Open-circuit voltage curve reconstruction method for vehicle-mounted lithium ion battery

Abstract

The invention provides an open-circuit voltage curve reconstruction method for a vehicle-mounted lithium ion battery, and belongs to the technical field of battery management systems, and the method comprises the steps: collecting and storing multi-stage constant-current charging data of the vehicle-mounted lithium ion battery, and carrying out the segmentation processing of the data; establishing a first-order equivalent circuit model of the battery, and identifying optimal parameters of the model by using a white whale optimization algorithm enhanced by a random reset strategy; solving an open-circuit voltage curve in each segment according to the model parameters; and reconstructing the open-circuit voltage curve combination. According to the invention, the reconstruction of the open-circuit voltage curve of the battery can be completed only through the data of the single charging of the vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery management systems, and in particular relates to a method for reconstructing an open circuit voltage curve of a vehicle-mounted lithium-ion battery. Background Art

[0002] The open circuit voltage curve is an important feature of lithium-ion batteries. It can be used to estimate the battery state of charge (SOC) and diagnose the health status, and plays an important role in the vehicle battery management system. However, as the battery ages, the shape and position of the open circuit voltage curve will change. If the curve is not updated, the performance of the battery management system will be reduced, such as an increase in SOC error.

[0003] At present, the research on the reconstruction of the open circuit voltage curve mainly falls into the following categories: 1) Test-based methods, which reduce the static time by applying different current excitations to the battery, thereby reducing the test time. However, the test requires special equipment and cannot be performed in daily use by vehicle users. 2) Model-based methods, by establishing a battery model, such as an equivalent circuit model, and using a parameter identification algorithm to identify the model parameters, the open circuit voltage is solved. However, the stability of the identification results is greatly affected by the operating conditions of the battery. 3) Data-driven methods, using Gaussian process regression, back propagation neural networks, deep neural networks, etc. to predict the open circuit voltage of the battery, but this method requires a lot of data support and is difficult to apply in practice.

[0004] Multi-stage constant current charging is a common charging method for electric vehicles. During the charging process, the charging current decreases step by step, which can effectively control the charging temperature rise and extend the battery life. Unlike constant current and constant voltage charging, the multi-stage constant current charging process includes the constant current stabilization process and the dynamic process when the constant current stage is switched, which can be used to reconstruct the open circuit voltage curve. Reconstructing the open circuit voltage curve using the battery charging data can greatly reduce the time cost required for testing and is also very convenient for users. However, few studies focus on reconstructing the open circuit voltage curve using the multi-stage constant current charging data of the battery.

[0005] Application number 202311360664.4, invention name A lithium-ion battery health diagnosis method based on open circuit voltage curve reconstruction, discloses the use of battery management system data of a cloud-based energy storage system, using the battery open circuit voltage data set collected and stored by BMS, analyzing the curve characteristics of the open circuit voltage data points and reconstructing an incomplete curve of the state of charge-open circuit voltage relationship; based on the offline complete open circuit voltage curve in the previous experiment, the complete open circuit voltage curve is reconstructed by using the voltage curve similarity matching method; based on the curve change characteristics of the high state of charge area of ​​the open circuit voltage curve after battery aging, battery health diagnosis is achieved.

[0006] Application number CN202110615632.9, invention name A data-driven method for reconstructing the open-circuit voltage curve of a lithium-ion battery, discloses collecting and storing the open-circuit voltage data set of the battery through the battery management system BMS, then preprocessing the stored open-circuit voltage data set, and finally reconstructing the open-circuit voltage curve using the preprocessed open-circuit voltage data. This invention solves the problem of interference with the open-circuit voltage curve caused by SOC estimation error in the traditional open-circuit voltage curve acquisition method, and avoids the problem that the traditional open-circuit voltage curve acquisition method relies on small current conditions.

[0007] Invention name: A method for reconstructing the open circuit voltage curve for on-board lithium-ion batteries, which uses multi-stage constant current charging data of on-board lithium-ion batteries, processes the data in segments and establishes a battery model, uses the White Whale optimization algorithm enhanced by a random reset strategy to identify the optimal parameters of the model, and then solves the open circuit voltage curve in each segment. The invention greatly reduces the time cost of obtaining the open circuit voltage curve and has high accuracy, providing an efficient solution for reconstructing the open circuit voltage curve using multi-stage constant current charging data.

[0008] Therefore, there is an urgent need to develop a method to reconstruct the battery open circuit voltage curve using charging data. Summary of the invention

[0009] The purpose of the present invention is to solve the defects of the above-mentioned prior art and provide a method for reconstructing the open circuit voltage curve of a vehicle-mounted lithium-ion battery, so that the open circuit voltage curve of the lithium-ion battery can be reconstructed using the charging data of the vehicle.

