A Fast Prediction Method for Battery SOP Based on Electrochemical Model

Abstract

The present invention relates to the technical field of battery SOP prediction, and discloses a fast battery SOP prediction method based on an electrochemical model, including the following steps: simplifying the P2D model into an e-SPM model, and coupling it with a thermal model to obtain an electrothermal coupling model; respectively identifying the thermal parameters and electrochemical parameters of the electrothermal coupling model by a variable-scale multi-objective grey wolf optimization algorithm; using an unscented Kalman filter as the state observer of the electrochemical model to observe the internal and external states of the battery. By means of a variable-scale multi-objective grey wolf optimization algorithm, the method quickly identifies 21 electrochemical model parameters, and by changing the input identification scale strategy during the optimization process, it avoids the non-convergence problem that occurs during the optimization process. By means of a multi-objective strategy, the proportion of the fitness values of various working conditions is selected through weights, avoiding the overfitting problem in parameter identification.

Description

Technical Field

[0001] The present invention relates to the technical field of battery SOP prediction, and specifically provides a fast prediction method for battery SOP based on an electrochemical model. Background Art

[0002] In order to cope with the growing environmental pollution and energy crisis, the development of clean energy has become a global urgent task and important trend. Due to the advantages of large specific energy, fast charging speed, and long life of lithium-ion batteries, they have been widely used, especially in the fields of electric vehicles and energy storage power stations. However, in actual applications, some safety problems may occur (for example, thermal runaway, internal short circuit, overcharging, and over-discharging). Therefore, to ensure the operation of the battery within the safe operating range SOA, a battery management system BMS is required to monitor the operating state of the battery, including the state of charge SOC and the power state SOP of the battery. SOC is defined as the ratio of the remaining capacity of the current battery to the maximum capacity. Accurate SOC can prevent the battery from overcharging or over-discharging. SOP is defined in most studies as the peak power capacity of the battery within a specified prediction time without exceeding the SOA boundary. However, since SOP cannot be measured and its magnitude is related to the state of the battery, accurate SOP prediction remains challenging.

[0003] Currently, the equivalent circuit model is mainly used for online prediction of lithium-ion battery SOP. However, the equivalent circuit model is difficult to reflect the internal physical state of the battery. It only limits the maximum current based on external signals, so it is difficult to ensure the safety of the battery and the accuracy of power prediction. When using the electrochemical model for online prediction of SOP, most current methods only use the surface lithium-ion concentration as the safe region, without considering other internal states and the influence of temperature and temperature on internal parameters. Therefore, obtaining the maximum power is also unsafe, and the time issue is also a major problem. Due to the slow convergence speed of the dichotomy method, it is difficult to meet online applications. Therefore, a fast prediction method for battery SOP based on an electrochemical model is proposed. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a fast prediction method for battery SOP based on an electrochemical model, which has the advantages of faster SOP prediction time and shortening the calculation time of the dichotomy method without loss of accuracy, and solves the problems mentioned in the above background art.

[0005] To achieve the above objectives of faster SOP prediction time and shortening the calculation time of the dichotomy method without loss of accuracy, the present invention provides the following technical solutions: A fast prediction method for battery SOP based on an electrochemical model, comprising the following steps:

[0006] S1: Simplify the P2D model to an e - SPM model and couple it with the thermal model to obtain an electro - thermal coupling model;

[0007] S2: Identify the thermal parameters and electrochemical parameters of the electro - thermal coupling model respectively through a variable - scale multi - objective grey wolf optimization algorithm;

[0008] S3: Use the unscented Kalman filter as the state observer of the electrochemical model to observe the internal and external states of the battery;

[0009] S4: Select the surface lithium - ion concentration, electrolyte concentration, lithium - plating over - potential, voltage, SOC, temperature as the safety constraints of the SOP. Select the bisection method for SOP prediction of the battery at different time scales, and select Gaussian process regression to accelerate the convergence rate of the bisection method for battery SOP prediction.

