Multi-parameter collaborative optimization solid-state battery life prediction method

Abstract

The invention relates to a multi-parameter collaborative optimization solid-state battery life prediction method, which comprehensively considers main parameters influencing the life of a solid-state battery, including ionic conductivity, an electrode material, working temperature, charge-discharge rate and the like, and the key of life evaluation of the solid-state battery is capacity attenuation and internal resistance increase of the solid-state battery. By establishing a multi-parameter collaborative optimization mathematical model, the accuracy and reliability of solid-state battery life prediction are improved, the optimization algorithm is adopted to optimize the model parameters, the optimal parameter combination can be quickly found, and the prediction efficiency is improved. The prediction method provided by the invention has relatively strong generalization ability, can be suitable for solid-state batteries of different types and specifications, and provides powerful support for design, manufacturing and application of the solid-state batteries.

Description

Technical Field

[0001] The present invention belongs to the field of solid-state battery technology, and in particular relates to a solid-state battery life prediction method for multi-parameter collaborative optimization Background Art

[0002] As a new type of battery technology, solid-state batteries have the advantages of high energy density, good safety, and fast charging speed, and are considered to be the development direction of future battery technology. However, the life of solid-state batteries is affected by many factors, such as the performance of solid electrolytes, the performance of electrode materials, the operating temperature of the battery, the charge and discharge rate, etc., which makes it difficult to accurately predict the life of solid-state batteries. At present, the existing battery life prediction methods mainly include experimental methods, methods based on physical models, and data-driven methods. Although the experimental method can directly obtain the battery life data, it is time-consuming and laborious, and it is difficult to consider all the influencing factors; the method based on physical models requires an in-depth understanding of the internal mechanism of the battery, and the determination of model parameters is relatively complicated; the data-driven method requires a large amount of experimental data to train the model, and the generalization ability of the model is limited. Therefore, a solid-state battery life prediction method that can comprehensively consider the influence of multiple factors is needed to improve the accuracy and reliability of the prediction. Summary of the invention

[0003] The purpose of the present invention is to provide a multi-parameter collaborative optimization method for predicting the life of solid-state batteries. By establishing a mathematical model, multiple parameters affecting the life of solid-state batteries are collaboratively optimized, thereby improving the accuracy of solid-state battery life prediction.

[0004] Based on current research, the main factors (parameters) affecting the life of solid-state batteries include ionic conductivity, electrode materials, operating temperature, and charge and discharge rates. The key to life assessment of solid-state batteries lies in their capacity decay and internal resistance growth. Generally speaking, when the capacity of a solid-state battery decays to a certain extent, it is considered scrapped; or when the internal resistance grows to a certain extent, it can also be used as a condition for the solid-state battery to be scrapped.

[0005] In order to achieve the above objectives, the technical solution of the present invention is: a multi-parameter collaborative optimization solid-state battery life prediction method, characterized in that it includes the following steps:

[0006] Step 1: Co-optimize the parameters that affect the life of solid-state batteries and build a life prediction model. The parameters include ionic conductivity, electrode materials, operating temperature, and charge and discharge rate. The key to the life assessment of solid-state batteries lies in their capacity decay and internal resistance growth; the capacity decay model C(n) = C 0 (1-αn β ), where n represents the number of charge and discharge cycles, C 0represents the initial capacity of the battery, and the relationship between the ionic conductivity σ and the capacity attenuation coefficient α is where a 1 and m 1 Combined with experimental data fitting to obtain; for the capacity decay model C(n) = C 0 (1-αn β ) in the equation, let its relationship with ionic conductivity be where b 1 and k 1 is a constant; suppose the relationship between the specific capacity Q of the electrode material and α is Here a 2 and m 2 Combined with experimental data fitting, the lower the specific capacity of the electrode material, the faster the capacity decay; the relationship between β and Q is: b 2 and k 2 is the corresponding constant; considering the form of the Arrhenius equation, where a 3 is the prefactor, E a1 is the activation energy related to capacity decay, R is the gas constant, and T is the operating temperature; for β, let b 3 is a constant, E a2 is the activation energy associated with β; where a 4 and m 4 is a constant. The higher the charge and discharge rate, the greater the capacity attenuation coefficient. b 4 and k 4 is a constant.