[0010] The present invention adopts the following technical solution:

[0011] A method for reconstructing an open circuit voltage curve of a vehicle-mounted lithium-ion battery, comprising:

[0012] S1. The on-board battery management system collects and stores the battery status data of the lithium-ion battery during the multi-stage constant current charging process, including current, voltage, state of charge (SOC) and temperature data;

[0013] S2. According to the current change during the multi-stage constant current charging process, the data is processed in segments;

[0014] S3. Establish a first-order RC equivalent circuit model of the battery for each of the segments in S2;

[0015] S4. Use the White Whale optimization algorithm enhanced by random reset strategy to optimize the model parameters;

[0016] S5. solving the open circuit voltage curve of each segment according to the optimal model parameters;

[0017] S6. Combine all segmented open circuit voltage curves, smooth the curves, use a polynomial model to fit the open circuit voltage-state of charge (SOC) curve, and the reconstruction is completed.

[0018] Furthermore, in step S1, the multi-stage constant current charging process should include at least 3 constant current stages, and the entire charging process starts when the SOC is lower than 10% and ends when the SOC is higher than 90%.

[0019] Furthermore, the step 2 is specifically as follows:

[0020] S2.1. Intercept all the segments where the current suddenly changes when charging starts and switching between the constant current phase, and record them as A, A = [A0, A1, ..., A n ], where A i represents the i-th segment, each segment length is m, m can be set to different time lengths, such as 300 seconds;

[0021] S2.2. After the entire charging process is cut off from segment A, all remaining segments are recorded as B, where B = [B0, B1, ..., B n ], where B j represents the jth segment.

[0022] Furthermore, in step S3, the first-order RC equivalent circuit model is only for segment A. i Modeling is performed, and the open circuit voltage source U in the model ocv Using a linear model instead, Where k is the sampling time point, k = 0, 1, 2, ..., Δt is the sampling time interval, μ and λ are parameters. The parameter set to be solved for the entire model is [μ, λ, R o , R p , C p ], where R o ,R p and C p They represent ohmic internal resistance, polarization internal resistance and polarization capacitance respectively.

[0023] Furthermore, in step S4, the White Whale optimization algorithm enhanced by the random reset strategy is specifically as follows: after the White Whale optimization algorithm updates the parameter position, it is checked whether the new parameter position exceeds the set parameter boundary. If it exceeds, the random reset strategy is used to adjust the parameter position, and the adjustment formula is:

[0024]

[0025] in, represents the new position of the i-th individual in the j-th dimension obtained by the white whale algorithm, T represents the current number of iterations, It represents the position after random reset strategy adjustment. lb and ub represent the lower and upper bounds of the parameter respectively. rand() is a function that generates random numbers between the interval [0,1]. d is an adjustment factor, which is set to 0.3 by default.

[0026] Furthermore, in step S4, when optimizing the model parameters, the loss function is defined as the root mean square error between the actual terminal voltage of the battery and the terminal voltage estimated by the model, and the calculation formula is:

[0027]

[0028] Where n is the number of data points, U t is the real terminal voltage, U est Estimate the terminal voltages for the model.

[0029] Furthermore, step S5 includes the following steps:

[0030] S5.1. According to the first-order RC equivalent circuit model, U ocv The linear model is used to solve the open circuit voltage curves of all segments in A;

[0031] S5.2. Solve the open circuit voltage curves of all segments in B. The calculation formula is:

[0032] U ocv =U t -I(R o +R p )

[0033] Where I is the current value, which is positive during charging. If the maximum temperature difference during charging does not exceed 10°C, then R o and R p Take the average of all fragment identification results in A. If the maximum temperature difference during charging exceeds 10°C, use a linear model to fit the temperature and R o and R p The relationship between R and o and R p The value of .

[0034] Furthermore, step S6 includes the following steps:

[0035] S6.1. Connect the open circuit voltage curves corresponding to all the segments in A and B according to their positional relationship, and use a Savitzky-Golay filter to smooth the combined open circuit voltage curve;

[0036] S6.2. Obtain a curve of open circuit voltage with respect to SOC according to the SOC data recorded by the battery management system;

[0037] S6.3. Use a p-order polynomial model to fit the relationship between open circuit voltage and SOC, where p is at least 6th order and can be adjusted based on the fitting effect.

[0038] Beneficial effects of the present invention:

[0039] The present invention can directly reconstruct the open circuit voltage curve of the battery pack by using the vehicle's single multi-stage constant current charging data, thereby saving time and cost while having good accuracy.