[0010] Preferably, the specific steps of step S2 are as follows:

[0011] S2.1: Use the 1C discharge and DST dynamic operating conditions as the objective fitness functions of the variable - scale multi - objective grey wolf optimization algorithm to obtain the fitness value fv of the mean square error between the simulated voltage V sim and the actual voltage V exp , and the expression is:

[0012]

[0013] where N is the number of voltage acquisitions, V exp is the actual voltage, and V sim is the simulated voltage;

[0014] S2.2: Calculate the absolute value fc of the difference between the positive - electrode capacity and the negative - electrode capacity, and the expression is:

[0015]

[0016] where A is the electrode area, L + is the positive - electrode thickness, L - is the negative - electrode thickness, is the positive - electrode solid - phase volume fraction, is the positive - electrode solid - phase volume fraction, is the maximum lithium - ion concentration of the positive electrode, is the maximum lithium - ion concentration of the negative electrode, F is the Faraday constant, SOL is the SOC of the single - cell electrode, SOC is the remaining charge of the single - cell battery, is the value corresponding to the positive electrode of the single - cell battery at 0, is the value corresponding to the negative electrode of the single - cell battery SOC at 0, is the value corresponding to the positive electrode of the single - cell battery at 100, The value corresponding to the negative electrode of the single battery at 100;

[0017] S2.3: The final fitness function fm of the electro-thermal coupling model is obtained by weighted calculation, and the expression is:

[0018] fm = w1fv1 + w2fv2 + w3fc

[0019] Among them, W1, W2 and W3 are the weights of different fitness functions, fv1 is the mean square error of the 1C discharge condition, and fv2 is the mean square error of the DST dynamic stress condition;

[0020] S2.4: Use the 1C discharge condition to identify the temperature parameter, and obtain the local root mean square error ft of the temperature. The expression is:

[0021]

[0022] Among them, ft is the local root mean square error, T exp refers to the experimentally measured temperature, T sim refers to the simulated temperature.

[0023] Preferably, the states inside and outside the battery include voltage, SOC, surface lithium ion concentration, electrolyte concentration, and lithium deposition overpotential.

[0024] Preferably, the specific steps of step S4 are:

[0025] S4.1: Select the surface lithium ion concentration, electrolyte concentration, lithium deposition overpotential, voltage, SOC, and temperature as the safety constraints of the SOP. The expression is:

[0026]

[0027] c e (x,t) ≥ 0

[0028]

[0029] 0 ≤ SOC(t) ≤ 1

[0030] V min ≤ V(t) ≤ V max

[0031] T min ≤ T(t) ≤ T max

[0032] Among them, and are the maximum lithium intercalation ranges of the positive and negative electrodes of the battery, is the maximum concentration of the positive and negative electrodes of the battery, is the surface lithium ion concentration of the positive and negative electrodes of the battery, η sr(x L- , t) is the lithium deposition overpotential, SOC(t) is the SOC of the current battery, V min is the minimum voltage of the battery, V max is the maximum voltage of the battery, V(t) is the current voltage of the battery, T min is the minimum temperature of the battery, T min is the maximum temperature of the battery, T min is the current temperature of the battery, c e (x, t) is the electrolyte concentration;

[0033] S4.2: Select the bisection method for SOP prediction of the battery at different time scales;

[0034] S4.3: Select Gaussian process regression for accelerated prediction of SOP, specifically:

[0035] Select the DST dynamic stress condition as the training set, take SOC, voltage, and temperature as inputs, and the SOP predicted by the bisection method and the 95% confidence interval of the SOP as outputs for offline training;

[0036] When predicting the SOP of other working conditions, obtain its 95% confidence interval through the offline model, and use the interval as the current maximum and minimum current boundaries, which speeds up the convergence rate of the bisection method.