[0007] Taking the above factors into consideration, the expression of capacity attenuation coefficient α is:

[0008]

[0009] Taking all the above factors into consideration,

[0010] The expression of capacity attenuation coefficient α is:

[0011]

[0012] The expression of β is:

[0013]

[0014] Internal resistance growth model R(n) = R 0 (1+γn δ ), where n represents the number of charge and discharge cycles, R 0represents the initial internal resistance value, and the relationship between the internal resistance growth coefficient γ and the ionic conductivity σ is c 1 and n 1 is a constant. The lower the ionic conductivity, the faster the internal resistance increases. For the internal resistance growth model, R(n) = R 0 (1+γn δ ) in δ, assuming d 1 and p 1 is a constant; c 2 and n 2 is a constant, the lower the specific capacity of the electrode material, the greater the internal resistance growth coefficient; d 2 and p 2 is a constant; where c 3 is the prefactor, E a3 is the activation energy associated with the growth of internal resistance; d 3 is a constant, E a4 is the activation energy associated with δ; c 4 and n 4 is a constant. The higher the charge and discharge rate, the faster the internal resistance increases. d 4 and p 4 is a constant;

[0015] Taken together,

[0016] The expression of internal resistance growth coefficient γ is:

[0017]

[0018] The expression of δ is:

[0019]

[0020] The life prediction model assumes that the end-of-life condition of the solid-state battery is that the capacity decays to x% of the initial capacity or the internal resistance increases to y% of the initial internal resistance. The life N of the solid-state battery can be expressed as:

[0021]

[0022] Substituting the above expressions of α, β, γ, and δ into the life prediction model, a complete mathematical model for solid-state battery life prediction based on multiple parameters is obtained;

[0023] Step 2: Collect parameter information and input the collected parameter information into the life prediction model to complete the solid-state battery life prediction.

[0024] Preferably, in the life prediction model in step 1, the life N of the solid-state battery is affected by parameters, wherein the capacity decay part involves parameters α, β, n and x%, and the internal resistance growth part involves parameters γ, δ, n and y%, and α, β, γ and δ are related to ionic conductivity σ, electrode material specific capacity Q, operating temperature T, and charge and discharge rate C. The model is optimized with minimizing the life prediction error as the optimization goal, and an optimization objective function is established. Among them C pred,i and R pred,i is the capacity and internal resistance predicted based on the current parameter combination, C actual,i and R actual,i It is the actual measured capacity and internal resistance.

[0025] Preferably, the steps for optimizing the life prediction model are as follows: 1) data acquisition, 2) data preprocessing, and 3) model parameter estimation.

[0026] Preferably, data collection includes: using an electrochemical impedance spectrometer to measure the ionic conductivity of the solid-state battery at different temperatures and charge and discharge states to obtain a set of ionic conductivity data σ 1 ,σ 2 ,…,σ n ; Through charge and discharge experiments, the specific capacity of the electrode material is measured to obtain the specific capacity data Q 1 ,Q 2 ,…,Q n During the battery charging and discharging process, the temperature sensor is used to record the working temperature and obtain the temperature data T 1 ,T 2 ,…,T n ; Set different charge and discharge rates, use the charge and discharge equipment to record the charge and discharge data of the battery at different rates, and calculate the charge and discharge rate data C 1 ,C 2 ,…,C n .

[0027] Preferably, data preprocessing specifically includes checking whether there are outliers in the data, such as data points that obviously deviate from the normal range. If there are outliers, the median filtering method is used to remove them; the min-max normalization method is used to normalize the data such as ionic conductivity, specific capacity, temperature and charge and discharge rate, and each data point is mapped to the interval [0,1]. Assuming that the original data is X={x 1 ,x 2 ,…,x n}, the normalized data is

[0028]

[0029] Preferably, the model parameter estimation is to perform multi-parameter identification on the solid-state battery life assessment model by an improved adaptive genetic algorithm, the main improvement of which includes an adaptive crossover probability (p c ) and mutation probability (p m ) and the introduction of catastrophe operations and return mechanisms.

[0030] Preferably, an improved adaptive genetic algorithm comprises the following steps:

[0031] ① Coding: Binary code the parameters to be optimized, set the parameter value range based on the empirical value of the parameter, and each range is represented by a binary string of a certain length;

[0032] ② Initialize the population: According to the binary coding method, a certain number of individuals are randomly generated to form the initial population. Each individual represents a set of possible parameter combinations.

[0033] ③ Calculate fitness: For each individual, decode its corresponding parameters and substitute them into the mathematical model of life prediction

[0034] , calculate its fitness value;

[0035] Among them, C 0 ,α,β,γ,δ are obtained by encoding and decoding of individuals;

[0036] ④ Selection operator: According to the fitness value of the individual, the roulette selection operator is used to select the individuals of the next generation population; Roulette selection: According to the fitness value of the individual, the proportion of the individual on the roulette wheel is calculated, and then a random selection is performed. The higher the fitness value, the greater the probability of the individual being selected; ⑤ Crossover operation, design adaptive crossover probability adjustment, and the crossover probability is dynamically adjusted according to the adaptive formula; when the fitness value of the individual is large and the number of individuals in the population whose fitness value is greater than the average fitness value is large, the crossover probability will be reduced accordingly to protect the excellent gene combination; the adaptive crossover probability and mutation probability formula are designed as follows:

[0037]

[0038] The values ​​"0.8" and "0.1" are empirical values. max Represents the maximum fitness value in the population, f avg represents the average fitness value of the group, f represents the larger value of the fitness value of the two crossed individuals, f' represents the fitness value of the individual to be mutated, and f min Represents the minimum fitness value in the group; M 1 Represents the number of individuals in each generation whose fitness value is greater than the average fitness value, M 2represents the number of individuals in each generation whose fitness value is less than the average fitness value; λ is an infinitesimal positive number, mainly to prevent the denominator from being equal to 0; other parameters can be defined according to empirical values. For example, p c1 = 0.9, p c2 = 0.6, p m1 = 0.1, p m2 = 0.01; perform crossover operation on the selected individuals to generate new individuals;

[0039] ⑥ Mutation operation, design an adaptive mutation probability adjustment. The mutation probability is adjusted according to the adaptive formula; when the fitness value of an individual is small or the population shows a premature trend, the mutation probability will increase to increase the diversity of the population and avoid falling into local optimum; the mutation probability formula is:

[0040]

[0041] where f' represents the fitness value of the individual to be mutated, and the value "0.1" is an empirical value. The meanings of other parameters are the same as those in the crossover probability formula; p m1 = 0.1, p m2 = 0.01, perform mutation operation on the newly generated individuals to introduce new genes;

[0042] ⑦ Catastrophe operation judgment. Assume that the number of iterations is M. After the m-th (m < M) generation, most individuals have a high similarity with the optimal individual (only the optimal individual and the worst individual have a low similarity). At this time, perform the following operations to generate a new population: save the current population state, the optimal individual and its fitness value; retain a small number of individuals with poor fitness values; delete most of the individuals with excellent fitness values in the population; randomly generate new individuals for supplementation;

[0043] a. Judgment condition:

[0044] After an appropriate number of iterations, judge whether the conditions for the catastrophe operation are met; it can be judged according to factors such as the diversity index of the population and the distribution of fitness values. For example, the similarity between individuals in the population can be calculated. If most individuals have a high similarity with the optimal individual (only the optimal individual and the worst individual have a low similarity), it is considered that the conditions for the catastrophe operation are met;

[0045] b. Mathematical description:

[0046] Assume that the catastrophe operation judgment is performed when the number of iterations reaches T. Calculate the similarity between individuals in the population using the Euclidean distance method. If the Euclidean distance between most individuals and the optimal individual is less than a certain threshold, it is considered that the conditions for the catastrophe operation are met;

[0047] ⑧ Return mechanism: In order to improve the global search performance of the algorithm and the traversal of the search, a return operation is added; when the crossover and mutation operations fail, that is, when the average fitness level of the newly generated population is lower than the average fitness level of the original population, the operation is negated and the population is restored to the previous state;

[0048] ⑨ Iteration: Repeat steps ③ to ⑧ until the stopping condition is met, such as reaching the maximum number of iterations or the fitness value reaches a certain accuracy requirement.

[0049] The technical effects of the present invention are:

[0050] 1. The present invention comprehensively considers the impact of various factors on the life of solid-state batteries, and improves the accuracy and reliability of solid-state battery life prediction by establishing a mathematical model for multi-parameter collaborative optimization. At the same time, the optimization algorithm is used to optimize the model parameters, which can quickly find the optimal parameter combination and improve the prediction efficiency; the prediction method of the present invention has a strong generalization ability and can be applied to solid-state batteries of different types and specifications, providing strong support for the design, manufacture and application of solid-state batteries.

[0051] 2. By continuously adjusting the crossover probability and mutation probability, as well as executing catastrophic operations and return mechanisms, the algorithm can better balance global search and local search, avoid premature convergence, and more effectively find the optimal combination of multiple parameters, so that the life prediction model can more accurately predict the life of solid-state batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of an improved adaptive genetic algorithm. DETAILED DESCRIPTION

[0053] The technical solution of the present invention will be clearly and completely described below in conjunction with embodiments.

[0054] Unless otherwise specified, the descriptions of "first", "second", etc. in the present invention are used to distinguish different objects, and are not used to indicate meanings such as size or timing, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second", etc. may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, the meaning of "plurality" is two or more.