[0040] The present invention can update the open circuit voltage curve of the battery in daily use of the vehicle without performing other special tests on the battery, thereby improving the reliability of the battery management system and playing an important role in the safety management of the vehicle battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 The overall flow chart of the method proposed by the present invention;

[0042] Figure 2 The data curves of the multi-stage constant current charging process in the method of the present invention; (a) represents the current curve; (b) represents the voltage curve; (c) represents the SOC curve; and (d) represents the temperature curve.

[0043] Figure 3 It is the result of segmenting the charging data in the method of the present invention.

[0044] Figure 4 It is a schematic diagram of a first-order equivalent current model in the method of the present invention.

[0045] Figure 5 It is a flow chart of the White Whale optimization algorithm enhanced by the random reset strategy in the method of the present invention.

[0046] Figure 6 It is the result of reconstructing the open circuit voltage curve in the preferred embodiment of the present invention; (a) represents the comparison between the reconstructed open circuit voltage curve and the real open circuit voltage curve; (b) represents the absolute error of the reconstructed open circuit voltage curve. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Example

[0049] like Figure 1As shown, a method for reconstructing an open circuit voltage curve of a vehicle-mounted lithium-ion battery of the present invention comprises the following steps:

[0050] S1. The on-board battery management system collects and stores the battery status data of the lithium-ion battery during the multi-stage constant current charging process, including current, voltage, state of charge (SOC) and temperature data.

[0051] S2. According to the current change during the multi-stage constant current charging process, the data is processed in segments;

[0052] S3. Establish a first-order RC equivalent circuit model of the battery for each of the segments in S2;

[0053] S4. Use the White Whale optimization algorithm enhanced by the random reset strategy to optimize the model parameters of S3;

[0054] S5. solving the open circuit voltage curve of each segment according to the optimal model parameters;

[0055] S6. Combine all segmented open circuit voltage curves, smooth the curves, use a polynomial model to fit the open circuit voltage-state of charge (SOC) curve, and the reconstruction is completed.

[0056] In a preferred embodiment of the present invention, in step S1, the multi-stage constant current charging process should include at least 3 constant current stages, and the entire charging process starts when the SOC is lower than 10% and ends when the SOC is higher than 90%.

[0057] In this embodiment: the experimental battery is a ternary lithium-ion battery with a rated capacity of 6.9Ah. Before the experiment, the battery is discharged to an SOC of less than 10%. Use charging and discharging equipment to perform multi-stage constant current charging on the battery: specifically, in the first stage, the battery is charged with a current of 1C until the battery voltage reaches 3.8V, in the second stage, the battery is charged with a current of 0.75C until the battery voltage reaches 3.95V, in the third stage, the battery is charged with a current of 0.5C until the battery voltage reaches 4.1V, and in the fourth stage, the battery is charged with a current of 0.25C until the battery voltage reaches 4.2V, and the charging is completed. Record the data (current, voltage, SOC, temperature) during the charging process, such as Figure 2 shown.

[0058] like Figure 3 As shown, in a preferred embodiment of the present invention, step S2 includes:

[0059] S2.1. Intercept all the segments where the current suddenly changes when charging starts and switching between the constant current phase, and record them as A, A = [A0, A1, ..., A n ], where A i represents the i-th segment, each segment length is m, m can be set to different time lengths, such as 300 seconds;

[0060] S2.2. After the entire charging process is cut off from segment A, all remaining segments are recorded as B, where B = [B0, B1, ..., B n ], where B j represents the jth segment.

[0061] For multi-stage constant current charging, there are several constant current stages corresponding to several A i and B i In the charging data of this embodiment, a 4-stage constant current charging is adopted, so there are 4 A i and 4 Bs i Fragments. The specific segmentation method is:

[0062] (1). Intercept A i Fragment, according to the setting A i The length of the segment is m. A1 corresponds to a segment of length m taken from the start of charging. A2 corresponds to switching from the first constant current stage to the second constant current stage, taking 2 data sampling points forward from the switching point as the starting point, and then cutting a segment of length m. A3 corresponds to switching from the second constant current stage to the third constant current stage, taking 2 data sampling points forward from the switching point as the starting point, and then cutting a segment of length m. A4 corresponds to switching from the third constant current stage to the fourth constant current stage, taking 2 data sampling points forward from the switching point as the starting point, and then cutting a segment of length m.

