break through. SUMMARY

[0004] In order to solve the technical problems existing in the prior art, the present application provides the following technical solutions: On the one hand, a thermal runaway prediction method for energy storage lithium batteries based on a neural network reduced-order model is provided, which comprises: S1, constructing a thermal runaway simulation model: first, an electrochemical-heat dissipation simulation basic model associated with electrochemistry and heat dissipation is established, then a thermal runaway mechanism model describing the characteristics of various chemical reactions in the thermal runaway process is constructed, the mechanism model is integrated into the electrochemical-heat dissipation simulation basic model to represent the related items of thermal runaway reaction heat, and a complete lithium battery thermal runaway simulation model is formed; S2, constructing an original data matrix: taking battery voltage, current and time as input, and taking battery time series temperature field data simulated by the thermal runaway simulation model as output, the input and output parameters are taken as target characteristic values, and an original data matrix for reduced order model training is constructed; S3, model reduction processing: performing eigenvalue decomposition on the original data matrix to complete data dimension reduction, combining a convolutional neural network to regress the reduced data and establish a working condition and physical field result mapping, and forming a preliminary reduced order model; S4, calculating the time of the thermal runaway stage: input the battery voltage and current collected on site into the preliminary reduced order model, perform time series calculation at a preset time step, read and discretize the storage time series temperature field simulation results; according to the preset thermal runaway stage warning threshold, take the highest temperature in the battery as the monitoring point, find the time corresponding to each threshold to obtain the arrival time of each stage of thermal runaway, and after delay compensation, perform inverse operation on the arrival time of each stage to obtain the simulation thermal runaway countdown time; S5, determining the final countdown: according to the measured temperature rise rate of the on-site temperature sensor, the measured thermal runaway countdown time is calculated by using the thermal runaway countdown formula; compare the simulation and measured thermal runaway countdown time, and select the smaller value as the final thermal runaway countdown time.

[0005] Further, the calculation formula of the electrochemical-heat dissipation simulation basic model in the S1, constructing a thermal runaway simulation model step is: ; ; , wherein, is the density of the battery material, is the specific heat capacity, is the battery temperature field, is time, is the material thermal conductivity, is the spatial differential operator, is the positive electrode effective conductivity, is the negative electrode effective conductivity, is the positive electrode potential distribution, is the negative electrode potential distribution, is the electrochemical reaction heat calculated by the electrochemical sub-model, is the Joule heat caused by internal short circuit, The heat of thermal runaway reaction activated only under conditions of thermal abuse. The current density of the electrochemical reaction. This represents the internal short-circuit current density. This represents the electrode current density.

[0006] Furthermore, in step S1, the calculation formula for the thermal runaway mechanism model in the step of constructing the thermal runaway simulation model is as follows: ; ; ; , in, It is a differential operator used to represent the instantaneous rate of change of a variable with respect to another variable; Time is used to characterize the time dimension in the reaction process; For dimensionless concentration characterization of the residual reactants in the SEI (solid electrolyte interface) decomposition reaction, This is a dimensionless concentration characterization of the remaining reactants in the negative electrode electrolyte reaction. For characterizing the dimensionless concentration of residual reactants in the electrolyte reaction. This is a dimensionless concentration characterization of the remaining reactants in the positive electrode electrolyte reaction; The pre-exponential factor for the SEI decomposition reaction. It is the pre-exponential factor for the negative electrode electrolyte reaction. It is the pre-exponential factor for the positive electrode electrolyte reaction. It is the pre-exponential factor for electrolyte reaction; The activation energy for the SEI decomposition reaction. The activation energy for the reaction in the negative electrode electrolyte is... This is the activation energy for the positive electrode electrolyte reaction. It is the activation energy of the electrolyte reaction; It is a universal gas constant. Battery temperature; The reaction order of the SEI decomposition reaction is given. This represents the reaction order of the negative electrode electrolyte reaction. , These are all reaction orders of the positive electrode electrolyte reaction. The reaction order of the electrolyte reaction; This is a dimensionless characterization of the SEI layer thickness. This is the reference thickness for the SEI layer.

[0007] Furthermore, in step S2, the input parameters for constructing the original data matrix, in addition to battery voltage, current and time, also include ambient temperature and cooling system flow rate parameters; wherein, ambient temperature is the temperature of the external environment in which the battery is located, and cooling system flow rate is the flow rate of the cooling medium in the battery cooling system, which are used to supplement the characterization of the external operating conditions of the battery.

[0008] Furthermore, in step S3, the eigenvalue decomposition calculation formula in the model reduction process is as follows: ,in, For the dimensionality-reduced data, It is a matrix containing the first k eigenvectors. This is the original data. This is the mean vector of the original data. for The transpose of the matrix, This is data after centralized processing.

[0009] Furthermore, in step S3, the calculation formula for the convolutional neural network in the model reduction process is as follows: ,in, This is the output of the l-th layer of the neural network. For the neural network The output of the layer, For the first The weight matrix of the layer, For the first The layer's bias vector, This is the activation function.

[0010] Furthermore, in the S3 model reduction step, the root mean square error (RMSE) is used to evaluate the accuracy of the initial reduced model. The formula for calculating the RMSE is: , in, The root mean square error, The total number of observations used for evaluation. For the first The actual data points of each observation. For the model to the first The prediction results for each observation, ( ) is the first The single-point error of each observation, ( )2 is the square of the corresponding single-point error; when the root mean square error is not less than 0.1%, return to step S2 to reconstruct the original data matrix, and repeat the eigenvalue decomposition dimensionality reduction and convolutional neural network modeling steps in step S3; when the root mean square error is less than 0.1%, the preliminary reduced-order model is determined to be a usable reduced-order model for fast calculation of battery thermal runaway.