[0055] The term "and / or" in the present invention is only a description of the association relationship of associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, B exists alone, and A and B exist at the same time. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0056] Based on current research, the main factors (parameters) that affect the life of solid-state batteries include ionic conductivity, electrode materials, operating temperature, and charge and discharge rates. The key to evaluating the life of solid-state batteries lies in their capacity decay and internal resistance growth. Generally speaking, when the capacity of a solid-state battery decays to a certain extent, it is considered scrapped; or when the internal resistance grows to a certain extent, it can also be used as a condition for the solid-state battery to be scrapped. The mathematical modeling of the life evaluation of solid-state batteries is described in detail as follows:

[0057] 1) Consider capacity attenuation

[0058] Capacity fading model C(n) = C 0 (1-αn β ), where n represents the number of charge and discharge cycles, C 0 Indicates the initial capacity of the battery. Other parameters are described in detail below.

[0059] 1. Effect of ionic conductivity on capacity fading

[0060] Assume that the relationship between ionic conductivity σ and capacity attenuation coefficient α is where a 1 and m 1 It needs to be obtained by combining experimental data fitting. For the capacity decay model C(n) = C 0 (1-αn β ) in the equation, let its relationship with ionic conductivity be where b 1 and k 1 is a constant.

[0061] 2. Effect of electrode material specific capacity on capacity decay

[0062] Assume that the relationship between the specific capacity Q of the electrode material and α is Here a 2 and m 2 It needs to be obtained by fitting with experimental data. The lower the specific capacity of the electrode material, the faster the capacity decay. At the same time, the relationship between β and Q is b 2 and k 2 is the corresponding constant.

[0063] 3. Impact of operating temperature on capacity attenuation

[0064] Consider the Arrhenius equation in the form, where a 3 is the prefactor, E a1 is the activation energy associated with capacity fade, R is the gas constant, and T is the operating temperature.

[0065] For β, let b 3 is a constant, E a2 is the activation energy associated with β.

[0066] 4. Effect of charge and discharge rate on capacity attenuation

[0067] where a 4 and m 4 is a constant. The higher the charge and discharge rate, the greater the capacity attenuation coefficient. b 4 and k 4 is a constant.

[0068] Taking the above factors into consideration, the expression of capacity attenuation coefficient α is:

[0069]

[0070] Taking all the above factors into consideration,

[0071] The expression of capacity attenuation coefficient α is:

[0072]

[0073] The expression of β is:

[0074]

[0075] 2) Consider the internal resistance growth

[0076] Internal resistance growth model R(n) = R 0 (1+γn δ ), where n represents the number of charge and discharge cycles, R 0 Indicates the initial internal resistance value. Other parameters are described in detail below.

[0077] 1. Effect of ionic conductivity on internal resistance growth

[0078] Assume that the relationship between the internal resistance growth coefficient γ and the ionic conductivity σ is c 1 and n 1 is a constant, the lower the ionic conductivity, the faster the internal resistance increases.

[0079] For the internal resistance growth model R(n) = R0 (1+γn δ ) in δ, assuming d 1 and p 1 is a constant.

[0080] 2. Effect of electrode material specific capacity on internal resistance growth

[0081] c 2 and n 2 is a constant. The lower the specific capacity of the electrode material, the greater the internal resistance growth coefficient. d 2 and p 2 is a constant.

[0082] 3. The influence of operating temperature on internal resistance growth

[0083] where c 3 is the prefactor, E a3 is the activation energy associated with the growth of internal resistance. d 3 is a constant, E a4 is the activation energy associated with δ.

[0084] 4. Effect of charge and discharge rate on internal resistance growth

[0085] c 4 and n 4 is a constant. The higher the charge and discharge rate, the faster the internal resistance increases.

[0086] d 4 and p 4 is a constant.

[0087] Taken together,

[0088] The expression of internal resistance growth coefficient γ is:

[0089]

[0090] The expression of δ is:

[0091]

[0092] Lifespan prediction model

[0093] Assuming that the end-of-life condition of a solid-state battery is that the capacity decays to x% of the initial capacity or the internal resistance increases to y% of the initial internal resistance, the life N of the solid-state battery can be expressed as:

[0094]

[0095] Substituting the above expressions of α, β, γ, and δ into the life prediction model, a complete mathematical model for solid-state battery life prediction based on multiple parameters is obtained. This model can comprehensively consider the synergistic effects of multiple parameters such as ionic conductivity, electrode material specific capacity, operating temperature, and charge and discharge rate on the life of solid-state batteries. The capacity decay model and internal resistance growth model in the model respectively introduce reasonable physical and chemical mechanisms related to each parameter. For example, the capacity decay model considers the effect of temperature on the reaction rate in the form of the Arrhenius equation, and the internal resistance growth model similarly considers the mechanism of action of each factor, so that the model is not only the result of data fitting, but also has clear physical meaning and theoretical support, which helps to deeply understand the internal reasons for battery life decay. The model gives clear analytical expressions for capacity decay coefficients α, β and internal resistance growth coefficients γ, δ, which describe in detail the quantitative relationship between each parameter and capacity decay and internal resistance growth. Through these analytical expressions, the degree of influence of different parameter changes on battery life can be analyzed more intuitively, and specific theoretical guidance can be provided for the optimization of battery design and use conditions.