[0037] Preferably, the specific steps of S4.2 are as follows:

[0038] S4.2.1: Set the maximum boundary current I 0，max to 20C, and set the minimum boundary current I 0，min to 0C;

[0039] S4.2.2: Let the initial maximum power current I 0，SOP be equal to the initial maximum boundary current I 0，max , and calculate the constraint state at different time intervals Δt through the electro-thermal coupling model with a constant current;

[0040] S4.2.3: Judge whether the six states of surface lithium ion concentration, electrolyte concentration, lithium deposition overpotential, voltage, SOC, and temperature all satisfy the safety constraints. If so, the initial maximum power current I 0,SOP is the maximum constant current I k,SOP in this state and this time period, enter step S4.2.8. Otherwise, update through the bisection method and enter step S4.2.4;

[0041] S4.2.4: Let the maximum constant current I k,sop be equal to half of the sum of the maximum boundary current I k，max and the minimum boundary current I k，min and use the maximum constant current Ik,sop Calculate the constraint status at different time intervals Δt through the electro-thermal coupling model;

[0042] S4.2.5: Determine whether the six states of surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature all satisfy the safety constraints,

[0043] S4.2.6: If not, set the maximum boundary current I k，max to the maximum constant current I k,sop , and calculate the constraint status at different time intervals Δt through the electro-thermal coupling model with the maximum constant current I k,sop , and return to step S4.2.5. If yes, enter step S4.2.7;

[0044] S4.2.7: Determine whether the maximum tolerance tol is satisfied. The judgment expression is:

[0045] |I k,max -I k,SOP | ≤ tol

[0046] |I k,min -I k,SOP | ≤ tol

[0047] where I k,max is the maximum boundary current at the k-th iteration, I k,min is the minimum boundary current at the k-th iteration, I k,SOP is the maximum constant current, and tol is the maximum tolerance;

[0048] If not, set the minimum boundary current I k，min equal to the maximum constant current I k,SOP , and return to step S4.2.5,

[0049] If yes, enter step S4.2.8:

[0050] S4.2.8: Stop the iteration and calculate the final maximum constant current I sop(t) , and the expression is:

[0051] I SOP (t) = I k,SOP

[0052] S4.2.9: Calculate the minimum terminal voltage at the time interval Δt through the maximum constant current I sop(t) and calculate the maximum charging power and the maximum discharging power , and the expressions are:

[0053] ​

[0054] Among them, is the maximum charging power at this moment, is the maximum charging current at this moment, is the terminal voltage at this moment, is the discharge charging power at this moment, is the maximum charging current at this moment, is the minimum terminal voltage after calculating Δt.

[0055] Compared with the prior art, the present invention provides a method for quickly predicting the battery SOP based on an electrochemical model, which has the following beneficial effects:

[0056] 1. The method for quickly predicting the battery SOP based on an electrochemical model uses a variable-scale multi-objective grey wolf optimization algorithm to quickly identify 21 electrochemical model parameters. By changing the input identification scale strategy during the optimization process, the problem of non-convergence during the optimization process is avoided; through a multi-objective strategy, the proportion of fitness values of various working conditions is selected by weights, avoiding the overfitting problem in parameter identification.

[0057] 2. The method for quickly predicting the battery SOP based on an electrochemical model uses Gaussian process regression to offline learn the relationship between SOP and its 95% confidence interval, voltage, SOC, and temperature. When predicting SOP, the interval range of the dichotomy method is shortened, making the dichotomy method converge more quickly, accelerating the prediction of SOP, and enabling the online use of SOP prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a schematic flow chart of a method for quickly predicting the battery SOP based on an electrochemical model proposed by the present invention;

[0059] Figure 2 is a schematic diagram of the 1s charging current of a method for quickly predicting the battery SOP based on an electrochemical model proposed by the present invention;

[0060] Figure 3 is a schematic diagram of the 1s charging power of a method for quickly predicting the battery SOP based on an electrochemical model proposed by the present invention;

[0061] Figure 4 is a schematic diagram of the 30s charging current of a method for quickly predicting the battery SOP based on an electrochemical model proposed by the present invention;

[0062] Figure 5 is a schematic diagram of the 30s charging power of a method for quickly predicting the battery SOP based on an electrochemical model proposed by the present invention;

[0063] Figure 6Schematic diagram of the 60s charging current of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0064] Figure 7 Schematic diagram of the 60s charging power of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0065] Figure 8 Schematic diagram of the 1s discharge power of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0066] Figure 9 Schematic diagram of the 1s discharge current of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0067] Figure 10 Schematic diagram of the 30s discharge current of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0068] Figure 11 Schematic diagram of the 30s discharge power of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0069] Figure 12 Schematic diagram of the 60s discharge current of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0070] Figure 13 Schematic diagram of the 60s discharge power of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0071] Figure 14 Schematic diagram of the parameter identification process of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention;