[0011] Furthermore, in S4, the thermal runaway stage warning threshold in the calculation of the thermal runaway stage time step is specifically as follows: the first-level alarm threshold is temperature > 55℃, the second-level alarm threshold is temperature rise rate ≥ 1℃ / min, the thermal runaway reversible window threshold is LFP battery temperature < 150℃, and the thermal runaway irreversible threshold is LFP battery temperature > 250℃; where LFP is the abbreviation for lithium iron phosphate battery, and the temperature rise rate is the change in battery temperature per unit time.

[0012] Furthermore, in step S4, the delay compensation duration is set to 5s to 20s in the thermal runaway stage calculation step to offset the time delay generated by model calculation and data processing; the time step of the time series calculation is 0.5s to 2s, and the discretized data is stored at time step intervals, where the time step is the time interval between two adjacent time series calculations, and discretization storage means extracting and saving the continuous temperature field simulation results at time step intervals.

[0013] Furthermore, in step S5, the formula for calculating the thermal runaway countdown in the final countdown step is determined as follows: , in, This is the countdown time of thermal runaway based on measured data. The irreversible threshold temperature for thermal runaway is set at 250℃. The current battery temperature is collected by a temperature sensor. The rate of battery temperature rise is monitored in real time by a temperature sensor.

[0014] On the other hand, an electronic device is provided, comprising: a processor; and a memory storing computer-readable instructions, wherein when executed by the processor, the computer-readable instructions implement any of the methods described above for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model.

[0015] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, the at least one instruction being loaded and executed by a processor to implement any of the above-described methods for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model.

[0016] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: I. This invention constructs a basic electrochemical-thermal simulation model of lithium batteries that incorporates electrochemical and heat dissipation correlations, and integrates a thermal runaway mechanism model to form a complete lithium battery thermal runaway simulation model. This model accurately characterizes the relevant terms of thermal runaway reaction heat. In data processing, eigenvalue decomposition is used to reduce the dimensionality of the original data matrix, and convolutional neural networks are combined for regression to establish a precise mapping between operating conditions and physical field results. This model construction and data processing method not only improves computational efficiency but also ensures the accuracy of prediction results, providing a more scientific and reliable method for predicting thermal runaway of energy storage lithium batteries.

[0017] Second, this invention achieves real-time dynamic prediction of the thermal runaway development process of energy storage lithium batteries by using a reduced-order neural network model, which improves the foresight and initiative of early warning. By integrating simulation data and measured data, it obtains a more accurate and reliable remaining thermal runaway time, upgrading the thermal runaway prevention and control of energy storage system batteries from "passive response" to "active predictive intervention". This not only reduces the risk of thermal runaway and fire, but also improves the safety performance of energy storage systems, providing a strong guarantee for the safe development of the energy storage field. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the overall method for predicting thermal runaway; Figure 2 This is a flowchart of the sub-process for calculating the thermal runaway phase time; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0021] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0022] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0023] In this embodiment of the invention, sometimes a subscript such as W1 may be mistakenly written as a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0024] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0025] This invention provides a method for predicting thermal runaway in energy storage lithium batteries based on a reduced-order neural network model. This method can be implemented by an electronic device, such as a terminal or a server. Figure 1 The flowchart shown is for a method to predict thermal runaway of energy storage lithium batteries based on a neural network reduced-order model. The processing flow of this method may include the following steps: Example 1: Prediction of Thermal Runaway in Lithium Iron Phosphate Battery Packs for Large-Scale Energy Storage Power Stations S1: Constructing a thermal runaway simulation model First, considering the characteristics of large-scale lithium iron phosphate battery packs in large-scale energy storage power stations—large capacity, long-term continuous operation, and complex operating conditions—a basic electrochemical-thermal simulation model for lithium batteries, incorporating electrochemical and heat dissipation correlations, is established. The calculation formula for the basic electrochemical-thermal simulation model is as follows: ; ; ,in, For battery material density, Heat capacity per unit mass For the battery temperature field, For time, For the material's thermal conductivity, For spatial differential operators, The effective conductivity of the positive electrode. The effective conductivity of the negative electrode. The potential distribution is positive. The potential distribution is negative. The electrochemical reaction heat is calculated by the electrochemical sub-model. Joule heating caused by an internal short circuit. The heat of thermal runaway reaction activated only under conditions of thermal abuse. The current density of the electrochemical reaction. This represents the internal short-circuit current density. The electrode current density is used. This model needs to encompass key physical properties such as battery material density, heat capacity per unit mass, and thermal conductivity, as well as electrochemical parameters such as effective conductivity and potential distribution of the positive and negative electrodes. It also needs to include calculations related to heat sources such as electrochemical reaction heat, Joule heating caused by internal short circuits, and thermal runaway reaction heat activated under thermal abuse conditions. Through the corresponding electrochemical-thermal simulation basic model calculation formulas, a basic simulation of temperature field changes during normal battery operation is achieved. The purpose of this process is to provide a basic framework for subsequent simulations of the battery transitioning from a normal state to a thermal runaway state, ensuring accurate capture of the temperature change patterns of the battery during normal operation. This lays a reliable foundation for the integration of subsequent thermal runaway mechanism models and avoids deviations in subsequent simulation results due to missing key parameters in the basic model. Figure 1 As shown.