[0096] Parameter collaborative algorithm design

[0097] In the above life prediction model, the life N of the solid-state battery is affected by multiple parameters. Among them, the capacity decay part involves parameters α, β, n and x%, and the internal resistance growth part involves parameters γ, δ, n and y%. α and β are related to parameters such as ionic conductivity σ, electrode material specific capacity Q, operating temperature T, and charge and discharge rate C. The specific relationship is as described above (for example

[0098] ). Similarly, γδ is also related to these parameters.

[0099] Taking minimizing the life prediction error as the optimization goal, the objective function is established Among them C pred,i and R pred,i is the capacity and internal resistance predicted based on the current parameter combination, C actual,i and R actual,i It is the actual measured capacity and internal resistance.

[0100] 1. Data Collection

[0101] The ionic conductivity of the solid-state battery at different temperatures and charge and discharge states was measured using an electrochemical impedance spectrometer to obtain a set of ionic conductivity data σ 1 ,σ 2 ,…,σ n ;

[0102] Through charge and discharge experiments, the specific capacity of the electrode material is measured to obtain the specific capacity data Q 1 ,Q2 ,…,Q n ;

[0103] During the battery charging and discharging process, the temperature sensor is used to record the working temperature and obtain the temperature data T 1 ,T 2 ,…,T n ;

[0104] Set different charge and discharge rates, use the charge and discharge equipment to record the charge and discharge data of the battery at different rates, and calculate the charge and discharge rate data C 1 ,C 2 ,…,C n .

[0105] 2. Data preprocessing

[0106] Check whether there are outliers in the data, such as data points that are significantly out of the normal range. If there are outliers, use the median filter method to remove them.

[0107] The min-max normalization method is used to normalize data such as ionic conductivity, specific capacity, temperature, and charge and discharge rate, mapping each data point to the [0,1] interval.

[0108] Assume that the original data is X = {x 1 ,x 2 ,…,x n}, the normalized data is

[0109]

[0110] 3. Model parameter estimation

[0111] In order to complete the multi-parameter identification of the solid-state battery life evaluation model mentioned above, an improved adaptive genetic algorithm (IAGA) is designed. Its main improvements include adaptive crossover probability (p c ) and mutation probability (p m ) and the introduction of catastrophic operations and return mechanisms to avoid premature convergence and improve the traversability of the search.

[0112] like Figure 1 As shown, the specific algorithm design steps are as follows:

[0113] ① Coding: Binary code the parameters to be optimized, set the parameter value range based on the empirical value of the parameter, and each range is represented by a binary string of a certain length;

[0114] ② Initialize the population: According to the binary coding method, a certain number of individuals are randomly generated to form the initial population. Each individual represents a set of possible parameter combinations; for example, for the objective function, the preprocessed data is substituted into the expression of the capacity attenuation coefficient.

[0115] and

[0116] An initial set of parameter combinations will be randomly generated:

[0117] (a 1 ,a 2 ,a 3 ,a 4 ,m 1 ,m 2 ,E a1 ,m 4 ,b 1 ,b 2 ,b 3 ,b 4 ,k 1 ,k 2 ,E a2 ,k 4 )

[0118] ③ Calculate fitness: For each individual, decode its corresponding parameters and substitute them into the mathematical model of life prediction

[0119] , calculate its fitness value.

[0120] Among them, C 0 ,Parameters such as α, β, γ, and δ are obtained by the encoding and decoding of the ,individual.

[0121] ④ Selection operator: According to the fitness value of the individual, the roulette selection operator is used to select the individuals of the next generation population. Roulette selection: According to the fitness value of the individual, the proportion of the individual on the roulette wheel is calculated, and then a random selection is performed. The higher the fitness value, the greater the probability of the individual being selected. ⑤ Crossover operation, design adaptive crossover probability adjustment, and the crossover probability is dynamically adjusted according to the adaptive formula. For example, when the fitness value of the individual is large and there are many individuals in the population whose fitness value is greater than the average fitness value, the crossover probability will be reduced accordingly to protect the excellent gene combination.

