Lithium battery internal temperature and surface temperature estimation method based on transfer model

Abstract

The present application relates to the technical field of lithium ion batteries, and provides a lithium battery internal temperature and surface temperature estimation method based on a migration model.The method comprises the following steps: obtaining monomer lithium battery charging and discharging data, constructing a basic model, fitting SOC-OCV, identifying parameters through machine learning, obtaining a fitting relationship between basic model parameters and internal and external temperatures, constructing a migration model and correcting the internal and external temperatures of the battery, and finally obtaining the internal and external temperatures of the lithium ion battery.The lithium battery internal temperature and surface temperature estimation method based on the migration model fully considers the influence of working environment temperature and aging state on the internal parameters of the battery by introducing the migration model, and avoids the particle degradation problem of the traditional particle filtering algorithm by using the particle filtering algorithm based on particle weight selection optimization, realizes the rapid migration construction of the lithium battery electro-thermal coupling model and the rapid acquisition of the temperature state of the lithium battery in the whole life cycle, reduces the demand for a large amount of new modeling data, and improves the calculation speed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of state estimation of lithium-ion batteries for vehicles, in particular to a lithium battery internal temperature and surface temperature estimation method based on a transfer model. BACKGROUND

[0002] Lithium batteries (lithium-ion batteries) as an excellent energy storage device have become the main power source of new energy vehicles, and their safety performance has also become the focus of attention. During the operation of electric vehicles, lithium-ion batteries often work under complex conditions, and the battery temperature state changes accordingly. Therefore, accurate estimation of the battery temperature state is crucial to ensure the safe operation of the battery and the safe operation of the vehicle. At the same time, accurate estimation of the battery temperature state is also a key point for efficient control of the thermal management system.

[0003] At the current stage, the main method to obtain the temperature state of lithium battery cells is to place temperature sensors inside and on the surface of the battery. However, placing sensors in a large number of modules in real vehicles will increase costs and increase the risk of losing control of management. The current estimation method for the internal temperature state of lithium-ion batteries mainly includes: using simulation software to establish a battery finite element model to simulate and analyze the internal heat generation and heat release temperature variation mechanism of lithium batteries. This method only considers the ideal environment under experimental conditions and ignores the interference of various complex external factors in real vehicle operation, resulting in an overly idealized result. Although the data-driven research method is more realistic and feasible, it requires a large amount of real vehicle operation data to support model training. The partial differential equation based on the lithium battery electrochemical reaction mechanism can reflect the internal reaction of the battery, but it usually requires a large number of experiments to calibrate its electrochemical parameters, and the parameter dependence is strong, the model structure is complex, and the calculation amount is large. In addition, the complex working environment temperature changes and the aging state of the battery itself will further increase the difficulty of accurate estimation, so it is of great practical significance to achieve accurate and efficient internal and external temperature state estimation of the battery within a wide temperature working range during the entire service process. SUMMARY

[0004] (I) The technical problem to be solved by the present application

[0005] In view of the deficiencies of the prior art, the lithium battery internal temperature and surface temperature estimation method based on the migration model is provided, the migration model is used to greatly reduce the data amount required by the traditional battery temperature estimation method, only a benchmark migration model of the environmental temperature and the aging state needs to be established, and the model information under other environmental temperatures and aging states can be obtained through online migration. In actual application, the battery internal temperature and surface temperature estimation can be quickly completed, when the battery temperature abnormality occurs, the battery temperature state can be fed back to the battery management system in time, a large amount of time is saved for thermal management, and the safety of real vehicle operation is greatly increased. The method has the advantages of small calculation amount, high precision and strong practicability, and solves the above technical problems.

[0006] (II) Technical scheme of the present application

[0007] To achieve the above object, the present application provides the following technical scheme: a lithium battery internal temperature and surface temperature estimation method based on a migration model, comprising the following steps:

[0008] S1: Obtain the charge and discharge data of lithium ion single batteries under wide temperature conditions in the whole life, including current, voltage, surface temperature and tab temperature data. The specific steps are as follows: cycle charge and discharge aging test is carried out on different batteries of the same type, the battery is fully charged by constant current and constant voltage, then discharged to the battery cut-off voltage, the above process is cycled, the battery is discharged to 90% and 80% of the nominal battery capacity respectively, then the UDDS-FUDS working condition and HPPC working condition experiments are carried out in the temperature range of-20℃ to 60℃, and the current, voltage, surface temperature and tab temperature data of the single battery are collected, that is, the UDDS-FUDS working condition and HPPC working condition experimental data, wherein the UDDS-FUDS working condition experimental data is used for model parameter identification, migration model building, temperature estimation and verification, and the HPPC working condition experimental data is used for fitting the relationship between SOC and open circuit voltage OCV.