[0026] Next, a thermal runaway mechanism model describing the characteristics of various chemical reactions during the thermal runaway process is constructed. The calculation formula for the thermal runaway mechanism model is as follows: ; ; ; ,in, It is a differential operator used to represent the instantaneous rate of change of a variable with respect to another variable; Time is used to characterize the time dimension in the reaction process; For dimensionless concentration characterization of the residual reactants in the SEI (solid electrolyte interface) decomposition reaction, This is a dimensionless concentration characterization of the remaining reactants in the negative electrode electrolyte reaction. For characterizing the dimensionless concentration of residual reactants in the electrolyte reaction. This is a dimensionless concentration characterization of the remaining reactants in the positive electrode electrolyte reaction; The pre-exponential factor for the SEI decomposition reaction. It is the pre-exponential factor for the negative electrode electrolyte reaction. It is the pre-exponential factor for the positive electrode electrolyte reaction. It is the pre-exponential factor for electrolyte reaction; The activation energy for the SEI decomposition reaction. The activation energy for the reaction in the negative electrode electrolyte is... This is the activation energy for the positive electrode electrolyte reaction. It is the activation energy of the electrolyte reaction; It is a universal gas constant. Battery temperature; The reaction order of the SEI decomposition reaction is given. This represents the reaction order of the negative electrode electrolyte reaction. , These are all reaction orders of the positive electrode electrolyte reaction. The reaction order of the electrolyte reaction; This is a dimensionless characterization of the SEI layer thickness. This serves as the reference thickness for the SEI layer. The study focuses on four core reaction types: SEI (solid electrolyte interface) decomposition, negative electrode electrolyte reaction, positive electrode electrolyte reaction, and electrolyte reaction. Key kinetic parameters such as pre-exponential factors, activation energies, and reaction orders for each reaction are identified. Using corresponding thermal runaway mechanism models, the study quantifies the extent and heat generation rate of each reaction. Its purpose is to deeply analyze the intrinsic chemical mechanisms of thermal runaway, clearly demonstrating the contribution of different chemical reactions to the battery temperature rise. This allows the subsequently integrated model to accurately reflect the root cause of the rapid temperature increase during thermal runaway, rather than merely simulating surface temperature changes, thus providing theoretical support at the chemical level for accurate prediction of the thermal runaway stage.

[0027] Finally, the thermal runaway mechanism model is integrated into the basic electrochemical-thermal simulation model, representing the thermal runaway reaction heat, to form a complete lithium battery thermal runaway simulation model capable of simulating the entire process from normal operation to thermal runaway. This step achieves full-chain simulation of the "electrochemical-thermal runaway chemical reaction," breaking the limitation of traditional models that can only simulate a single stage. The model can cover all stages of the battery from normal operation to thermal runaway, accurately simulating everything from the slow temperature rise during normal operation to the intensification of chemical reactions in the early stages of thermal runaway, and finally the rapid temperature increase. This provides a comprehensive and realistic source of simulation data for the subsequent construction of the original data matrix.

[0028] S2: Construct the original data matrix

[0029] Based on the actual operating data of lithium iron phosphate battery packs in large-scale energy storage power stations, the input and output parameters of the original data matrix are determined. In addition to battery voltage, current, and time, the input parameters include ambient temperature (i.e., the temperature inside the power station's equipment room where the battery is located, covering different operating conditions such as day-night temperature differences and seasonal variations) and cooling system flow parameters (i.e., the flow rate of the cooling medium in the battery cooling system, including values ​​under different cooling states such as normal operation and fault derating), to comprehensively characterize the external operating conditions of the battery. The output parameters are set as battery time-series temperature field data, which must cover the temperature values ​​of the battery surface and key internal monitoring points at different time points.

[0030] During the data collection and screening process, it is necessary to collect battery operation data under different operating loads and environmental conditions throughout the year at the power station, and remove outliers and invalid data to ultimately construct the original data matrix for training the reduced-order model. This step serves two main purposes: First, supplementing the data with environmental temperature and cooling system flow parameters allows for a more comprehensive reconstruction of the actual battery operating scenario, preventing the model training data from becoming disconnected from reality due to neglecting external operating conditions, and ensuring that the subsequently trained reduced-order model can accurately predict under different external environments. Second, screening effective data and constructing the matrix provides a high-quality, highly relevant data source for subsequent model reduction processing, reducing the interference of invalid data on model training, improving model training efficiency and final prediction accuracy, and providing reliable data support for eigenvalue decomposition dimensionality reduction and convolutional neural network modeling.

[0031] S3: Model order reduction processing

[0032] First, eigenvalue decomposition is performed on the constructed original data matrix to reduce its dimensionality. The eigenvalue decomposition formula transforms the high-dimensional original data into low-dimensional feature vectors, preserving key information and reducing the complexity of subsequent model calculations. The eigenvalue decomposition formula is as follows: ,in, For the dimensionality-reduced data, It is a matrix containing the first k eigenvectors. This is the original data. This is the mean vector of the original data. for The transpose of the matrix, This is the data after centralized processing. The purpose of this operation is to solve the problems of high dimensionality and redundant information in the original data. While removing irrelevant or low-relevance redundant data, it retains the core feature information related to battery thermal runaway to the greatest extent. This reduces the computational load of subsequent convolutional neural network modeling, shortens the model training time, and avoids the model overfitting problem that may be caused by high-dimensional data, thus creating conditions for efficient and accurate modeling in the future.

[0033] Subsequently, regression analysis was performed on the dimensionality-reduced data using a convolutional neural network. Through the convolutional neural network calculation formula, a mapping relationship was constructed between the operating conditions (input parameter combinations) and the physical field results (output temperature field data), forming a preliminary dimensionality reduction model. The convolutional neural network calculation formula is as follows: ,in, This is the output of the l-th layer of the neural network. For the neural network The output of the layer, For the first The weight matrix of the layer, For the first The layer's bias vector, The activation function is selected by continuously adjusting the number of layers, weight matrices, and bias vectors of the neural network during model training. This step leverages the powerful nonlinear fitting capabilities of convolutional neural networks to establish a precise correlation between battery operating conditions and temperature field results. Compared to traditional linear models, it better captures the complex nonlinear relationship between operating parameters and temperature changes, giving the initial reduced-order model superior temperature prediction capabilities. It can quickly output the corresponding temperature field results based on the input operating parameters, meeting the real-time requirements of large-scale energy storage power stations for thermal runaway prediction.

