A method and apparatus for extracting parameters of battery thermal runaway reaction

Abstract

This invention relates to a method and apparatus for extracting parameters of a battery thermal runaway reaction. The method includes the following steps: 1) obtaining a thermal runaway heat flow curve of a lithium-ion battery; 2) adaptively dividing the heat flow curve into multiple fitting regions; 3) for each fitting region, initializing a chemical reaction kinetic model based on a pre-built offline equivalent kinetic database, and fitting the model using the initialized model; 4) selecting the optimal solution for the reaction parameters based on the fitting results, wherein the reaction parameters include the number of peaks. Compared with the prior art, this invention has the advantages of high computational efficiency and high accuracy.

Description

Technical Field

[0001] This invention relates to the field of lithium battery safety technology, and in particular to a method and apparatus for extracting parameters of battery thermal runaway reaction. Background Technology

[0002] Lithium-ion batteries, with their advantages of high energy density, good cycle performance, and no memory effect, have been widely used in portable electronic devices, electric vehicles, and energy storage systems. However, their application is often limited by operating voltage, temperature range, and charge / discharge rate range. Various electrical, thermal, and mechanical abuse conditions accelerate battery performance degradation and can even lead to safety issues such as thermal runaway. Furthermore, in recent years, lithium-ion batteries have been increasingly developed towards higher specific energy and specific power, further reducing their thermal stability. Differential scanning calorimetry (DSC) is often used to interpret the endothermic and exothermic processes within the battery to assess battery safety and thermal runaway mechanisms. Extracting thermal runaway reaction kinetic parameters through DSC can further construct thermal runaway models for safety warnings. However, due to the complex reactions between electrode and electrolyte materials during thermal runaway, determining the optimal number of peaks is difficult. Existing mechanism extraction methods often rely on expert experience, resulting in large errors and poor versatility. Simultaneously, the numerous and interconnected parameters of thermal runaway reactions make extracting all reaction parameters from DSC extremely challenging. Therefore, it is necessary to find a method that can improve the accuracy, calculation speed and rationality of the mechanism to help improve the reliability of thermal runaway early warning.

[0003] Currently, commonly used methods for extracting thermal runaway mechanisms based on DSC (Digital Subtraction Angiography) are empirical formula methods and model-based methods. Empirical methods first obtain the peak temperature of each peak at different heating rates through manual peak division, and then use empirical formulas to identify the kinetic parameters of each peak. This method often uses Gaussian functions or other asymmetric Gaussian functions for manual peak division, but there are currently no standard rules to guide the selection of function forms. Furthermore, there may be deviations between the peaks obtained through the selected function and the kinetic mechanism, and errors accumulate during both manual peak division and parameter identification. Moreover, this method requires testing multiple material combinations at multiple DSC heating rates, resulting in high experimental costs. Model-based methods minimize the error between the model-predicted thermal effect and the tested thermal effect by constructing a least-squares model. It uses a global optimization algorithm to extract parameters. However, when there are multiple overlapping peaks in the DSC curve, the initial values ​​of the parameters, boundary conditions, and a reasonable number of peaks are difficult to determine. Currently, there is no model-based method that can effectively extract parameters from complex multi-peak overlapping DSC curves. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art by providing a method and apparatus for extracting battery thermal runaway reaction parameters with high computational efficiency and high accuracy.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for extracting parameters of battery thermal runaway reaction includes the following steps:

[0007] 1) Obtain the thermal runaway heat flux curve of the lithium-ion battery;

[0008] 2) The heat flux curve is adaptively divided into fitting regions to obtain multiple fitting regions;

[0009] 3) For each fitting region, the chemical reaction kinetic model is initialized based on the pre-built offline equivalent kinetic database, and the initialized chemical reaction kinetic model is used for fitting.

[0010] 4) Select the optimal solution for the reaction parameters based on the fitting results, wherein the reaction parameters include the number of peaks.

[0011] Further, in step 2), the adaptive fitting region division specifically includes:

[0012] Perform data smoothing on the heat flux profile;

[0013] Find local extrema in the smoothed heat flow curve, and define the boundary of each fitted region according to x% of the local extrema, where x is a positive number less than 10;

[0014] Based on the location of local extrema, the number of initial peaks in each fitting region is obtained.