[0072] Figure 15 Schematic diagram of the dichotomy process of a battery SOP rapid prediction method based on an electrochemical model proposed by the present invention. Detailed implementation manners

[0073] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0074] Please refer to Figures 1-15 , a battery SOP rapid prediction method based on an electrochemical model, includes the following steps:

[0075] S1: Simplify the P2D model to an e - SPM model and couple it with the thermal model to obtain an electro - thermal coupling model, with the expression:

[0076]

[0077] Among them, in the two electrodes, the solid - phase concentration C s (r, x, t) remains constant in the x - dimension, C e，0 represents the initial lithium - ion concentration in the electrolyte, and represent the electrolyte volume fractions of the negative electrode, separator, and positive electrode respectively, L n , L s and L p represent the lengths of the negative electrode, separator, and positive electrode respectively, S represents the cross - sectional area of the battery cell, and represent the initial lithium - ion concentrations of the negative - electrode and positive - electrode respectively, and represent the volume fractions of the negative - electrode and positive - electrode respectively;

[0078] When the battery temperature T(x, t) varies little in the x - dimension and only changes with time t, after obtaining the output vector through the state estimator, the battery terminal voltage can be described as:

[0079]

[0080] Among them, is related to the equilibrium potentials of the anode and cathode, representing the estimated value of the equilibrium potential, The first part of is the ohmic potential drop caused by the electrolyte conductivity, the second part is the estimated value of the electrolyte concentration over - potential, ΔU film (t) represents the voltage across the thin - film resistance, represents the over - potential.

[0081] U + , U - represent the equilibrium potentials of the positive and negative electrodes respectively; represent the surface lithium - ion concentrations of the negative - electrode and positive - electrode respectively;

[0082] represent the effective reaction rate constants of the negative electrode, separator, and positive electrode respectively; I represents the charge - discharge current, R represents the universal gas constant; F represents the Faraday constant; represents the lithium - ion transference number;

[0083] represent the thin - film resistances of the positive and negative electrodes respectively; represent the interface areas of the spherical particles of the positive and negative electrodes respectively;

[0084] α represents the charge transfer coefficient of the electrode; represent the exchange current densities of the positive and negative electrodes respectively.

[0085] S2: Identify the thermal parameters and electrochemical parameters of the electro-thermal coupling model respectively through the variable-scale multi-objective grey wolf optimization algorithm;

[0086] S3: Use the unscented Kalman filter as the state observer of the electrochemical model to observe the internal and external states of the battery. The internal and external states of the battery include voltage, SOC, surface lithium-ion concentration, electrolyte concentration, lithium plating overpotential. For temperature, it is only used as a correction of the coupling parameter, and the temperature measured in the previous second is used to correct the coupling parameter in the next second;

[0087] S4: Select the surface lithium-ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature as the safety constraints of SOP. Select the bisection method to predict the SOP of the battery at different time scales, and select the Gaussian process regression to accelerate the convergence speed of the bisection method for predicting the battery SOP;

[0088] The power of the lithium battery is not only limited by the external terminal voltage, current and temperature, but also limited by the internal electrochemical reactions. Therefore, accurate power prediction cannot be carried out only relying on the externally measurable states. Therefore, several internal electrochemical states of the lithium battery are selected as additional safety constraints. Compared with the power prediction of the ECM model, the current provided by the manufacturer is not selected as the constraint condition, because the maximum current is calculated according to other limiting conditions.

[0089] Step S2 is specifically as follows:

[0090] Since the simultaneous identification of the electro-chemical-thermal coupling model will cause mutual coupling between variables, resulting in greater difficulty in parameter identification, the electrochemical model and the thermal model are identified separately, and the grey wolf optimization algorithm is used to identify the relevant parameters.

[0091] For the parameters of the electrochemical model, the main problems in the identification process are the identification time problem, overfitting problem and convergence problem.