[0034] Finally, the root mean square error (RMSE) is used to assess the accuracy of the preliminary reduced-order model. The RMSE calculation formula is used to compare the difference between the model's predictions and the actual data. The RMSE calculation formula is as follows: ,in, The root mean square error, The total number of observations used for evaluation. For the first The actual data points of each observation. For the model to the first The prediction results for each observation, ( ) is the first The single-point error of each observation, ( Step S2 represents the square of the corresponding single-point error. If the root mean square error (RMSE) is not less than 0.1%, return to step S2 to reconstruct the original data matrix and repeat the eigenvalue decomposition dimensionality reduction and convolutional neural network modeling steps in step S3. If the RMSE is less than 0.1%, the preliminary reduced-order model is determined to be a usable reduced-order model for rapid calculation of battery thermal runaway. If the RMSE is not less than 0.1%, return to step S2 to reconstruct the original data matrix, supplement with more diverse operating condition data, and repeat the eigenvalue decomposition dimensionality reduction and convolutional neural network modeling steps. If the RMSE is less than 0.1%, the preliminary reduced-order model is determined to be a usable reduced-order model for rapid calculation of battery thermal runaway. The purpose of this accuracy assessment and model optimization process is to strictly control the model quality, ensure that the prediction accuracy of the final reduced-order model meets the needs of actual applications, and avoid deviations in thermal runaway prediction due to insufficient model accuracy, which could lead to safety accidents. Through repeated iterative optimization, the model can achieve high prediction accuracy while ensuring computational speed, providing reliable model support for subsequent thermal runaway stage time calculations.

[0035] S4: Calculate the thermal runaway phase time

[0036] In the daily operation of large-scale energy storage power stations, voltage and current data of lithium iron phosphate battery packs are collected in real time and input into a pre-defined, available reduced-order model. Time-series calculations are performed with a fixed 1-second time step, reading the battery time-series temperature field simulation results every 1 second. A discretized storage method is used to extract and save continuous temperature field simulation results at 1-second intervals, forming a complete temperature change time-series dataset. This process enables real-time dynamic monitoring and data recording of the battery temperature field. The 1-second time step ensures data timeliness, promptly capturing subtle temperature changes, while avoiding excessive data volume and increased storage and processing pressure due to excessively short time intervals. Discretized storage facilitates rapid retrieval of time points corresponding to various temperature thresholds, providing a clear and orderly data foundation for the division of thermal runaway stages. Figure 2 As shown.

[0037] Based on preset thermal runaway stage warning thresholds (Level 1 alarm threshold: temperature > 55℃; Level 2 alarm threshold: temperature rise rate ≥ 1℃ / min; reversible thermal runaway window threshold: lithium iron phosphate battery temperature < 150℃; irreversible thermal runaway threshold: lithium iron phosphate battery temperature > 250℃), using the highest internal temperature of the battery as the key monitoring point, the system searches for the time points corresponding to each threshold in the discretely stored temperature dataset, thereby obtaining the arrival time of each stage of thermal runaway (Level 1 alarm stage, Level 2 alarm stage, reversible window stage, and irreversible stage). Its function is to refine the battery thermal runaway process into different stages through clear threshold divisions, enabling personnel to clearly understand the current thermal runaway risk level of the battery, rather than just knowing the vague information that "there is a risk of thermal runaway." This provides a basis for subsequent targeted countermeasures; for example, during the reversible window stage, enhanced cooling can be used to prevent further development of thermal runaway.

[0038] Considering the time delays that occur during model calculations and data processing, delay compensation is applied to the arrival times of each stage (the compensation duration is set to 10 seconds). Then, the reciprocal calculation is performed on the compensated arrival times of each stage to obtain the countdown time for thermal runaway based on the model simulation. The purpose of delay compensation is to offset the time loss during model calculations and data transmission, ensuring that the obtained thermal runaway stage arrival times are consistent with the actual situation, avoiding delayed warnings and missing the optimal intervention opportunity due to time delays. The reciprocal calculation of the arrival times transforms the "stage arrival time" into a more intuitive concept of "remaining time," making it easier for staff to quickly understand the remaining time before each stage of thermal runaway, improving the readability and practicality of the warning information, and providing a clear simulation time reference for subsequent comparison with measured data to determine the final countdown.

[0039] S5: Determine the final countdown

[0040] Temperature sensors installed on lithium iron phosphate battery packs in large-scale energy storage power stations are used to collect the current temperature and temperature rise rate (i.e., the change in battery temperature per unit time) of the battery in real time. Based on the measured temperature rise rate, a thermal runaway countdown calculation formula is used. Substituting the thermal runaway irreversible threshold temperature, the current battery temperature, and the measured temperature rise rate into the formula, the thermal runaway countdown time based on the measured data is calculated. The thermal runaway countdown calculation formula is as follows: ,in, This is the countdown time of thermal runaway based on measured data. The irreversible threshold temperature for thermal runaway is set at 250℃. The current battery temperature is collected by a temperature sensor. This step involves using a temperature sensor to monitor the battery temperature rise rate in real time. The purpose of this step is to obtain a thermal runaway time prediction result that most closely approximates the actual battery state, based on actual hardware monitoring data. This avoids the potential theoretical and practical discrepancies that may arise from relying solely on model simulations. By introducing measured data, the thermal runaway time prediction becomes more realistic and reliable, providing crucial measured evidence for the final determination of the countdown.

[0041] The thermal runaway countdown time obtained through model simulation is compared with that obtained based on measured data, and the smaller of the two values ​​is selected as the final thermal runaway countdown time. This comparative selection process employs a "conservative prediction" principle to minimize safety risks caused by prediction errors. Since model simulations may not account for extreme conditions, and measured data may fluctuate due to instantaneous sensor errors, selecting the smaller value as the final countdown ensures that the remaining thermal runaway time obtained by personnel is the most critical reference value, prompting them to take proactive measures and avoiding complacency due to excessively long predicted times, thus preventing delays in response.

[0042] When the countdown time for thermal runaway reaches the preset warning trigger condition, the corresponding warning mechanism of the energy storage power station is immediately activated, such as issuing audible and visual alarms, activating the backup cooling system, and disconnecting some battery pack circuits. This step transforms the prediction results into actual safety protection actions, forming a complete closed loop of "prediction-warning-response." By promptly activating the warning mechanism, the occurrence of thermal runaway accidents can be effectively prevented or the losses caused by such accidents reduced, ensuring the safety of equipment and personnel in large-scale energy storage power stations and fully leveraging the practical application value of thermal runaway prediction methods.