[0015] Furthermore, the heat flow curve is smoothed using the moving average method.

[0016] Furthermore, the initial number of fitting peaks is the number of local extrema contained within the fitting region.

[0017] Furthermore, in step 3), the offline equivalent kinetic database stores the mapping relationship between the chemical reaction kinetic model and the Gaussian approximate kinetic model.

[0018] Furthermore, the initialization of the chemical reaction kinetic model based on the pre-built offline equivalent kinetic database is specifically as follows:

[0019] For each fitted region, a Gaussian approximate dynamic model is fitted to obtain a set of Gaussian parameters;

[0020] The Gaussian parameter set is input into an offline equivalent kinetic database to obtain the initial parameter set of the chemical reaction kinetic model, which includes the initial parameters and initial enthalpy change values ​​of the chemical reaction kinetic model.

[0021] Furthermore, the mapping relationship is obtained through a global optimization algorithm.

[0022] Further, in step 4), the selection of the optimal solution for the reaction parameters based on the fitting results specifically includes:

[0023] Using R 2 The termination criterion filters all feasible peak numbers. For each fitting region, if the termination criterion is not met, the number of peaks in the fitting region is increased by repeating the chemical reaction kinetic model fitting process in step 3) until the termination criterion is met and a feasible solution for the number of peaks is obtained.

[0024] Among all feasible solutions, the optimal solution is selected using an information-based Bayesian criterion;

[0025] Among them, R 2 As the determining factor, the termination criterion is: if R is the coefficient of determination for the current number of peaks N, then... 2 (N) satisfies R 2 (N)>0.995 or satisfy R 2 (N)>0.990 and R 2 (N-1)>0.990 and R 2 (N)>R 2 (N-1), then the maximum feasible solution for the number of peaks is the current number of peaks N; if R under the current number of peaks N 2 (N) satisfies R 2 (N)>0.990 and R 2 (N-1)>0.990 and R 2 (N) <R 2 If (N-1), then the maximum feasible solution for the number of peaks is N-1.

[0026] Furthermore, the Bayesian criterion is as follows:

[0027]

[0028] Where SSE is the sum of squared residuals of different feasible solutions, K represents the number of samples, p represents the number of model parameters, and d is the penalty exponent for the model parameters.

[0029] The present invention also provides a battery thermal runaway reaction parameter extraction device, comprising a microcontroller, a memory, a differential scanning calorimeter, and a kinetic calculator, wherein the memory is used to store one or more programs, the differential scanning calorimeter measures and obtains the thermal runaway heat flow curve of a lithium-ion battery under the control of the microcontroller, and the kinetic calculator calls the one or more programs under the control of the microcontroller to execute the battery thermal runaway reaction parameter extraction method as described in any one of claims 1-9.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. This invention develops a method for extracting parameters of the thermal runaway reaction of lithium-ion batteries based on the multi-peak overlapping DSC curves, so as to extract the thermal runaway kinetic reaction mechanism of lithium batteries. By constructing an offline equivalent kinetic database, effective initial parameter values ​​are provided for each reaction peak, thereby improving computational efficiency. An adaptive partitioning fitting method is used to reduce the computational difficulty and assist in peak extraction. The optimal number of peaks is achieved by using error and information-based criteria.

[0032] 2. This invention constructs an offline equivalent kinetic database, and uses Gaussian functions to initialize the parameters of the chemical reaction kinetic model effectively, solving the problem of getting trapped in local optima during the calculation process. Moreover, the parameters of the Gaussian approximation kinetic model have clear physical meanings, making them easier to identify and obtain, thus improving computational efficiency.