[0092] Identification time problem: Even for the reconstructed electrochemical model, during the parameter identification process, if the amount of input data is too large, it will also take a long time to identify the parameters.

[0093] Convergence problem: During the identification process, some intermediate calculation values may exceed the boundary, resulting in complex numbers appearing during the calculation. This situation where complex numbers appear is called non-convergence.

[0094] Overfitting problem: During the identification process, it may occur that a set of parameters is only applicable to one working condition and not applicable to other working conditions. This situation is called overfitting.

[0095] A multi-objective grey wolf optimization algorithm with variable input is designed for the three problems of electrochemical parameter identification that occur.

[0096] Since the non-convergence problem mainly exists in the best calculated data, the identification data is divided according to the number of iterations. As the number of iterations increases, the identification data also increases. This not only avoids the non-convergence problem but also improves the identification speed.

[0097] S2.1: Use the two working conditions of 1C discharge and DST dynamic working conditions as the objective fitness function of the variable-scale multi-objective grey wolf optimization algorithm to obtain the simulated voltage V of the electro-thermal coupling model. sim and the actual voltage V exp The fitness value fv of the mean square error between them is expressed as:

[0098]

[0099] where N is the number of voltage acquisitions, V exp is the actual voltage, and V sim is the simulated voltage;

[0100] S2.2: Calculate the absolute value fc of the difference between the positive electrode capacity and the negative electrode capacity, and the expression is:

[0101]

[0102] where A is the electrode area, L + is the positive electrode thickness, L - is the negative electrode thickness, is the positive electrode solid-phase volume fraction, is the negative electrode solid-phase volume fraction, is the maximum lithium-ion concentration in the positive electrode, is the maximum lithium-ion concentration in the negative electrode, F is the Faraday constant, SOL is the SOC of the single-cell electrode, SOC is the remaining charge of the single-cell battery, is the value corresponding to the positive electrode of the single-cell battery at 0, is the value corresponding to the negative electrode of the single-cell battery SOC at 0, is the value corresponding to the positive electrode of the single-cell battery at 100, is the value corresponding to the negative electrode of the single-cell battery at 100;

[0103] S2.3: Calculate the weighted final fitness function fm of the electro-thermal coupling model, and the expression is:

[0104] fm = w1fv1 + w2fv2 + w3fc

[0105] Among them, W1, W2, and W3 are the weights of different fitness functions, fv1 is the mean square error of the 1C discharge condition, and fv2 is the mean square error of the DST dynamic stress condition;

[0106] S2.4: Identify the temperature parameters using the 1C discharge condition to obtain the local root mean square error ft of the temperature. The expression is:

[0107]

[0108] Among them, ft is the local root mean square error, T exp refers to the experimentally measured temperature, T sim refers to the simulated temperature.

[0109] The specific steps of step S4 are as follows:

[0110] S4.1: Select the surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature as the safety constraints of the SOP. The expression is:

[0111]

[0112] c e (x,t) ≥ 0

[0113]

[0114] 0 ≤ SOC(t) ≤ 1

[0115] V min ≤ V(t) ≤ V max

[0116] T min ≤ T(t) ≤ T max

[0117] Among them, and are the maximum lithium intercalation ranges of the positive and negative electrodes of the battery, is the maximum concentration of the positive and negative electrodes of the battery, is the surface lithium ion concentration of the positive and negative electrodes of the battery, η sr (x L- ,t) is the lithium plating overpotential, SOC(t) is the SOC of the current battery, V min is the minimum voltage of the battery, V max is the maximum voltage of the battery, V(t) is the current voltage of the battery, T min is the minimum temperature of the battery, T min is the maximum temperature of the battery, T min is the current temperature of the battery, c e (x, t) is the electrolyte concentration;

[0118] The power prediction of a lithium battery can be defined as the maximum constant charging or discharging current that can be applied to the battery within a certain time period Δt without exceeding several SOA ranges defined in 3.31. However, due to different time periods, the power prediction of the battery under the same working conditions is also different. Therefore, the power prediction for different time ranges is also important in battery management. In the technical solution, the power prediction for three time periods of 1s, 30s, and 60s is considered, meeting long-term and short-term predictions.