[0043] In summary, accurate prediction of thermal runaway in large-scale energy storage power station lithium iron phosphate battery packs is achieved through five core steps. First, a complete simulation model integrating an electrochemical-thermal simulation model and a thermal runaway mechanism model is constructed, providing a basic framework for subsequent predictions. Next, a high-quality raw data matrix is ​​built using multi-dimensional operating parameters to ensure the reliability of model training data. Then, dimensionality reduction through eigenvalue decomposition and convolutional neural network modeling are performed, with root mean square error used to control accuracy, resulting in a usable reduced-order model. Subsequently, real-time data is input to calculate the time for each stage of thermal runaway, and delay compensation ensures timeliness. Finally, measured data is used to determine the final countdown and trigger an early warning. The entire process covers the entire workflow of "model construction - data support - accuracy optimization - time calculation - early warning handling," effectively adapting to the complex operating conditions of power station batteries and providing crucial protection for the safe operation of the power station.

[0044] Example 2: Prediction of Thermal Runaway in New Energy Vehicle Power Batteries (Lithium Iron Phosphate)

[0045] S1: Constructing a thermal runaway simulation model

[0046] Considering the frequent start-stop, acceleration, deceleration, and multi-road-condition driving characteristics of new energy vehicle power batteries, a basic electrochemical-thermal simulation model of lithium batteries, incorporating electrochemical and heat dissipation relationships, is first established. The calculation formula for the basic electrochemical-thermal simulation model is as follows: ; ; This model needs to consider the dynamic charging and discharging behavior of the battery during vehicle operation, incorporating physical parameters such as battery material density, heat capacity per unit mass, and thermal conductivity, as well as electrochemical parameters such as effective conductivity and potential distribution of the positive and negative electrodes. Simultaneously, through the corresponding electrochemical-thermal simulation basic model calculation formulas, it accurately calculates the heat of electrochemical reactions, Joule heating caused by internal short circuits, and thermal runaway reaction heat under thermal abuse conditions, achieving a basic simulation of temperature field changes under different battery driving conditions. The purpose of this step is to provide a foundational framework that fits the actual application scenario for subsequent integration of thermal runaway mechanism models, avoiding discrepancies between simulation results and reality due to the model's failure to consider the dynamic driving characteristics of the vehicle. This ensures accurate capture of the temperature fluctuation patterns of the battery during normal driving, laying a solid foundation for subsequent in-depth simulation of the thermal runaway process.

[0047] Then, a thermal runaway mechanism model was constructed, focusing on analyzing the SEI decomposition reaction, negative electrode electrolyte reaction, positive electrode electrolyte reaction, and electrolyte reaction that may trigger thermal runaway during the operation of new energy vehicles. The kinetic parameters of each reaction, such as pre-exponential factor, activation energy, and reaction order, were clarified. Using the corresponding thermal runaway mechanism model calculation formulas, the changes in each reaction with temperature and time, as well as the heat generation, were described. Its purpose is to analyze the occurrence process of thermal runaway from a chemical perspective, clearly presenting the driving effect of different chemical reactions on the sudden rise in battery temperature. This breaks through the limitation of traditional models that can only simulate surface temperature changes, allowing the subsequently integrated model to not only predict temperature changes but also reflect the chemical causes behind these changes. This provides a scientific basis for accurately classifying thermal runaway stages and setting early warning thresholds.

[0048] Finally, the thermal runaway mechanism model is integrated into the relevant terms characterizing the thermal runaway reaction heat in the basic electrochemical-thermal simulation model, forming a complete lithium battery thermal runaway simulation model suitable for new energy vehicle power batteries. The calculation formula for the thermal runaway mechanism model is as follows: ; ; ; The purpose of this step is to achieve a full-chain simulation of "driving conditions - electrochemical changes - heat dissipation - thermal runaway chemical reactions," enabling the model to cover the entire process of a battery from its normal state to thermal runaway under diverse scenarios such as normal driving, fast charging, and collisions. Whether it is the gradual temperature rise during low-speed driving, the rapid temperature rise during fast charging, or the intensified chemical reactions that may be triggered after a collision, the model can accurately simulate these phenomena, providing comprehensive and realistic simulation data support for the subsequent construction of the original data matrix. This ensures that the subsequent model training data can cover the core application scenarios of automotive power batteries.

[0049] S2: Construct the original data matrix

[0050] Based on the actual operating data of new energy vehicle power batteries, the input and output parameters of the original data matrix are determined. In addition to battery voltage, current, and time, the input parameters include ambient temperature (such as high temperatures in summer, low temperatures in winter, and temperatures under different regional climate conditions) and cooling system flow parameters (the flow rate of the cooling medium under different operating modes of the cooling system during vehicle operation) to comprehensively reflect the external operating conditions of the battery during vehicle operation. The output parameters are battery time-series temperature field data, which must include temperature information at different times for different locations within the battery cell and battery module (such as the positive electrode, negative electrode, and the center of the cell).

[0051] During the data acquisition and processing phase, battery operation data from multiple new energy vehicles of the same model under different mileages, road conditions (urban roads, highways, mountain roads), and charging / discharging methods (slow charging, fast charging) needs to be collected. The data is preprocessed to remove invalid data from abnormal driving conditions such as severe collisions and extreme overloads, ultimately constructing the original data matrix for training the reduced-order model. This step serves two main purposes: first, it supplements environmental temperature and cooling system flow parameters, more comprehensively recreating the actual operating environment of the vehicle's power battery, avoiding the problem of ignoring external environmental differences and ensuring that the subsequently trained reduced-order model can accurately predict under different climates and driving modes; second, it filters effective data and constructs a matrix, providing a high-quality, highly relevant data source for subsequent model reduction processing, reducing interference from invalid data, improving training efficiency, and ensuring that the model can learn the temperature change patterns of the battery under diverse operating conditions, laying a data foundation for accurate prediction of thermal runaway.