[0033] 3. This invention proposes an adaptive partitioning fitting method and a calculation termination criterion based on posterior error, combined with an information-based optimal solution selection criterion, to achieve automatic fitting of thermal runaway dynamics. This method requires no reliance on expert knowledge, has a fast calculation time (no more than 1000s), and high calculation accuracy (coefficient of determination R0). 2 (Not less than 0.99). Attached Figure Description

[0034] Figure 1 This is a schematic flowchart of the method of the present invention;

[0035] Figure 2 This is a schematic diagram of the structure of the device of the present invention;

[0036] Figure 3 This is a schematic diagram of the multi-peak overlapping DSC curve to which this invention applies;

[0037] Figure 4 This is a schematic diagram of the error-based termination criterion determination process of the present invention. Detailed Implementation

[0038] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0039] This invention provides a method for extracting parameters of battery thermal runaway reaction, such as... Figure 1 As shown, it includes the following steps:

[0040] 1) Obtain the thermal runaway heat flux curve of the lithium-ion battery;

[0041] 2) The heat flux curve is adaptively divided into fitting regions to obtain multiple fitting regions;

[0042] 3) For each fitting region, the chemical reaction kinetic model is initialized based on the pre-built offline equivalent kinetic database, and the initialized chemical reaction kinetic model is used for fitting.

[0043] 4) Select the optimal solution for the reaction parameters based on the fitting results, wherein the reaction parameters include the number of peaks.

[0044] In the steps described above, an offline equivalent kinetic database is pre-built to provide initial parameter values ​​for the Chemical Reaction Kinetic Model (CKM), reducing the risk of getting trapped in local optima during computation. This database is based on geometric similarity and incorporates a Gaussian approximation kinetic model (GAM). The advantage of GAMs is that their parameters have explicit physical meanings and are easier to identify. During database construction, a series of standard GAM models are first determined based on DSC test conditions, and then the CKM model is used to fit each standard GAM. The database input is the set of GAM parameters, and the output is the corresponding initial parameter values ​​for the CKM.

[0045] During thermal runaway, the reaction rate was described using the Arrhenius CKM:

[0046] Arrhenius equation:

[0047] Where A, E, and T represent the pre-exponential factor, activation energy, and temperature, respectively. R represents the ideal gas constant, with a value of 8.314.

[0048] Chemical reaction rates are related not only to temperature but also to reactant concentrations. This invention uses an exponential form to represent the effect of reactant concentration on the reaction rate: f(c) = c n

[0049] After considering the effects of temperature and concentration, the chemical reaction rate and the heat production rate can be expressed as:

[0050] Chemical reaction rate:

[0051] Heat generation rate:

[0052] Here, ΔH represents the enthalpy change of the reaction; the enthalpy change of an exothermic reaction is positive, and the enthalpy change of an endothermic reaction is negative. When multiple reactions occur simultaneously, the total heat production rate is the sum of the CKMs of each reaction. However, due to the large differences in the parameters of each CKM, they are difficult to identify. Therefore, this invention introduces a Gaussian approximate kinetic model:

[0053] Standard Gaussian function:

[0054] Among them, a, μ and σ are related to peak height, peak mean and peak width, respectively, and have relatively clear physical meanings.

[0055] DSC temperature rise relationship: T=kt+T0

[0056] Where k is the temperature rate of the DSC test, and T0 is the initial temperature. This can be obtained through integration:

[0057] Gaussian approximate enthalpy change:

[0058] Gaussian approximation dynamics:

[0059] The mapping relationship between GAM parameters [k,μ,σ] and CKM parameters [A,E,n] is obtained through global optimization algorithms such as genetic algorithms. The coefficient of determination (R²) 2 Used to evaluate the fitting effect:

[0060] Determination coefficient:

[0061] In the above steps, the adaptive fitting region division is based on prior local extremum information to divide the DSC curve into several fitting regions, so as to reduce the computational difficulty and assist in peak separation decision-making.

[0062] In a specific implementation, the adaptive fitting region division includes:

[0063] 201) The heat flow curve is smoothed by using the moving average method to reduce noise when searching for local extrema.

[0064] Moving average method:

[0065] Where y and yy represent the experimentally obtained and smoothed DSC heat flux curves, respectively. S represents the number of data points used to calculate the average value, and the value is an odd number.

[0066] 202) Find local extrema in the smoothed heat flux curve, and denote all found local extrema as [T]. l H l ](l=1:N l ), where N represents the peak temperature (position) and height, respectively. l The number of local extrema obtained.