[0122] The adaptive crossover probability and mutation probability formulas are designed as follows:

[0123]

[0124] The value "0.8" is an empirical value, f max Represents the maximum fitness value in the population, f avgrepresents the average fitness value of the population, f represents the larger value of the fitness values of the two individuals to be crossed, f' represents the fitness value of the individual to be mutated, f min represents the minimum fitness value in the population; M 1 represents the number of individuals in each generation whose fitness values are greater than the average fitness value, M 2 represents the number of individuals in each generation whose fitness values are less than the average fitness value; λ is an infinitesimal positive number, mainly to prevent the denominator from being equal to 0; other parameters can be defined according to empirical values. For example, p c1 = 0.9, p c2 = 0.6, p m1 = 0.1, p m2 = 0.01. Perform crossover operations on the selected individuals to generate new individuals.

[0125] ⑥ Mutation operation, design an adaptive mutation probability adjustment, and the mutation probability is also adjusted according to the adaptive formula. When the fitness value of an individual is small or the population shows a premature trend, the mutation probability will increase to increase the diversity of the population and avoid falling into a local optimum. The mutation probability formula is:

[0126]

[0127] Among them, f' represents the fitness value of the individual to be mutated, the value "0.1" is an empirical value, and the meanings of other parameters are the same as those in the crossover probability formula; p m1 = 0.1, p m2 = 0.01. Perform mutation operations on the newly generated individuals to introduce new genes.

[0128] ⑦ Catastrophe operation judgment. Assume that the number of iterations is M. After the m-th (m < M) generation, most individuals have a high similarity to the optimal individual (only the optimal individual and the worst individual have a low similarity). At this time, perform the following operations to generate a new population: Save the current population state, the optimal individual and its fitness value; retain a small number of individuals with poor fitness values; delete most of the individuals with excellent fitness values in the population; randomly generate new individuals for supplementation.

[0129] a. Judgment conditions:

[0130] After an appropriate number of iterations, judge whether the conditions for the catastrophe operation are met. Usually, it can be judged according to factors such as the diversity index of the population and the distribution of fitness values. For example, the similarity between individuals in the population can be calculated. If most individuals have a high similarity to the optimal individual (only the optimal individual and the worst individual have a low similarity), it is considered that the conditions for the catastrophe operation are met.

[0131] b. Mathematical description:

[0132] Assume that the catastrophic operation is judged when the number of iterations reaches T. Calculate the similarity between individuals in the population using the Euclidean distance method. If the Euclidean distance between most individuals and the optimal individual is less than a certain threshold, it is considered that the conditions for catastrophic operation are met. For example, there is a population with a population size of N = 50, and the catastrophic operation is judged when the number of iterations reaches T = 100. Calculate the Euclidean distance between each individual and the optimal individual, and find that the Euclidean distance between 40 individuals and the optimal individual is less than 0.5, then it is considered that the conditions for catastrophic operation are met. Perform the catastrophic operation, retain the 10 individuals with the worst fitness values, delete the remaining 40 individuals, and then randomly generate 40 new individuals to form a new population.

[0133] ⑧ Return mechanism: In order to improve the global search performance of the algorithm and the traversal of the search, a return operation is added. When the crossover and mutation operations fail, that is, when the average (optimal) fitness level of the newly generated population is lower than the average (optimal) fitness level of the original population, the operation is negated and the population is restored to the previous state. For example, after a crossover and mutation operation, the newly generated population has N = 50 individuals, and its average fitness value is calculated as f' avg The average fitness value of the original population is f avg If f' avg <f avg , then execute the return operation to restore to the previous population state.

[0134] ⑨ Iteration: Repeat steps ③ to ⑧ until the stopping condition is met, such as reaching the maximum number of iterations or the fitness value reaches a certain accuracy requirement. Throughout the process, by continuously adjusting the crossover probability and mutation probability, as well as performing catastrophic operations and return mechanisms, the algorithm can better balance global search and local search, avoid premature convergence, and more effectively find the optimal combination of multiple parameters, so that the life prediction model can more accurately predict the life of solid-state batteries.

[0135] The present invention comprehensively considers the impact of various factors on the life of solid-state batteries, and improves the accuracy and reliability of solid-state battery life prediction by establishing a mathematical model for multi-parameter collaborative optimization. At the same time, the optimization algorithm is used to optimize the model parameters, which can quickly find the optimal parameter combination and improve the prediction efficiency. The prediction method of the present invention has a strong generalization ability and can be applied to solid-state batteries of different types and specifications, providing strong support for the design, manufacture and application of solid-state batteries.