[0119] S4.2: Select the bisection method for SOP prediction of the battery at different time scales;

[0120] S4.3: Select Gaussian process regression for accelerated prediction of SOP. Specifically:

[0121] Select the DST dynamic stress condition as the training set, use SOC, voltage, and temperature as inputs, and the SOP predicted by the bisection method and the 95% confidence interval of the SOP as outputs for offline training;

[0122] When predicting the SOP under other working conditions, obtain its 95% confidence interval through the offline model, and use the interval as the current maximum and minimum current boundaries, which speeds up the convergence rate of the bisection method.

[0123] The specific steps of S4.2 are as follows:

[0124] S4.2.1: Set the maximum boundary current I 0，max to 20C, and set the minimum boundary current I 0，min to 0C;

[0125] S4.2.2: Let the initial maximum power current I 0，SOP be equal to the initial maximum boundary current I 0，max , and calculate the constraint state for different time periods Δt by passing a constant current through the electrothermal coupling model;

[0126] S4.2.3: Determine whether the six states of surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature all satisfy the safety constraints. If so, the initial maximum power current I 0,SOP is the maximum constant current I k,SOP for this state and this time period, and enter step S4.2.8. Otherwise, update through the bisection method and enter step S4.2.4;

[0127] S4.2.4: Let the maximum constant current I k,sop be equal to half of the sum of the maximum boundary current I k，max and the minimum boundary current I k，min , and calculate the constraint state for different time periods Δt by passing the maximum constant current I k,sop through the electrothermal coupling model;

[0128] When predicting the SOP at a certain point, the state of this point is used as the initial value of eSPM, and a constant current is applied for a period of time. Although the current is not used as a constraint for power prediction, the maximum current and the minimum current are required as the boundaries of the dichotomy method. Therefore, the current boundaries (also the initial minimum current I 0,min and the maximum current I 0，max ) are set to a minimum of 0C and a maximum of 20C. Then, the initial maximum power current I 0，SOP is set equal to the initial maximum current I 0，max . Then, with this constant current, the constraint states at different time intervals Δt are calculated through eSPM. If all six final states satisfy the safety constraints, then the I 0,SOP at this time is the maximum constant current for this state and this time period; if any one of the states exceeds the safety constraints, update and iterate through the dichotomy method. Until the maximum constant current I k,SOP in the Kth iteration satisfies all constraints. However, since it is the maximum current that needs to satisfy the constraints, in order to make the predicted current close to the maximum current that satisfies the constraints, a maximum tolerance tol equal to 0.1 is set, so that the predicted current is as close as possible to the boundary current

[0129] Judge whether the six states of the surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature all satisfy the safety constraints,

[0130] S4.2.6: No, then set the maximum boundary current I k，max as the maximum constant current I k,sop . With the maximum constant current I k,sop , calculate the constraint states at different time intervals Δt through the electro-thermal coupling model, and return to step S4.2.5. Yes, then enter step S4.2.7;

[0131] S4.2.7: Judge whether the maximum tolerance tol is satisfied. The judgment expression is:

[0132] Ik,max - Ik, SOP | ≤ tol

[0133] Ik,min - Ik, SOP | ≤ tol

[0134] where I k,max is the maximum boundary current in the kth iteration, Ik ,min is the minimum boundary current in the kth iteration, I k,SOP is the maximum constant current, and tol is the maximum tolerance;

[0135] No, then set the minimum boundary current I k，min equal to the maximum constant current Ik,SOP , and return to step S4.2.5,

[0136] If yes, enter step S4.2.8:

[0137] S4.2.8: Iteration stops, calculate the final maximum constant current I sop(t) , and the expression is:

[0138] I SOP (t) = I k,SOP

[0139] S4.2.9: Through the maximum constant current I sop(t) and the electro-thermal coupling model, calculate the minimum terminal voltage under the time period of Δt and calculate the maximum charging power and the maximum discharging power The maximum charging power can be obtained from the maximum constant charging current at the current moment and the minimum voltage during Δt. The maximum discharging power can be obtained from the maximum constant discharging current at the current moment and the minimum voltage during Δt. The expression is:

[0140]

[0141] Among them, is the maximum charging power at this moment, is the maximum charging current at this moment, is the terminal voltage at this moment, is the discharging and charging power at this moment, is the maximum charging current at this moment, is the minimum terminal voltage after calculating Δt.