[0067] 203) The regional boundary is determined according to the local extremum (H) l The region is defined by x%, where x is a positive value and generally does not exceed 10%. When dividing the region, if a local extremum is contained within an already divided region, the number of peaks in that region increases by 1. The initial number of peaks for the fitted region is the number of local extrema contained within the fitted region. The initial number of peaks for the m-th region is denoted as N. 0,m (ΣN 0,m =N0).

[0068] Step 3) involves performing partitioned fitting for each fitting region. Specifically, for each fitting region, GAM-based fitting, parameter initialization, and CKM-based fitting are performed sequentially.

[0069] First, based on GAM, a nonlinear least squares fitting method is used for fitting. During the fitting process:

[0070] Optimization goal:

[0071] Where q is the experimentally measured DSC heat flux value, q g This is the heat flux value simulated by GAM. N m This is the number of peaks in the fitted region, with an initial value of N. 0,m K m It is the number of points in the m-th fitted region.

[0072] Then, the parameter set [k,μ,σ] obtained through GAM fitting is input into the offline equivalent kinetic database to obtain the initial parameters [A0,E0,n0] of CKM. The GAM parameters [a,σ] are used to calculate the initial value of enthalpy change ΔH0.

[0073] Finally, based on the initialization parameters [A0,E0,n0,ΔH0], using N m A CKM model is fitted to this region, and its parameters are adjusted. The optimization objective of this process is defined as:

[0074] Optimization goal:

[0075] Where, q c It is the heat flux value simulated by GAM.

[0076] In step 4), for each region, the number of peaks (N)m The value of ) is not fixed. This invention proposes a method based on R. 2 The termination criteria screen all feasible peak counts, such as... Figure 4 As shown, R 2 As the determining factor, the termination criterion is: if R is the coefficient of determination for the current number of peaks N, then... 2 (N) satisfies R 2 (N)>0.995 or satisfy R 2 (N)>0.990 and R 2 (N-1)>0.990 and R 2 (N)>R 2 If the number of peaks is (N-1), then the maximum feasible solution for the number of peaks is the current number of peaks, i.e., N. max,m =N; if R = N; 2 (N) satisfies R 2 (N)>0.990 and R 2 (N-1)>0.990 and R 2 (N) <R 2 If (N-1), then the maximum feasible solution with the number of peaks is N. max,m =N-1. If the termination accuracy is not met, then the number of peaks in that region increases by one, and the above fitting process is repeated. The feasible solutions for all peak counts are denoted as [N]. 0,m N 0,m +1,...,N max,m ].

[0077] Among all feasible solutions, the optimal solution is selected by balancing model accuracy and complexity using an information-based Bayesian criterion.

[0078] Bayes' criterion:

[0079] Where SSE is the sum of squared residuals of different feasible solutions, which is related to the model longitude. K represents the number of samples, p represents the number of model parameters, and d is the penalty exponent for the model parameters, which takes a value greater than 1.

[0080] Number of model parameters: p = 4N A

[0081] Where, N A This represents the number of peaks in all regions after combining feasible solutions for each region.

[0082] This invention provides a precise, rapid, and automated method for extracting thermal runaway reaction parameters from DSC experimental results. By identifying prior extreme values ​​of the curves and using posterior fitting error information, a standardized peak number (reaction mechanism) optimization strategy is constructed. This eliminates the need for human intervention, enabling adaptive selection and calculation of the number of thermal runaway reaction peaks. This method is applicable to methods such as… Figure 3The reaction mechanism of the complex multi-peak overlapping DSC curves shown is extracted, enabling effective and rapid parameter extraction, which facilitates the prediction and alarm of battery thermal runaway. The kinetic parameters are initialized by constructing an equivalent kinetic database offline. The database introduces a series of standard Gaussian approximate kinetic models based on the DSC test range, constructs a least-squares optimization model, and establishes the relationship between the Gaussian approximate kinetic model and the chemical reaction kinetic model, thereby improving computational efficiency.

[0083] If the above methods 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.