[0136] It should be noted that the sequence of the embodiments of the present invention described above is for description only and does not represent the advantages and disadvantages of the embodiments. The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0137] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0138] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-parameter collaborative optimization method for predicting the life of a solid-state battery, characterized in that: It includes the following steps: Step 1: Co-optimize the parameters affecting the life of the solid-state battery, and construct a life prediction model. The parameters include ionic conductivity, electrode material, operating temperature, and charge-discharge rate. The key to the life assessment of the solid-state battery lies in its capacity attenuation and internal resistance growth; Capacity decay model C(n) = C0(1-αn β ), where n represents the number of charge and discharge cycles, C0 represents the initial capacity of the battery, and the relationship between the ionic conductivity σ and the capacity attenuation coefficient α is: Where a1 and m1 are obtained by fitting the experimental data; for the capacity decay model C(n) = C0(1-αn β ) in the equation, let its relationship with ionic conductivity be Where b1 and k1 are constants; the relationship between the specific capacity Q of the electrode material and α is: Here a2 and m2 are obtained by fitting experimental data; the lower the specific capacity of the electrode material, the faster the capacity decay; the relationship between β and Q is b2 and k2 are corresponding constants; considering the form of the Arrhenius equation, Where a3 is the pre-factor, E a1 is the activation energy related to capacity decay, R is the gas constant, and T is the operating temperature; for β, let b3 is a constant, E a2 is the activation energy associated with β; Among them, a4 and m4 are constants. The higher the charge and discharge rate, the greater the capacity attenuation coefficient. b4 and k4 are constants. Considering the above factors, the expression of the capacity attenuation coefficient α is: Considering the above factors, The expression of the capacity attenuation coefficient α is: The expression of β is: Internal resistance growth model R(n) = R0(1+γn δ ), where n represents the number of charge and discharge cycles, R0 represents the initial internal resistance, and the relationship between the internal resistance growth coefficient γ and the ionic conductivity σ is: c1 and n1 are constants. The lower the ionic conductivity, the faster the internal resistance increases. For the internal resistance growth model, R(n) = R0(1+γn δ ) in δ, assuming d1 and p1 are constants; c2 and n2 are constants. The lower the specific capacity of the electrode material, the greater the internal resistance growth coefficient. d2 and p2 are constants; Where c3 is the pre-factor, E a3 is the activation energy associated with the growth of internal resistance; d3 is a constant, E a4 is the activation energy associated with δ; c4 and n4 are constants. The higher the charge and discharge rate, the faster the internal resistance increases. d4 and p4 are constants; Taken together, The expression of the internal resistance growth coefficient γ is: d 的表达式为： Life prediction model Assume that the life termination condition of the solid-state battery is that the capacity decays to x% of the initial capacity or the internal resistance grows to y% of the initial internal resistance. Then the life N of the solid-state battery can be expressed as: Substitute the expressions of the above α, β, γ, and δ into the life prediction model to obtain a complete mathematical model for predicting the life of the solid-state battery based on multiple parameters; Step 2: Collect parameter information and input the collected parameter information into the life prediction model to complete the life prediction of the solid-state battery.

2. The multi-parameter collaborative optimization solid-state battery life prediction method according to claim 1 is characterized in that: In the life prediction model in step 1, the life N of the solid-state battery is affected by parameters, among which the capacity decay part involves parameters α, β, n and x%, and the internal resistance growth part involves parameters γ, δ, n and y%. α, β, γ and δ are related to ionic conductivity σ, electrode material specific capacity Q, operating temperature T, and charge and discharge rate C. The model is optimized with minimizing the life prediction error as the optimization goal, and the optimization objective function is established. Among them C pred,i and R pred,i is the capacity and internal resistance predicted based on the current parameter combination, C actual,i and R actual,i It is the actual measured capacity and internal resistance.

3. The multi-parameter collaborative optimization solid-state battery life prediction method according to claim 2 is characterized in that: The optimization steps for the life prediction model are as follows: 1) Data collection, 2) Data preprocessing, 3) Model parameter estimation.

4. The multi-parameter collaborative optimization solid-state battery life prediction method according to claim 3 is characterized in that: Data collection includes: using an electrochemical impedance spectrometer to measure the ionic conductivity of the solid-state battery at different temperatures and charge and discharge states, and obtaining a set of ionic conductivity data σ1, σ2, …, σ n ; Through charge and discharge experiments, the specific capacity of the electrode material is measured to obtain specific capacity data Q1, Q2, ..., Q n ; During the battery charging and discharging process, the temperature sensor is used to record the working temperature and obtain the temperature data T1, T2, ..., T n ; Set different charge and discharge rates, use the charge and discharge equipment to record the charge and discharge data of the battery at different rates, and calculate the charge and discharge rate data C1, C2, ..., C n .