[0009] S2, construct an equivalent circuit model and a battery thermal model as a basic electro-thermal coupling model, wherein the equivalent circuit model is a second-order RC equivalent circuit model, the battery thermal model includes a heat generation model and a heat transfer model, the heat generation model adopts a simplified Bernardi equation as a heat generation model, and a two-state lumped parameter thermal model is adopted as a heat transfer model to simulate the heat transfer in the battery, and the second-order RC equivalent circuit model, the simplified Bernardi equation and the two-state lumped parameter thermal model are discretely expressed;

[0010] S3, curve fitting: using the experimental data under the HPPC working condition, the relationship curve between SOC and open circuit voltage OCV is obtained by six-order polynomial fitting, and the fitting expression is as follows:

[0011]

[0012] wherein, denotes the polynomial fit parameters, denotes the sum of the entire sixth order polynomial, denotes the different power terms of the SOC;

[0013] In the normal temperature full new state, based on the UDDS-FUDS operating condition experimental data obtained in S1, the recursive least square method with forgetting factor is used to identify the parameters of the second order RC equivalent circuit and thermal model established in S2, and the identified parameters are the parameters of the basic electro-thermal coupling model, and the identification process is as follows:

[0014] In the normal temperature full new state, based on the UDDS-FUDS operating condition experimental data obtained in S1, the recursive least square method with forgetting factor is used to identify the parameters of the second order RC equivalent circuit and thermal model established in S2, and the identification process is as follows:

[0015] Let the model parameter vector be θ=[αβγ] T

[0016] The observation vector is:

[0017]

[0018] The recursive loop formula is:

[0019]

[0020]

[0021] In the formula, K is the gain matrix; P is the covariance matrix; is the system estimated reference value; y(k+1) is the actual observation value of the system; is the observation value matrix; Δt is the sampling time interval 1s;

[0022] The battery internal heat generation power Q c According to Bernardi equation, the heat capacity C of the aluminum plastic film on the surface of the battery air According to the data provided by the battery manufacturer, the value is a constant,

[0023]

[0024] From the above formula, the parameters C c , R c , R air can be obtained.

[0025] S4, obtaining a fitting relationship between the basic electro-thermal coupling model parameters and the internal and external temperatures of the battery: based on the internal and external temperature values of the battery obtained in S1 and the electro-thermal coupling model parameters identified in S3, the relationship curves between the model parameters and the internal and surface temperatures of the battery are obtained by a neural network fitting method respectively;

[0026] S5, constructing a migration model and correcting the internal and surface temperatures of the single battery, based on the fitting relationship between the basic model parameters and the internal and surface temperatures of the battery, a migration framework is built, under different environmental temperatures and aging conditions, the online migration of the migration factor is completed by the weight selection particle filtering algorithm, so that the inaccurate internal and surface temperature values of the single battery are migrated, and the real internal and surface temperature values under the influence of environmental temperature and aging are obtained.

[0027] As a preferred technical scheme of the present application, the working condition charge-discharge data of the lithium ion single battery under different environmental temperatures and different aging states collected in step S1 include: voltage, current, surface temperature, tab temperature data.

[0028] As a preferred technical scheme of the present application, the discrete expression of the second-order RC equivalent circuit model established in step S2 is:

[0029]

[0030] In the above formula, U t (k) represents the terminal voltage of the circuit at the current moment; U ocv (k) represents the open circuit voltage of the circuit at the current moment; I(k) represents the current of the circuit at the current moment; U1(k), U2(k) represent the terminal voltage corresponding to the polarization resistance and the polarization capacitance at the current moment; U1(k+1), U2(k+1) represent the terminal voltage corresponding to the polarization resistance and the polarization capacitance at the next moment; (k) represents the state quantity at the current moment; (k+1) represents the state quantity at the next moment; R0 is the ohmic resistance; R1, R2 represent the polarization resistance; C1, C2 represent the polarization capacitance; τ1, τ2 represent the time constant; Δt represents the sampling time interval 1s.

[0031] In step S2, the discrete expression of the heat generation equation is:

[0032]

[0033] In the above formula, T(k) represents the battery core temperature at the current moment, and T represents the battery core temperature.

[0034] The discrete expression of the heat transfer equation is:

[0035]

[0036] In the above formula, Qc (k) represents the heat generation power of the single battery at the current time; C c represents the heat capacity of the battery; R c represents the thermal resistance between the inside of the battery and the surface of the battery; C air represents the heat capacity of the surface shell of the battery; R air represents the thermal resistance between the surface of the battery and the external environment; T c (k), T s (k), T air (k) respectively represent the temperature of the inside of the battery, the temperature of the surface of the battery and the temperature of the environment around the battery at the current time.

[0037] As a preferred technical solution of the present application, the steps of obtaining the SOC-OCV fitting curve in step S3 and identifying the basic model parameters by using the FFRLS algorithm are as follows:

[0038] S3.1, the voltage and current data of the single battery in step S1 under the HPPC working condition are extracted, and a 6-order polynomial fitting method is used to fit the relationship curve of the open circuit voltage OCV and the SOC.

[0039] S3.2, the internal parameters of the electrical model and the thermal model are obtained by using the FFRLS algorithm respectively, wherein the identified parameters of the electrical model are ohmic resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration difference polarization resistance R2 and concentration difference polarization capacitance C2; the identified parameters of the thermal model are: battery heat capacity C c , thermal resistance R c between the inside of the battery and the surface of the battery, heat capacity C air of the surface shell of the battery, and thermal resistance R air .