[0142] In summary, for the battery SOP fast prediction method based on the electrochemical model, through a variable-scale multi-objective grey wolf optimization algorithm, 21 electrochemical model parameters are quickly identified. By changing the input identification scale strategy during the optimization process, the non-convergence problem that appears during the optimization process is avoided; through a multi-objective strategy, by selecting the proportion of fitness values of various working conditions through weights, the overfitting problem in parameter identification is avoided;

[0143] Moreover, through Gaussian process regression, the relationship between SOP and its 95% confidence interval and voltage, SOC, and temperature is learned offline. When predicting SOP, the interval range of the bisection method is shortened, making the bisection method converge more quickly, accelerating the prediction of SOP, and enabling the online use of SOP prediction.

[0144] It should be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or apparatus including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the phrase "including an..." does not exclude the presence of additional identical elements in the process, method, article or apparatus including the said element.

[0145] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A fast prediction method for battery SOP based on an electrochemical model, characterized in that, It includes the following steps: S1: Simplify the P2D model to an e-SPM model and couple it with the thermal model to obtain an electro-thermal coupling model; S2: Identify the thermal parameters and electrochemical parameters of the electro-thermal coupling model respectively through a variable-scale multi-objective grey wolf optimization algorithm; S3: Use the unscented Kalman filter as the state observer of the electrochemical model to observe the internal and external states of the battery; S4: Select the surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature as the safety constraints of the SOP. Select the bisection method to predict the SOP of the battery at different time scales, and select the Gaussian process regression to accelerate the convergence speed of the bisection method for predicting the battery SOP; The specific content of step S2 is as follows: S2.1: Use the 1C discharge and DST dynamic conditions as the objective fitness functions of the variable-scale multi-objective grey wolf optimization algorithm to obtain the simulated voltage V of the electro-thermal coupling model sim and the actual voltage V exp The fitness value fv of the mean square error between them is expressed as: Among them, N is the number of voltage acquisitions, V exp is the actual voltage, V sim is the analog voltage; S2.2: Calculate the absolute value fc of the difference between the positive electrode capacity and the negative electrode capacity. The expression is: Among them, A is the electrode area, L + is the positive electrode thickness, L - is the negative electrode thickness, is the positive electrode solid-phase volume fraction, is the positive electrode solid-phase volume fraction, is the maximum lithium-ion concentration of the positive electrode, is the maximum lithium-ion concentration of the negative electrode, F is the Faraday constant, SOL is the SOC of the single-cell electrode, SOC is the remaining charge of the single-cell battery, is the value corresponding to the positive electrode of the single-cell battery at 0, is the value corresponding to the negative electrode of the single-cell battery SOC at 0, is the value corresponding to the positive electrode of the single-cell battery at 100, is the value corresponding to the negative electrode of the single-cell battery at 100; S2.3: Calculate the final fitness function fm of the electro-thermal coupling model through weighted calculation. The expression is: fm = w1fv1 + w2fv2 + w3fc Among them, W1, W2, and W3 are the weights of different fitness functions. fv1 is the mean square error under the 1C discharge condition, and fv2 is the mean square error under the DST dynamic stress condition; S2.4: Use the 1C discharge condition to identify the temperature parameter and obtain the local root mean square error ft of the temperature. The expression is: Among them, ft is the root mean square error of the official side, and T exp refers to the temperature measured in the experiment, and T sim refers to the simulated temperature.

2. The method for quickly predicting the battery SOP based on the electrochemical model according to claim 1, characterized in that: The internal and external states of the battery include voltage, SOC, surface lithium ion concentration, electrolyte concentration, and lithium plating overpotential.