[0084] Example 2

[0085] This embodiment provides a device for extracting parameters of battery thermal runaway reaction, such as... Figure 2 As shown, the system includes a microcontroller 100, a memory 102, a differential scanning calorimeter 104, and a kinetic calculator 106. The memory 102 stores one or more programs. The differential scanning calorimeter 104, under the control of the microcontroller, measures and obtains the thermal runaway heat flow curve of the lithium-ion battery. The kinetic calculator 106, under the control of the microcontroller, calls the one or more programs to execute the battery thermal runaway reaction parameter extraction method as described in Example 1, including an offline equivalent kinetic database and a partitioned fitting algorithm.

[0086] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0087] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0088] This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0089] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A battery thermal runaway reaction parameter extraction method, characterized in that, The method comprises the following steps: 1) obtaining a thermal runaway heat flow curve of a lithium ion battery; 2) performing adaptive fitting region division on the heat flow curve to obtain a plurality of fitting regions; 3) initializing a chemical reaction kinetics model based on a pre-constructed offline equivalent kinetics database for each fitting region, and performing fitting using the initialized chemical reaction kinetics model; 4) selecting an optimal solution of reaction parameters based on the fitting result, the reaction parameters including the number of peaks; In step 3), the offline equivalent kinetics database stores a mapping relationship between the chemical reaction kinetics model and the Gaussian approximation kinetics model; Initializing the chemical reaction kinetics model based on the pre-constructed offline equivalent kinetics database specifically comprises: For each fitting region, performing fitting based on the Gaussian approximation kinetics model to obtain a Gaussian parameter set; Inputting the Gaussian parameter set into the offline equivalent kinetics database to obtain an initial parameter set of the chemical reaction kinetics model, the initial parameter set including initial parameters of the chemical reaction kinetics model and an initial value of enthalpy change; In step 4), the optimal solution selection of reaction parameters based on the fitting result specifically comprises: All feasible peak numbers are screened using a termination criterion based on R 2 For each fitting region, the peak number is increased by one and the chemical reaction kinetic model fitting process of step 3) is repeated until the termination criterion is met, obtaining a feasible solution of the peak number. Among all feasible solutions, the optimal solution is selected based on the information-based Bayesian criterion; Among them, R 2 The determinant coefficient is used as the termination criterion: if the current number of peaks... N R below 2 ( N Satisfying R 2 ( N )>0.995 or satisfy R 2 ( N )>0.990 and R 2 ( N -1)>0.990 and R 2 ( N R 2 ( N -1), then the maximum feasible solution with the number of peaks is the current number of peaks. N If the current peak number N R below 2 ( N Satisfying R 2 ( N )>0.990 and R 2 ( N -1)>0.990 and R 2 ( N ) <R 2 ( N -1), then the maximum feasible solution with the number of peaks is N -1.

2. The battery thermal runaway reaction parameter extraction method of claim 1, wherein, In step 2), the adaptive fitting region division specifically comprises: Performing data smoothing processing on the heat flow curve; Finding local extrema in the smoothed heat flow curve, and dividing the region boundaries of each fitting region according to x% of the local extrema, x being a positive number less than 10; According to the positions of the local extrema, obtaining the fitting initial peak number of each fitting region.

3. The battery thermal runaway reaction parameter extraction method of claim 2, wherein, The data smoothing processing on the heat flow curve is performed by using the moving average method.

4. The battery thermal runaway reaction parameter extraction method of claim 2, wherein, The fitting initial peak number is the number of local extrema contained in the fitting region.

5. The battery thermal runaway reaction parameter extraction method of claim 1, wherein, The mapping relationship is obtained by a global optimization algorithm.

6. The battery thermal runaway reaction parameter extraction method of claim 1, wherein, The Bayesian criterion includes: where SSE is the sum of squared residuals of different feasible solutions, K represents the number of samples, p represents the number of model parameters, d is the penalty index of model parameters.

7. A battery thermal runaway reaction parameter extraction apparatus, characterized by, A microcontroller, a memory, a differential scanning calorimeter, and a kinetics calculator, wherein the memory is used to store one or more programs, the differential scanning calorimeter measures and obtains the thermal runaway heat flow curve of the lithium ion battery under the control of the microcontroller, and the kinetics calculator invokes the one or more programs to execute the battery thermal runaway reaction parameter extraction method according to any one of claims 1-6 under the control of the microcontroller.

Images