[0040] In the normal temperature full new state, based on the UDDS-FUDS working condition experimental data obtained in S1, the parameters of the second-order RC equivalent circuit and the thermal model established in S2 are identified by using the recursive least square method with forgetting factor, and the identification process is as follows:

[0041] Let the model parameter vector be θ=[αβγ] T ;

[0042] The observation vector is:

[0043]

[0044] The recursive loop formula is:

[0045]

[0046]

[0047] In the formula, K is the gain matrix; P is the covariance matrix; y(k+1) represents the system's estimated reference value; y(k+1) represents the system's actual observed value. This is the observation matrix; Δt is the sampling time interval (1 second).

[0048] Battery internal heat generation power Q c The heat capacity C of the aluminum-plastic film on the battery surface can be estimated using the Bernardi equation. air This value can be obtained from data provided by the battery manufacturer; it is a fixed value.

[0049]

[0050] From the above formula, we can obtain the parameter C. c R c R air .

[0051] As a preferred embodiment of the present invention, step S4, which uses a neural network fitting method to fit the basic model parameters with the internal and external temperatures of the battery, is as follows:

[0052] S4.1. Prepare training data, including the experimentally obtained battery internal and external temperature data T. c T s As input data, the electrical and thermal model parameters identified using the FFRLS algorithm will be used, including ohmic internal resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration difference polarization resistance R2, concentration difference polarization capacitance C2, and battery thermal capacity C. c The thermal resistance R between the inside of the battery and the surface of the battery c Thermal resistance R between the battery surface and the external environment air As the target output data.

[0053] S4.2 Create a neural network. Create a feedforward neural network. Select a network with a single hidden layer and select an appropriate number of hidden nodes through cross-validation. Specify the number of hidden layer nodes as 10.

[0054] S4.3. Set the neural network parameters and training function, using the Levenberg-Marquardt algorithm. This algorithm is an optimization algorithm for solving nonlinear least squares problems and is widely used in curve fitting, parameter estimation, and other fields. It combines the advantages of gradient descent and Gauss-Newton methods, and can flexibly adjust the optimization strategy under different conditions; set the normalization range for the input and output.

[0055] S4.4, training the neural network, training the neural network model, using the train function for fitting, which is commonly used in machine learning or deep learning frameworks for training models, the main function is to adjust the parameters of the model through optimization algorithms to minimize the loss function. The training process adjusts the weights and biases of the network so that the network output matches the target data as much as possible.

[0056] S4.5, evaluating model performance and visualizing results, after training, draw a comparison chart between the real target data and the neural network prediction results, and evaluate the model performance by comparison.

[0057] As a preferred technical solution of the present application, the step S5 of constructing the transfer model is as follows: constructing the battery internal temperature T c The transfer model between the basic model parameters is as follows:

[0058]

[0059] Constructing the battery surface temperature T s The transfer model between the basic model parameters is as follows:

[0060]

[0061] In the above formula, X = [x1, x2, x 3, …, x 22 ] represents 22 uncertain transfer factors; T air , β represents the influence factor of the model on the surrounding environment temperature and battery aging; represents the parameter corrected based on the transfer model at the current time corresponding to the parameter, * represents the model parameters, k represents the current state quantity, m represents the correction after the transfer model; g * (T i c / s,k ,T air ,β) represents a function of correcting each battery model parameter, T i c / s,k represents the inaccurate internal and external temperature values of the battery under the influence of the environment temperature and aging; f * (x1T i c / s,k +x2,T air ,β) represents the mapping relationship between each model parameter and the internal and external temperature of the battery, x1T i c / s,k +x2 represents the corrected internal and external temperature values of the battery; I k represents the current at time k.

[0062] As the preferred technical scheme of the present application, the step S5 is completed by the weight selection particle filter algorithm to correct the inaccurate internal and external temperature values of the single battery, and to obtain the accurate internal and external temperature values under the influence of the environment temperature and aging, with the specific steps as follows:

[0063] S5.1, with the migration of time, the migration factor matrix X=[x1, x2, x 3, …,x 22 ]obeys the Gaussian distribution, therefore, the matrix is taken as the state variable of the system, and the terminal voltage of the battery is taken as the observation of the system, and the discrete state equation of the system is established as:

[0064]

[0065]

[0066]

[0067]

[0068] In the above formula, x 1,k , x 2,k , x 3,k …,x 22,k and U t,k are the state equation and observation equation of the system respectively, x ×,k represents the xth migration factor at the kth moment, x ×,k-1 represents the xth migration factor at the k-1th moment, is the system noise, is the variance of the system measurement noise, and Rand is a number randomly generated in N particles, represents the system output value of the ith particle at the kth moment;

[0069] S5.2, particle initialization. N initial particles are generated by using the prior probability p(x k ): The weight is where x k is the particle at the kth moment, is the ith particle at the kth moment, and δ(·) is the Dirichlet function.

[0070] S5.3, next step particle state prediction. The state of the N particles at the kth moment is predicted by the state x k-1 at the k-1th moment and the predicted terminal voltage value corresponding to each is calculated, wherein is the system noise at the k-1th moment.

[0071]

[0072] S5.4, Update particle weights.

[0073]

[0074] In the above formula, Let U be the system output value of the i-th particle at time k. t,k Let be the terminal voltage at time k. This represents the particle weight at the current moment. σ represents the updated particle weights. 23 exp represents the standard deviation of the system measurement noise.