3. The rapid prediction method of battery SOP based on an electrochemical model according to claim 1, characterized in that The specific steps of step S4 are as follows: S4.1: Select the surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature as the safety constraints of the SOP. The expression is: c e (x, t) ≥ 0 η sr (x L- ,t)≥0 0 ≤ SOC(t) ≤ 1 V min V(t) is less than or equal to V and greater than or equal to V max T min T(t) is less than or equal to T and greater than or equal to T max Among them, and are the maximum lithium intercalation ranges of the positive and negative electrodes of the battery, is the maximum concentration of the positive and negative electrodes of the battery, is the surface lithium ion concentration of the positive and negative electrodes of the battery, is the lithium deposition overpotential, SOC(t) is the SOC of the current battery, V min is the minimum voltage of the battery, V max is the maximum voltage of the battery, V(t) is the current voltage of the battery, T min is the minimum temperature of the battery, T min is the maximum temperature of the battery, T min is the current temperature of the battery, c e (x, t) is the electrolyte concentration; S4.2: Select the bisection method to predict the SOP of the battery at different time scales; S4.3: Select the Gaussian process regression for accelerated prediction of the SOP.

4. A method for quickly predicting the battery SOP based on an electrochemical model according to claim 3, characterized in that: The specific steps of S4.2 are as follows: S4.2.1: Set the maximum boundary current I 0，max to 20C and the minimum boundary current I 0，min to 0C; S4.2.2: Let the initial maximum power current I 0，SOP be equal to the initial maximum boundary current I 0，max , and calculate the constraint states at different time intervals Δt by passing a constant current through the electro-thermal coupling model; S4.2.3: Determine whether the six states of surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature all meet the safety constraints. If so, the initial maximum power current I 0,SOP is the maximum constant current I in this state and this time period k,SOP , enter step S4.2.

8. If not, update by the bisection method and enter step S4.2.4; S4.2.4: Let the maximum constant current I k,sop be equal to half of the sum of the maximum boundary current I k，max and the minimum boundary current I k，min to calculate the constraint states at different time intervals Δt through the electrothermal coupling model with the maximum constant current I k,sop ; S4.2.5: Judge whether the six states of the surface lithium ion concentration, electrolyte concentration, lithium plating overpotential, voltage, SOC, and temperature all satisfy the safety constraints, S4.2.6: No, then let the maximum boundary current I k，max is the maximum constant current I k,sop , with a maximum constant current I k,sop Calculate the constraint state of different time periods Δt through the electrothermal coupling model and return to step S4.2.

5. If yes, proceed to step S4.2.7; S4.2.7: Judge whether the maximum tolerance tol is satisfied. The judgment expression is: |I k,max -I k,SOP |≤tol |I k,min -I k,SOP |≤tol where I k,max is the maximum boundary current at the k-th iteration, I k,min is the minimum boundary current at the k-th iteration, I k,SOP is the maximum constant current, and tol is the maximum tolerance; Otherwise, set the minimum boundary current I k，min equal to the maximum constant current I k,SOP , and return to step S4.2.5 If yes, enter step S4.2.8: S4.2.8: Iteration stops and the final maximum constant current I is calculated sop(t) , and the expression is: I SOP (t) = I k,SOP S4.2.9: Through the maximum constant current I sop(t) and calculate the minimum terminal voltage under the time period Δt through the electro-thermal coupling model and calculate the maximum charging power and the maximum discharging power The expressions are as follows: Among them, is the maximum charging power at this moment, is the maximum charging current at this moment, is the terminal voltage at this moment, is the discharge charging power at this moment, is the maximum charging current at this moment, is the minimum terminal voltage after calculating Δt.

5. The rapid prediction method of battery SOP based on an electrochemical model according to claim 3, characterized in that The specific content of step S4.3 is as follows: Select the DST dynamic stress condition as the training set, use SOC, voltage, and temperature as the inputs, and use the SOP predicted by the bisection method and the 95% confidence interval of the SOP as the outputs for offline training; When predicting the SOP under other conditions, obtain its 95% confidence interval through the offline model, and use the interval as the current maximum and minimum current boundaries, which accelerates the convergence speed of the bisection method.

Images