[0052] S3: Model order reduction processing

[0053] First, eigenvalue decomposition is performed on the original data matrix to reduce its dimensionality. The formula for eigenvalue decomposition is as follows: By using eigenvalue decomposition formulas, key feature vectors are extracted from high-dimensional raw data, reducing data dimensionality and decreasing model computational complexity while retaining core information. This operation addresses the problem of high-dimensionality and redundant information in the original data, eliminating redundant data with low correlation to battery thermal runaway, and maximizing the retention of core correlation information between parameters such as voltage, current, and ambient temperature and temperature field changes. This reduces the computational load of subsequent convolutional neural network modeling, shortens model training time, and avoids overfitting issues that may occur with high-dimensional data. It ensures the model can more efficiently learn the intrinsic relationship between key parameters and temperature changes, laying the foundation for rapid prediction of thermal runaway.

[0054] Next, a regression model is performed on the dimensionality-reduced data using a convolutional neural network. The convolutional neural network calculation formula is then used to construct a mapping model between battery operating conditions (input parameters) and temperature field results (output parameters). The convolutional neural network calculation formula is as follows: During model training, the model performance is continuously optimized by adjusting the layer structure, weight matrix, bias vector, and activation function of the neural network. This step leverages the powerful nonlinear fitting capabilities of convolutional neural networks to accurately capture the complex nonlinear relationship between battery operating conditions and temperature field changes. Compared to traditional linear models, it better adapts to the variable operating conditions of automotive power batteries, such as accurately fitting the correlation between a sudden increase in current and a rapid rise in temperature during rapid acceleration, and the impact of cooling system flow changes on temperature control in low-temperature environments. This gives the initial reduced-order model better temperature prediction accuracy and operating condition adaptability, meeting the dual requirements of real-time performance and accuracy in predicting thermal runaway in new energy vehicles.

[0055] Finally, the root mean square error (RMSE) was used to assess the accuracy of the preliminary reduced-order model. Based on the RMSE calculation formula, the predicted temperature data and the actual collected temperature data were compared. The RMSE calculation formula is as follows: If the root mean square error (RMSE) is ≥0.1%, return to step S2 to supplement with more diverse vehicle driving condition data (such as data under different altitudes and humidity environments), reconstruct the original data matrix, and repeat the eigenvalue decomposition dimensionality reduction and convolutional neural network modeling steps. If the RMSE is <0.1%, the preliminary reduced-order model is determined to be a usable reduced-order model for rapid calculation of thermal runaway in new energy vehicle power batteries. The purpose of this accuracy assessment and iterative optimization process is to strictly control model quality, ensuring that the prediction accuracy of the final reduced-order model meets the safety requirements of vehicle operation, and avoiding prediction deviations in thermal runaway due to insufficient model accuracy, which could lead to safety accidents. Through repeated iterations, the model achieves high prediction accuracy while maintaining computational speed, providing reliable model support for subsequent thermal runaway stage time calculations.

[0056] S4: Calculate the thermal runaway phase time

[0057] During the operation of new energy vehicles, the on-board battery management system collects the voltage and current data of the power battery in real time and inputs it into a usable reduced-order model. Time-series calculations are performed with a fixed time step of 1 second, reading the battery time-series temperature field simulation results every second. The continuous temperature field simulation results are then discretized and stored at 1-second intervals, forming a real-time temperature change data sequence. This process enables dynamic real-time monitoring and data recording of the battery temperature field. The 1-second time step ensures data timeliness, promptly capturing rapid temperature changes during scenarios such as rapid acceleration and fast charging, while avoiding excessive data storage and processing pressure on the on-board system due to excessively short time intervals. Discretized storage facilitates quick retrieval of the time points corresponding to each temperature threshold, providing a clear and orderly data foundation for the division of thermal runaway stages, ensuring timely identification of the battery's thermal runaway risk stage.

[0058] Based on preset thermal runaway stage warning thresholds (Level 1 alarm threshold: temperature > 55℃; Level 2 alarm threshold: temperature rise rate ≥ 1℃ / min; reversible thermal runaway window threshold: lithium iron phosphate battery temperature < 150℃; irreversible thermal runaway threshold: lithium iron phosphate battery temperature > 250℃), using the highest internal temperature of the battery as the monitoring point, the system accurately locates the time corresponding to each threshold in the discretely stored temperature data sequence, thereby determining the arrival time of each thermal runaway stage (Level 1 alarm, Level 2 alarm, reversible window, irreversible). Its function is to refine the battery thermal runaway process into stages with different risk levels through clear threshold divisions, enabling the vehicle system and driver to clearly understand the current level of thermal runaway risk. For example, when the Level 1 alarm threshold is reached, only the driver needs to be alerted to the battery temperature; when the reversible window stage is entered, timely intervention measures such as enhanced cooling should be taken, providing a clear basis for subsequent targeted handling and avoiding inappropriate responses due to ambiguity in risk levels.

[0059] Due to the time delay in model calculation and data processing within the vehicle system, a 10-second delay compensation is applied to the arrival times of each stage. Then, the reciprocal calculation is performed on the compensated arrival times of each stage to obtain the thermal runaway countdown time based on the model simulation. The delay compensation serves to offset the time loss during vehicle system calculation and data transmission, ensuring that the obtained thermal runaway stage arrival times are consistent with the actual battery state. This avoids delayed warnings caused by time delays, preventing missed opportunities for optimal intervention within the reversible window. The reciprocal calculation transforms the "stage arrival time" into a more intuitive "remaining time," facilitating the vehicle system to deliver warning information to the driver in a "countdown" format. This improves the readability and urgency of the warning information and provides a clear simulation time reference for subsequent comparison with measured data to determine the final countdown.