[0075] S5.5 Optimal Selection. Select the N particles with the largest weight from the N particles. S One particle.

[0076] S5.6, Particle weight normalization.

[0077]

[0078] In the above formula, For the updated N S Individual particle weights The normalized particle weights, The particle weights are updated in step S5.4, where l represents the number of particles. Indicates the internal N S Summing of data points.

[0079] S5.7 Filter estimation, calculate the selected N S The posterior probability density p(x) of each particle k |U 1,k ).

[0080]

[0081] In the above formula, U 1,k This represents the actual voltage of U1 at time k. For time k, N S The state of a particle, x k Let N be the state of N particles at time k.

[0082] S5.8 Weight Restoration and Normalization.

[0083]

[0084] In the above formula, The normalized weights for all particles.

[0085] S5.9, calculate the model end voltage.

[0086]

[0087] S5.10, internal and external temperature estimation.

[0088]

[0089] In the above formula, is the accurate internal temperature value under the influence of ambient temperature and aging state, is the accurate surface temperature value under the influence of ambient temperature and aging state, is the first particle state at time k, is the second particle state at time k, is the normalized weight of all particles.

[0090] S5.11, filter and denoise the estimated internal and surface temperatures in the migration model. A Savitzky-Golay filter is used, which smoothes the signal by fitting local data points with polynomials, effectively removing high-frequency noise and preserving the trend of the signal. The internal and external temperatures calculated by the migration model are used as the input of the filter to smooth and denoise the internal and surface temperature values obtained in step S5.10.

[0091] Compared with the prior art, the present application provides a lithium battery internal and surface temperature estimation method based on a migration model, which has the following beneficial effects:

[0092] 1. The present application uses a simplified Bernardi equation as a heat generation model and a two-state lumped parameter heat model as a heat transfer model to simulate the internal heat transfer of a lithium ion battery. The two models are coupled with a second-order RC equivalent circuit model as a basic electrical model to build a complete basic electrical-thermal coupling model.

[0093] 2. The present application uses a machine learning method to consider the influence of different aging degrees and working environment temperatures on the electrical-thermal model parameters. Based on the fitting relationship between the basic electrical-thermal coupling model parameters and the internal and surface temperatures, a migration model is built. The influence of the environment temperature and aging on the model accuracy is determined as an uncertain quantity. The existing model parameters are corrected by the slope and bias, and then the real model parameters are obtained to make linear changes to the basic model. The migration factor is determined online to realize online determination of the real parameter information of the model. Further, the inaccurate internal and surface temperature values are migrated to obtain the accurate internal and surface temperature values under the influence of temperature and aging, so that the lithium ion battery has good application prospects in the field of electric vehicles. BRIEF DESCRIPTION OF DRAWINGS

[0094] Fig. 1 The flowchart is a schematic diagram of the present application.

[0095] Fig. 2 The second-order RC equivalent circuit model is a schematic diagram of the present application.

[0096] Fig. 3 The electro-thermal coupling model is a schematic diagram of the present application. DETAILED DESCRIPTION

[0097] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0098] Please refer to Figs. 1-3 The lithium battery internal temperature and surface temperature estimation method based on the migration model includes the following steps:

[0099] S1: Obtain the charge and discharge data of the lithium ion single battery under the wide temperature condition of the whole life, including the current, voltage, surface temperature and tab temperature data of the single battery.

[0100] The specific steps are as follows: cycle charge and discharge aging test is performed on different batteries of the same model, the battery is fully charged by constant current and constant voltage, and then discharged to the battery cut-off voltage, the above process is cycled, the battery is discharged to 90% and 80% of the nominal battery capacity respectively, and then the working condition experiment is carried out in the temperature range of-20℃ to 60℃, and the current, voltage, surface temperature and tab temperature data of the single battery are collected; the UDDS-FUDS working condition experiment data are as follows:

[0101] Table 1 Experimental record data

[0102]

[0103] S2, construct the equivalent circuit model and battery thermal model of the lithium ion single battery, wherein the equivalent circuit model is a second-order RC equivalent circuit model, and the battery thermal model includes a heat generation model and a heat transfer model.

[0104] Specifically, the simplified Bernardi equation is used as the heat generation model, the two-state lumped parameter thermal model is used as the heat transfer model to simulate the internal heat transfer of the battery, and the second-order RC equivalent circuit model, the heat generation model and the heat transfer model are discretely expressed;

[0105] The discretization expression of the second-order RC equivalent circuit model established in step S2 is:

[0106]

[0107] In the above formula, T(k) represents the current time battery core temperature, and T represents the battery core temperature. t (k) represents the current time circuit open circuit voltage; I(k) represents the current time circuit current; U1(k), U2(k) represent the current time polarization resistance, capacitor corresponding to the terminal voltage; U1(k+1), U2(k+1) represent the next time polarization resistance, capacitor corresponding to the terminal voltage; (k) represents the state quantity at the current time; (k+1) represents the state quantity at the next time; R0 is the ohmic resistance; R1, R2 represent the polarization resistance; C1, C2 represent the polarization capacitance; τ1, τ2 represent the time constant; Δt represents the sampling time interval 1s. ocv (k) represents the current time circuit open circuit voltage; I(k) represents the current time circuit current; U1(k), U2(k) represent the current time polarization resistance, capacitor corresponding to the terminal voltage; U1(k+1), U2(k+1) represent the next time polarization resistance, capacitor corresponding to the terminal voltage; (k) represents the state quantity at the current time; (k+1) represents the state quantity at the next time; R0 is the ohmic resistance; R1, R2 represent the polarization resistance; C1, C2 represent the polarization capacitance; τ1, τ2 represent the time constant; Δt represents the sampling time interval 1s.