5. The multi-parameter collaborative optimization solid-state battery life prediction method according to claim 4 is characterized in that: Data preprocessing specifically includes checking whether there are outliers in the data, such as data points that obviously deviate from the normal range. If there are outliers, the median filtering method is used to remove them; the min-max normalization method is used to normalize the data such as ionic conductivity, specific capacity, temperature and charge and discharge rate, and each data point is mapped to the interval [0,1]. Assuming that the original data is X = {x1, x2,…, x n }, the normalized data is 6. The multi-parameter collaborative optimization solid-state battery life prediction method according to claim 5, characterized in that: Model parameter estimation is to identify the multi-parameters of the solid-state battery life assessment model through an improved adaptive genetic algorithm. The main improvements include adaptive crossover probability (p c ) and mutation probability (p m ) and the introduction of catastrophe operations and return mechanisms.

7. The multi-parameter collaborative optimization solid-state battery life prediction method according to claim 5, characterized in that: The steps of an improved adaptive genetic algorithm are as follows: ① Encoding: Binary encoding is performed on the parameters to be optimized. According to the empirical values of the parameters, the value range of the parameters is set, and each range is represented by a binary string of a certain length; ② Initializing the population: According to the binary encoding method, a certain number of individuals are randomly generated to form an initial population, and each individual represents a possible combination of parameters; ③ Calculating fitness: For each individual, after decoding the corresponding parameters, substitute them into the mathematical model of life prediction , calculate its fitness value; Among them, C0, α, β, γ, and δ are obtained by encoding and decoding of the individual; ④ Selection operator: According to the fitness values of the individuals, use the roulette wheel selection operator to select the individuals of the next generation population; Roulette wheel selection: Calculate the proportion of each individual's fitness value on the roulette wheel, and then perform random selection. The higher the fitness value of an individual, the greater the probability of being selected; ⑤ Crossover operation, design an adaptive crossover probability adjustment. The crossover probability is dynamically adjusted according to the adaptive formula; When the fitness value of an individual is large and the number of individuals with fitness values greater than the average fitness value in the population is large, the crossover probability will be correspondingly reduced to protect excellent gene combinations; The adaptive crossover probability and mutation probability formulas are designed as follows: Among them, the values ​​"0.8" and "0.1" are empirical values, f max Represents the maximum fitness value in the population, f avg represents the average fitness value of the group, f represents the larger value of the fitness value of the two crossed individuals, f' represents the fitness value of the individual to be mutated, and f min represents the minimum fitness value in the population; M1 represents the number of individuals in each generation whose fitness value is greater than the average fitness value, and M2 represents the number of individuals in each generation whose fitness value is less than the average fitness value; λ is an infinitesimal positive number, mainly to prevent the denominator from being equal to 0; other parameters can be defined according to empirical values, for example, p c1 =0.9,p c2 =0.6,p m1 =0.1,p m2 =0.01; perform crossover operation on the selected individuals to generate new individuals; ⑥ Mutation operation, design an adaptive mutation probability adjustment. The mutation probability is adjusted according to the adaptive formula; When the fitness value of an individual is small or the population shows a premature trend, the mutation probability will increase to increase the diversity of the population and avoid falling into local optimum; The mutation probability formula is: Among them, f' represents the fitness value of the individual to be mutated, the value "0.1" is an empirical value, and the other parameters have the same meaning as in the crossover probability formula; p m1 =0.1,p m2 =0.01, perform mutation operation on the newly generated individuals and introduce new genes; ⑦ Catastrophe operation judgment, assume that the number of iterations is M. After the m-th (m < M) generation, most individuals have a high similarity to the optimal individual (only the optimal individual and the worst individual have a low similarity). At this time, perform the following operations to generate a new population: Save the current population state, the optimal individual and its fitness value; Retain a small number of individuals with poor fitness values; Delete most of the individuals with excellent fitness values in the population; Randomly generate new individuals for supplementation; a. Judgment condition: After an appropriate number of iterations, determine whether the conditions for the catastrophic operation are met; this can be determined based on factors such as the diversity index of the population and the distribution of fitness values. For example, the similarity between individuals in the population can be calculated. If most individuals have a high similarity to the best individual (only the best individual and the worst individual have a low similarity), then the conditions for the catastrophic operation are considered to be met. b. Mathematical description: Assuming that the catastrophic operation is judged when the number of iterations reaches T, the similarity between individuals in the population is calculated, and the Euclidean distance method is used. If the Euclidean distance between most individuals and the optimal individual is less than a certain threshold, it is considered that the conditions for catastrophic operation are met; ⑧ Return mechanism: In order to improve the global search performance of the algorithm and improve the traversal of the search, a return operation is added; when the crossover and mutation operations fail, that is, when the average fitness level of the newly generated population is lower than the average fitness level of the original population, the operation is negated and the population is restored to the previous state; ⑨ Iteration: Repeat steps ③ to ⑧ until the stopping condition is met, such as reaching the maximum number of iterations or the fitness value reaches a certain accuracy requirement.

Images