[0060] S5: Determine the final countdown

[0061] Temperature sensors installed on the power batteries of new energy vehicles are used to obtain the current temperature and measured temperature rise rate of the battery in real time. Based on the measured temperature rise rate, a thermal runaway countdown calculation formula is used. Substituting the thermal runaway irreversible threshold temperature, the current battery temperature, and the measured temperature rise rate into the formula, the thermal runaway countdown time based on the measured data is calculated. The thermal runaway countdown calculation formula is as follows: The purpose of this step is to obtain the thermal runaway time prediction result that most closely reflects the actual state of the battery by relying on real-time monitoring data from hardware sensors. This avoids the theoretical deviations that may exist when relying solely on model simulations. For example, when the battery suffers internal damage that is not detected by the sensors due to a minor collision, the measured temperature rise rate will more accurately reflect the risk. By introducing measured data, the thermal runaway time prediction becomes more realistic and reliable, providing key measured evidence for the final determination of the countdown.

[0062] By comparing the simulated countdown time of thermal runaway with the measured countdown time, the smaller value is selected as the final countdown time. This comparative selection process employs a "conservative prediction" principle, minimizing safety risks caused by prediction errors. Since the model simulation may not consider individual battery differences and sensor instantaneous errors, selecting the smaller value as the final countdown ensures that the remaining thermal runaway time obtained by the driver is the "most critical" reference value, prompting them to take proactive measures, such as slowing down or stopping fast charging, avoiding complacency due to an excessively long predicted time and delaying appropriate action.

[0063] When the countdown time for thermal runaway reaches the preset warning level threshold of the vehicle system, the system immediately triggers corresponding warning measures, such as issuing a voice alarm to the driver and displaying warning information on the instrument panel. In case of emergency, it will also automatically activate the battery cooling enhancement mode, limit the battery charging and discharging power, and even disconnect the battery main circuit in extreme cases. The purpose of this step is to transform the prediction results into actual safety protection actions, forming a complete closed loop of "real-time monitoring - accurate prediction - graded warning - proactive handling". By promptly activating warning and intervention measures, it effectively prevents thermal runaway accidents from occurring or reduces accident losses, ensuring vehicle driving safety and the safety of passengers, and fully leveraging the practical application value of thermal runaway prediction methods in new energy vehicle scenarios.

[0064] In summary, a prediction process based on the dynamic driving characteristics of vehicles is designed for predicting thermal runaway in new energy vehicle power batteries. First, a thermal runaway simulation model closely matches the vehicle's operating conditions, taking into account electrochemistry, heat dissipation, and thermal runaway chemical reactions. Then, diverse driving scenario data is collected to construct an original data matrix, improving the model's adaptability to different operating conditions. Through eigenvalue decomposition for dimensionality reduction and convolutional neural network modeling, combined with root mean square error iterative optimization, a reduced-order model meeting automotive-grade requirements is obtained. Real-time data collection calculates the thermal runaway stage time, and delay compensation eliminates the impact of onboard system latency. Finally, a conservative final countdown is determined based on measured data, triggering tiered warnings and intervention measures. This process accurately matches the characteristics of frequent start-stop cycles and variable operating conditions of automotive power batteries, forming a closed loop of "prediction-warning-response," effectively ensuring the safety of vehicle operation and passengers.

[0065] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, such as... Figure 3 As shown. Optionally, the electronic device 410 may include a first processor 2001.

[0066] Optionally, the electronic device 410 may also include a memory 2002 and a transceiver 2003.

[0067] The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus.

[0068] The following is combined with Figure 3 A detailed description of each component of electronic device 410 is provided below: The first processor 2001 is the control center of the electronic device 410. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0069] Optionally, the first processor 2001 can perform various functions of the electronic device 410 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0070] In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, for example... Figure 3 CPU0 and CPU1 are shown in the diagram.

[0071] In a specific implementation, as one example, the electronic device 410 may also include multiple processors, for example... Figure 3 The first processor 2001 and the second processor 2004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, a processor can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).

[0072] The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0073] Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently and be connected via the interface circuit of the electronic device 410. Figure 3 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0074] The transceiver 2003 is used to communicate with network devices or with terminal devices.

[0075] Alternatively, transceiver 2003 may include a receiver and a transmitter. Figure 3 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0076] Optionally, the transceiver 2003 can be integrated with the first processor 2001, or it can exist independently and be connected via the interface circuit of the electronic device 410. Figure 3 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0077] It should be noted that, Figure 3 The structure of the electronic device 410 shown does not constitute a limitation on the router. Actual knowledge structure identification devices may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0078] Furthermore, the technical effects of the electronic device 410 can be referred to the technical effects of the XXX method described in the above method embodiments, and will not be repeated here.

[0079] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0080] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can 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. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0081] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0082] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0083] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0084] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0085] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed 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 implementations should not be considered beyond the scope of this invention.

[0086] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0087] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0088] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0089] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0090] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0091] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model, characterized in that, The specific steps of this method are as follows: S1. Constructing a thermal runaway simulation model: First, establish a basic simulation model of lithium battery electrochemistry-heat dissipation containing electrochemistry and heat dissipation correlation. Then, construct a thermal runaway mechanism model that describes the characteristics of various chemical reactions in the thermal runaway process. Integrate this mechanism model into the relevant terms characterizing the thermal runaway reaction heat in the basic electrochemistry-heat dissipation simulation model to form a complete lithium battery thermal runaway simulation model. S2, Construct the original data matrix: With battery voltage, current and time as input, and battery time-series temperature field data obtained by the thermal runaway simulation model as output, the input and output parameters are used as target feature values ​​to construct the original data matrix for training the reduced-order model; S3, Model Dimensionality Reduction: Perform eigenvalue decomposition on the original data matrix to reduce the dimensionality of the data, and combine convolutional neural network to regress the dimensionality-reduced data and establish a mapping between working conditions and physical field results to form a preliminary dimensionality reduction model; S4, Calculate the thermal runaway stage time: Input the battery voltage and current collected on site into the preliminary reduced-order model, perform time-series calculation with a preset time step, and read and discretize the simulation results of the time-series temperature field. Based on the preset thermal runaway stage warning threshold, the highest internal temperature of the battery is used as the monitoring point. The arrival time of each stage of thermal runaway is obtained by finding the time corresponding to each threshold. After adding delay compensation, the arrival time of each stage is counted in reverse to obtain the simulated thermal runaway countdown time. S5, Determine the final countdown: Based on the actual temperature rise rate measured by the on-site temperature sensor, calculate the actual thermal runaway countdown time using the thermal runaway countdown formula; compare the simulation and actual thermal runaway countdown times, and select the smaller value as the final thermal runaway countdown time.

2. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, The calculation formula for the basic electrochemical-heat dissipation simulation model in step S1, which involves constructing the thermal runaway simulation model, is as follows: ； ； ， in, For battery material density, Heat capacity per unit mass For the battery temperature field, For time, For the material's thermal conductivity, For spatial differential operators, The effective conductivity of the positive electrode. The effective conductivity of the negative electrode. The potential distribution is positive. The potential distribution is negative. The electrochemical reaction heat is calculated by the electrochemical sub-model. Joule heating caused by an internal short circuit. The heat of thermal runaway reaction activated only under conditions of thermal abuse. The current density of the electrochemical reaction. This represents the internal short-circuit current density. This represents the electrode current density.

3. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, The calculation formula for the thermal runaway mechanism model in step S1, which involves constructing the thermal runaway simulation model, is as follows: ； ； ； ， in, It is a differential operator used to represent the instantaneous rate of change of a variable with respect to another variable; Time is used to characterize the time dimension in the reaction process; For dimensionless concentration characterization of the residual reactants in the SEI decomposition reaction, This is a dimensionless concentration characterization of the remaining reactants in the negative electrode electrolyte reaction. For characterizing the dimensionless concentration of residual reactants in the electrolyte reaction. This is a dimensionless concentration characterization of the remaining reactants in the positive electrode electrolyte reaction; The pre-exponential factor for the SEI decomposition reaction. It is the pre-exponential factor for the reaction of the negative electrode electrolyte. It is the pre-exponential factor for the positive electrode electrolyte reaction. It is the pre-exponential factor for electrolyte reaction; The activation energy for the SEI decomposition reaction. The activation energy for the reaction in the negative electrode electrolyte is... The activation energy for the positive electrode electrolyte reaction. It is the activation energy of the electrolyte reaction; It is a universal gas constant. Battery temperature; The reaction order of the SEI decomposition reaction is given. This represents the reaction order of the negative electrode electrolyte reaction. , These are all reaction orders of the positive electrode electrolyte reaction. The reaction order of the electrolyte reaction; This is a dimensionless characterization of the SEI layer thickness. This is the reference thickness for the SEI layer.

4. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, In step S2, when constructing the original data matrix, the input parameters, in addition to battery voltage, current and time, also include ambient temperature and cooling system flow rate. The ambient temperature is the temperature of the external environment where the battery is located, and the cooling system flow rate is the flow rate of the cooling medium in the battery cooling system, which is used to supplement the characterization of the external operating conditions of the battery.

5. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, The eigenvalue decomposition calculation formula in step S3, the model order reduction process, is as follows: ， in, For the dimensionality-reduced data, It is a matrix containing the first k eigenvectors. The original data, This is the mean vector of the original data. for The transpose of the matrix, This is data after centralized processing.

6. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, The formula for calculating the convolutional neural network in step S3, the model reduction process, is as follows: ， in, This is the output of the l-th layer of the neural network. For the neural network The output of the layer, For the first The weight matrix of the layer, For the first The layer's bias vector, This is the activation function.

7. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, In the S3 model reduction step, the root mean square error (RMSE) is used to evaluate the accuracy of the initial reduced model. The formula for calculating the RMSE is: ， in, The root mean square error, The total number of observations used for evaluation. For the first The actual data points of each observation. For the model to the first The prediction results for each observation, ( ) is the first The single-point error of each observation, ( )2 is the square of the corresponding single-point error; when the root mean square error is not less than 0.1%, return to step S2 to reconstruct the original data matrix, and repeat the eigenvalue decomposition dimensionality reduction and convolutional neural network modeling steps in step S3; when the root mean square error is less than 0.1%, the preliminary reduced-order model is determined to be a usable reduced-order model for fast calculation of battery thermal runaway.

8. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, In step S4, the thermal runaway stage warning threshold in the time step of calculating the thermal runaway stage is as follows: the first-level alarm threshold is temperature > 55℃, the second-level alarm threshold is temperature rise rate ≥ 1℃ / min, the thermal runaway reversible window threshold is LFP battery temperature < 150℃, and the thermal runaway irreversible threshold is LFP battery temperature > 250℃; where LFP is the abbreviation for lithium iron phosphate battery, and the temperature rise rate is the change in battery temperature per unit time.

9. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, In step S4, the delay compensation duration is set to 5s to 20s in the thermal runaway stage calculation step to offset the time delay generated by model calculation and data processing. The time step of the time series calculation is 0.5s to 2s. The discretized data is stored at time step intervals, where the time step is the time interval between two adjacent time series calculations. Discretization storage means extracting and saving the continuous temperature field simulation results at time step intervals.

10. The method for predicting thermal runaway of energy storage lithium batteries based on a reduced-order neural network model according to claim 1, characterized in that, In step S5, the formula for calculating the thermal runaway countdown in the final countdown step is determined as follows: ， in, This is the countdown time of thermal runaway based on measured data. The irreversible threshold temperature for thermal runaway is set at 250℃. The current battery temperature is collected by a temperature sensor. The rate of battery temperature rise is monitored in real time by a temperature sensor.

11. An electronic device, characterized in that, The electronic device includes: A processor; a memory storing computer-readable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 10.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that can be invoked by a processor to execute the method as described in any one of claims 1 to 10.

Images