[0108] In step S2, the heat generation equation is discretized as follows:

[0109]

[0110] In the above formula, T(k) represents the current time battery core temperature, and T represents the battery core temperature.

[0111] The heat transfer equation is discretized as follows:

[0112]

[0113] In the above formula, Q c (k) represents the current time single battery heat generation; C c represents the heat capacity of the battery; R c represents the thermal resistance between the battery interior and the battery surface; C air represents the heat capacity of the battery surface shell; R air represents the thermal resistance between the battery surface and the external environment; T c (k), T s (k), T air (k) respectively represent the current time battery interior temperature, battery surface temperature and the ambient temperature around the battery.

[0114] S3, the steps of obtaining the SOC-OCV fitting curve and identifying the basic model parameters by using the FFRLS algorithm are as follows:

[0115] S3.1, the voltage and current data of the single battery in step S1 under HPPC working condition are extracted, and the 6-order polynomial fitting method is used to fit the relationship curve of open circuit voltage OCV and SOC.

[0116] Curve fitting: the relationship curve between SOC and OCV is obtained by six-order polynomial fitting using the experimental data under HPPC condition, and the fitting expression is as follows:

[0117]

[0118] Wherein, represents the polynomial fitting parameter, represents the sum of the entire six-order polynomial, represents the different power terms of SOC;

[0119] S3.2, the internal parameters of the electric model and the thermal model are obtained by using the FFRLS algorithm respectively, wherein the electric model identification parameters are ohmic resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration difference polarization resistance R2 and concentration difference polarization capacitance C2; the thermal model identification parameters are: battery heat capacity C c , thermal resistance R c between the inside of the battery and the surface of the battery, air thermal capacity C air of the surface shell of the battery, and thermal resistance R T between the surface of the battery and the external environment.

[0120] In the normal temperature full new state, based on the UDDS-FUDS operating condition experimental data obtained in S1, the parameters of the second-order RC equivalent circuit and the thermal model established in S2 are identified by using the recursive least square method with forgetting factor, and the identification process is as follows:

[0121] Let the model parameter vector be θ=[αβγ] T

[0122] The observation vector is:

[0123]

[0124] The recursive loop formula is:

[0125]

[0126]

[0127] In the formula, K is the gain matrix; P is the covariance matrix; is the system estimated reference value; y(k+1) is the actual observation value of the system; is the observation value matrix; Δt is the sampling time interval 1s;

[0128] The internal heat generation power Q c of the battery can be estimated according to Bernardi equation, and the heat capacity C air of the surface aluminum plastic film of the battery can be obtained according to the data provided by the battery manufacturer, which is a constant value,

[0129]

[0130] The parameter C is obtained from the above formula c , R c , R air .

[0131] S4, obtaining the fitting relationship between the basic electro-thermal coupling model parameters and the internal and external temperatures of the battery: based on the internal and external temperature values of the battery obtained in S1 and the FFRLS identified electro-thermal coupling model parameters in S3, the relationship curves between the model parameters and the internal and surface temperatures of the battery are obtained by neural network fitting method;

[0132] Step S4 specifically includes the following steps:

[0133] S4.1, preparing training data, respectively taking the experimentally obtained internal and external temperature data T c , T s as input data, and taking the FFRLS algorithm identified electrical model, thermal model parameters ohmic resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration difference polarization resistance R2, concentration difference polarization capacitance C2, battery heat capacity C c , the thermal resistance R c between the battery interior and the battery surface, and the thermal resistance R air between the battery surface and the external environment as target output data.

[0134] S4.2, creating a neural network, creating a feedforward neural network, selecting a network with a single hidden layer, and selecting a suitable number of hidden nodes through cross-validation, and specifying the number of hidden layer nodes as 10.

[0135] S4.3, setting the neural network parameters, setting the training function, using the Levenberg-Marquardt algorithm, which is an optimization algorithm for solving nonlinear least squares problems, widely used in curve fitting, parameter estimation and other fields. It combines the advantages of gradient descent method and Gauss-Newton method, and can flexibly adjust the optimization strategy under different conditions; set the normalization range of input and output.

[0136] S4.4, training the neural network, training the neural network model, using the train function for fitting, which is commonly used in machine learning or deep learning frameworks for training models, and the main function is to adjust the parameters of the model through optimization algorithm to minimize the loss function. The training process adjusts the weights and biases of the network so that the network output matches the target data as much as possible.

[0137] S4.5, evaluate the model performance and visualize the results, after training, draw a comparison chart between the real target data and the neural network prediction results, and evaluate the model performance by comparison.

[0138] S5, build a migration model and correct the internal and external temperatures of the single battery: based on the fitting relationship between the basic electro-thermal coupling model parameters obtained in S4 and the internal and external temperatures of the battery, a migration framework is built, and under different environmental temperatures and aging conditions, the online migration of the migration factor is completed through the weight selection particle filtering algorithm, so as to migrate the inaccurate internal and external temperature values of the single battery, and obtain accurate internal and external temperature values under different environmental temperatures and aging conditions. The migration model construction formula is as follows:

[0139] The internal temperature T of the battery is constructed as follows: c The migration model between the basic model parameters is as follows:

[0140]

[0141] The surface temperature T of the battery is constructed as follows: s The migration model between the basic model parameters is as follows:

[0142]

[0143] In the above formula, X = [x1, x2, x 3, …, x 22 ] represents 22 uncertain migration factors; T air , β represents the influence factors of the model on the surrounding environmental temperature and the battery aging; represents the parameters corrected based on the migration model at the current time, * represents the model parameters, k represents the current state quantity, and m represents the correction after the migration model; g * (T i c / s,k , T air , β) represents a function for correcting each battery model parameter, T i c / s,k represents the inaccurate internal and external temperature values of the battery under the influence of environmental temperature and aging; f * (x1T i c / s,k +x2, T air , β) represents the mapping relationship between each model parameter and the internal and external temperatures of the battery, x1T i c / s,k +x2 represents the corrected internal and external temperature values of the battery; I k represents the current at time k.

[0144] Step S5 specifically includes the following steps:

[0145] S5.1, with the passage of time, the transition factor matrix X = [x1, x2, x 3, …, x 22 ] obeys Gaussian distribution, so the matrix is taken as the state variable of the system, and the terminal voltage of the battery is taken as the observation of the system, and the discrete state equation of the system is established as:

[0146]

[0147]

[0148]

[0149]

[0150] In the above formula, x 1,k , x 2,k , x 3,k …, x 22,k and U t,k are the state equation and observation equation of the system respectively, x ×,k represents the xth transition factor at k time, x ×,k-1 represents the xth transition factor at k-1 time, is the system noise, is the variance of the system measurement noise, and Rand is a random number generated in N particles, represents the output value of the i th particle system at k time;

[0151] S5.2, particle initialization. N initial particles are generated by using the prior probability p(x k ): The weight of which is where x k is the particle at the kth time, is the i th particle at the kth time, and δ(·) is the Dirichlet function.

[0152] S5.3, next step particle state prediction. The state of N particles at k time is predicted by the state x k-1 at k-1 time And the predicted terminal voltage value corresponding to each is calculated, wherein is the system noise at k-1 time.

[0153]

[0154] S5.4, update particle weight.

[0155]

[0156] In the above equation, is the system output value of the i-th particle at time k, U t,k is the terminal voltage at time k, denotes the particle weight at the current time, denotes the updated particle weight, σ 23 denotes the standard deviation of the system measurement noise, exp denotes the natural logarithm.

[0157] S5.5, Selection. Select the N S particles with the largest weights from the N particles.

[0158] S5.6, Normalization of Particle Weights.

[0159]

[0160] In the above equation, is the weight of the i-th particle after updating, S is the weight of the i-th particle after normalization, is the weight of the i-th particle after normalization, is the weight of the i-th particle after updating in step S5.4, and I denotes the number of particles, denotes the summation of the internal N S particles.

[0161] S5.7, Filtering Estimation, calculate the posterior probability density p(x S |U k ) of the selected N 1,k particles.

[0162]

[0163] In the above equation, U 1,k denotes the actual voltage of U1 at time k, is the state of the N S particles at time k, x k is the state of the N particles at time k.

[0164] S5.8, Weight Recovery and Normalization.

[0165]

[0166] In the above equation, is the normalized weight of all particles.

[0167] S5.9, Calculation of Model Terminal Voltage.

[0168]

[0169] S5.10, Internal and External Temperature Estimation.

[0170]

[0171] in the above formula, is the accurate internal temperature value under the influence of ambient temperature and aging state, is the accurate surface temperature value under the influence of ambient temperature and aging state, is the first particle state at time k, is the second particle state at time k, is the normalized weight of all particles.

[0172] S5.11, filter and denoise the estimated values of internal temperature and surface temperature in the migration model. A Savitzky-Golay filter is adopted, which smoothes the signal by fitting local data points with a polynomial, can effectively remove high-frequency noise and retain the trend of the signal. The internal and external temperatures calculated by the migration model are taken as the input of the filter, and the internal and surface temperature values obtained in step S5.10 are smoothed and denoised.

[0173] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for estimating internal and surface temperature of a lithium battery based on a transfer model, characterized in that: The method comprises the following steps: S1: obtaining the charging and discharging data of a lithium ion single battery under a wide temperature condition in a whole life cycle, including the current, voltage, surface temperature and tab temperature data of the single battery; S2, constructing an equivalent circuit model and a battery thermal model of the lithium ion single battery, wherein the equivalent circuit model is a second-order RC equivalent circuit model, and the battery thermal model comprises a heat generation model and a heat transfer model; S3, SOC-OCV curve fitting: using the experimental data under the HPPC condition, the relationship curve between SOC and OCV is obtained by six-order polynomial fitting, and the fitting expression is as follows: wherein, represents a polynomial fit parameter, represents the sum of the entire sixth order polynomial, represents different power terms of the SOC; Under the normal temperature full new state, based on the UDDS-FUDS operating condition experimental data obtained in S1, the parameters of the second-order RC equivalent circuit and the thermal model established in S2 are identified by using the recursive least square method with a forgetting factor, and the identification process is as follows: Let the model parameter vector be: θ = [α β γ] T , The observation vector is: The recursive circulation formula is: In the formula, K is a gain matrix; P is a covariance matrix; is a system estimated reference value; y(k+1) is a system actual observation value; is an observation value matrix; Δt is a sampling time interval 1s; The heat generation power Q inside the battery c The heat capacity C of the aluminum plastic film on the surface of the battery can be estimated according to the Bernardi equation air The value can be obtained according to the data provided by the battery manufacturer, which is a constant value, From the above equation, the parameter C c , R c , R air ; S4, obtaining the fitting relationship between the basic electro-thermal coupling model parameters and the internal and external temperatures of the battery: based on the internal and external temperature values of the battery obtained in S1 and the electro-thermal coupling model parameters identified in S3, the relationship curves between the model parameters and the internal and surface temperatures of the battery are obtained by a neural network fitting method; S5, constructing a migration model and correcting the internal and external temperatures of the single battery: based on the fitting relationship between the basic electro-thermal coupling model parameters and the internal and external temperatures of the battery obtained in S4, a migration framework is built, and under different environmental temperatures and aging conditions, the online migration of the migration factor is completed by using the weight selection particle filtering algorithm, so that the inaccurate internal and external temperature values of the single battery are migrated, and the accurate internal and external temperature values under different environmental temperatures and aging conditions are obtained. 2.The lithium battery internal temperature and surface temperature estimation method based on a migration model of claim 1, wherein, The single battery operating charging and discharging data collected in S1 under different environmental temperatures and different aging conditions comprises voltage, current, surface temperature and tab temperature data. 3.The lithium battery internal temperature and surface temperature estimation method based on a migration model of claim 1, wherein, The second-order equivalent circuit model established in S2 is: The discrete model expression of the second-order RC equivalent circuit model is: In the above formula, U t (k) represents the open-circuit voltage of the circuit at the current time; I(k) represents the current of the circuit at the current time; U1(k), U2(k) represent the terminal voltage corresponding to the polarization resistance and the polarization capacitance at the current time; U1(k+1), U2(k+1) represent the terminal voltage corresponding to the polarization resistance and the polarization capacitance at the next time; (k) represents the state quantity at the current time; (k+1) represents the state quantity at the next time; R0 is the ohmic resistance; R1, R2 represent the polarization resistance; C1, C2 represent the polarization capacitance; τ1, τ2 represent the time constant; Δt represents the sampling time interval 1s; ocv (k) represents the open-circuit voltage of the circuit at the current time; I(k) represents the current of the circuit at the current time; U1(k), U2(k) represent the terminal voltage corresponding to the polarization resistance and the polarization capacitance at the current time; U1(k+1), U2(k+1) represent the terminal voltage corresponding to the polarization resistance and the polarization capacitance at the next time; (k) represents the state quantity at the current time; (k+1) represents the state quantity at the next time; R0 is the ohmic resistance; R1, R2 represent the polarization resistance; C1, C2 represent the polarization capacitance; τ1, τ2 represent the time constant; Δt represents the sampling time interval 1s; In step S2, the discrete expression of the heat generation equation is: In the formula, T(k) represents the current battery core temperature, and T represents the battery core temperature; The discrete expression of the heat transfer equation is: In the above formula, Q c (k) represents the heat generation power of the single battery at the current time; C c represents the heat capacity of the battery; R c represents the thermal resistance between the inside of the battery and the surface of the battery; C air represents the heat capacity of the surface shell of the battery; R air represents the thermal resistance between the surface of the battery and the external environment; T c (k), T s (k), T air (k) respectively represent the temperature of the inside of the battery, the temperature of the surface of the battery and the temperature of the environment around the battery at the current time. 4.The lithium battery internal temperature and surface temperature estimation method based on a migration model of claim 1, wherein, The steps of obtaining the SOC-OCV fitting curve in S3 and identifying the basic model parameters by using the FFRLS algorithm are as follows: S3.1, extracting the voltage and current data of the single battery in S1 under the HPPC condition, and using a 6-order polynomial fitting method to fit the relationship curve between the open circuit voltage OCV and SOC; S3.2, respectively, using FFRLS algorithm to obtain the internal parameters of the electrical model and thermal model, wherein the electrical model identification parameters are ohmic resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration difference polarization resistance R2, and concentration difference polarization capacitance C2; the thermal model identification parameters are: battery heat capacity C c , thermal resistance R between the battery interior and the battery surface c , heat capacity C of the battery surface shell air , thermal resistance R between the battery surface and the external environment air . 5.The lithium battery internal temperature and surface temperature estimation method based on a migration model of claim 1, wherein, The steps of fitting the basic model parameters and the internal and external temperatures of the battery by using the neural network fitting method in S4 are as follows: S4.1, prepare training data, respectively, the battery internal and external temperature data T c s As input data, the electric model, thermal model parameters ohmic resistance R0, electrochemical polarization resistance R1, electrochemical polarization capacitance C1, concentration difference polarization resistance R2, concentration difference polarization capacitance C2, battery heat capacity C c The thermal resistance R c The thermal resistance R air As the target output data;​ S4.2, create a neural network, create a feedforward neural network, select a network with a single hidden layer, and select the appropriate number of hidden nodes through cross-validation, and specify the number of hidden layer nodes as 10; S4.3, set the neural network parameters, set the training function, use the Levenberg-Marquardt algorithm, which is an optimization algorithm for solving nonlinear least squares problems, widely used in curve fitting, parameter estimation field, it combines the advantages of gradient descent method and Gauss-Newton method, can adjust the optimization strategy flexibly under different circumstances; Set the input and output normalization range; S4.4, train the neural network, train the neural network model, use the train function for fitting, which is commonly used in machine learning or deep learning framework for training model, the main function is to adjust the parameters of the model through optimization algorithm, in order to minimize the loss function, the training process will adjust the weight and bias of the network, so that the network output and target data match as much as possible; S4.5, evaluate the model performance and visualize the results, after training, draw the comparison chart between the real target data and the neural network prediction results, and evaluate the model performance by comparison. 6.The lithium battery internal and surface temperature estimation method based on a migration model of claim 1, wherein, The S5 constructs the battery internal temperature T c The migration model between the base model parameters is as follows: Constructing the battery surface temperature T s The transfer model between the base model parameters is as follows: In the above formula, X = [x1, x2, x 3, …,x 22 ] represents 22 uncertain migration factors; T air , β represents the factors affecting the model by the surrounding temperature and battery aging; represents the parameters of the corresponding parameters at the current time based on the migration model correction, * represents the model parameters, k represents the current state quantity, and m represents the correction after passing through the migration model; g * (T i c / s,k ,T air ,β) represents a function for correcting each battery model parameter, T i c / s,k represents the inaccurate internal and external temperature values of the battery under the influence of the environment temperature and aging; f * (x1T i c / s,k +x2,T air ,β) represents the mapping relationship between each model parameter and the internal and external temperature of the battery, x1T i c / s,k +x2 represents the corrected internal and external temperature values of the battery; I k represents the current at time k. 7.The lithium battery internal and surface temperature estimation method based on a migration model of claim 1, wherein, S5.4, update particle weight: S5.1, With the passage of time, the transition factor matrix X = [x1, x2, x 3, …,x 22 ] obeys Gaussian distribution, so the matrix is taken as the state variable of the system, and the terminal voltage of the battery is taken as the observation of the system, and the discrete state equation of the system is established as: In the above formula, x 1,k ,x 2,k ,x 3,k ……,x 22,k and U t,k are the state equation and observation equation of the system respectively, x ×,k represents the xth transition factor at time k, x ×,k-1 represents the xth transition factor at time k-1, is the system noise, is the variance of the system measurement noise, and Rand is a random number generated in N particles, represents the output value of the ith particle system at time k; S5.2, particle initialization: N initial particles are generated using the prior probability p(x k ): with weights where x k is the kth particle, is the ith particle at the kth time, and δ(·) is the Dirac function. S5.3, Next step particle state prediction: predict the state of N particles at time k by the state x k-1 at time k-1 and calculate the predicted end voltage value of each corresponding to the state of the particle, where is the system noise at time k-1; S5.6, particle weight normalization: In the above equation, U is the system output value of the i-th particle at the k-th time, t,k V is the terminal voltage at the k-th time, Wk represents the particle weight at the current time, Wk represents the updated particle weight, σ 23 σ represents the standard deviation of the system measurement noise, exp represents the natural logarithm; S5.5, Selection: Select the N particles with the largest weights among the N particles S ​ S5.8, weight recovery and homogenization: In the above formula, Nnewis the updated N S particle weights, Nnewis the normalized particle weights, Nnewis the updated particle weights in step S5.4, and l represents the number of particles, Nnewrepresents the summation of the internal N S data; S5.7 Filter estimation, calculate the selected N S The posterior probability density p(x) of each particle k |U 1,k ): In the above formula, U 1,k represents the actual voltage of U1 at time k, is the state of N S particles at time k, x k is the state of N particles at time k; S5.9, calculate the model terminal voltage: In the above formula, is the normalized weight of the entire particle; S5.10, internal and external temperature estimation: S5.11, filter and denoise the estimated value of internal and surface temperature in the migration model: use Savitzky-Golay filter, which can smooth the signal by polynomial fitting of local data points, can effectively remove high frequency noise and retain the trend of signal; The internal and external temperature calculated by the migration model is used as the input of the filter, and the internal and surface temperature values obtained by S5.10 are smoothed and denoised. In the above formula, is the precise internal temperature value under the influence of the ambient temperature and the aging state, is the precise surface temperature value under the influence of the ambient temperature and the aging state, is the first particle state at time k, is the second particle state at time k, is the normalized weight of all particles; ​

Images