[0063] (2) Determine B i Segments, B1 corresponds to the data segment from the end point of segment A1 to the start point of segment A2. B2 corresponds to the data segment from the end point of segment A2 to the start point of segment A3. B3 corresponds to the data segment from the end point of segment A3 to the start point of segment A4. B4 corresponds to the data segment from the end point of segment A4 to the end of charging. In this way, all A1 and B are obtained. i Fragment.

[0064] like Figure 4 As shown, in a preferred embodiment of the present invention, in step S3, the first-order RC equivalent circuit model is only for segment A i Modeling is performed, and the open circuit voltage source U in the model ocv Using a linear model instead, Where k is the sampling time point, k = 0, 1, 2, ..., Δt is the sampling time interval, μ and λ are parameters. The parameter set to be solved for the entire model is [μ, λ, R o , R p , C p ], where R o ,R p and C pThey represent ohmic internal resistance, polarization internal resistance and polarization capacitance respectively.

[0065] The specific formula of the first-order RC equivalent circuit model in this embodiment is as follows:

[0066] U t =U ocv +IR o +U p (1)

[0067]

[0068] U p represents the polarization voltage, represents the time derivative of the polarization voltage, U ocv represents the open circuit voltage OCV, and I is the current (positive during charging and negative during discharging). Writing the above equations (1) and (2) in discrete form yields:

[0069] U t (k+1)=U ocv (k+1)+R0I(k+1)+U p (k+1) (3)

[0070]

[0071] k is the discrete time step, Δt is the sampling time interval, and in the charging data used for verification, the sampling time interval is 1s. Substituting formula (4) into formula (3), eliminating U p , we can get:

[0072]

[0073] Since the open circuit voltage OCV changes slowly, it can be assumed that the open circuit voltages in adjacent time steps are equal, that is, U ocv (k+1)=U ocv (k) In addition, Formula (5) can be simplified as:

[0074] U t (k) = (1-a)U ocv (k)+aU t (k-1)+cR p -a(R o +R p ))I(k-1)+R o I(k) (6)

[0075] At this time, the linear model of OCV is as follows:

[0076]

[0077] Substituting into formula (6), we can get formula (8):

[0078]

[0079] Observing formula (8), we can see that:

[0080] U t is the terminal voltage, corresponding to the voltage in the charging data, which is known; Δt is the sampling time interval, which is 1s, and I is the charging current, which is known in the charging data:

[0081] Therefore, the unknown parameters θ = [μ, λ, R o , R p , C p ]

[0082] like Figure 5 As shown, in a preferred embodiment of the present invention, in step S4, after the White Whale optimization algorithm updates the parameter position, it is checked whether the new parameter position exceeds the set parameter boundary. If it exceeds, the parameter position is adjusted using a random reset strategy, and the adjustment formula is:

[0083]

[0084] in, represents the new position of the i-th individual in the j-th dimension obtained by the White Whale optimization algorithm, T represents the current number of iterations, It represents the position after random reset strategy adjustment. lb and ub represent the lower and upper bounds of the parameter respectively. rand() is a function that generates random numbers between the interval [0,1]. d is an adjustment factor, which is set to 0.3 by default.

[0085] In a preferred embodiment of the present invention, in step S4, when optimizing the model parameters, the loss function is defined as the root mean square error between the actual terminal voltage of the battery and the terminal voltage estimated by the model, and the calculation formula is:

[0086]

[0087] Where n is the number of data points, U t is the real terminal voltage, U est Estimate the terminal voltages for the model.

[0088] In this embodiment:

[0089] After determining the parameters to be optimized, substitute the following specific numerical values ​​for calculation:

[0090] Define input data X(k) = [U t(k-1), I(k-1), I(k), k], k = 1, 2, 3, ..., Substitute the voltage and current data collected during the charging process into X(k). The right side of the equal sign in formula (8) is actually a function of the input data X(k) and the parameter set θ, denoted as f(X(k), θ). Now we can get U t (k) = f(X(k), θ), in this formula, only θ is unknown.

[0091] The following uses the White Whale optimization algorithm enhanced by the random reset strategy to identify the optimal parameter θ of the model. The optimal θ is the voltage value calculated by f(X(k),θ) and the actual voltage value U t The error between (k) is the smallest, and the root mean square error is used here to measure it. The calculation formula is:

[0092]

[0093] U est (k) is the voltage value calculated by f(X(k), θ). This formula calculates the error between the voltage estimated by the battery model and the actual voltage.

[0094] The specific steps of the random reset strategy enhanced Beluga optimization algorithm are as follows: (1) Set the population size and number of iterations, randomly initialize the position of the Beluga population, and each position of a Beluga corresponds to a candidate solution θ. According to the loss function, find the Beluga individual with the smallest current loss function value, that is, the optimal individual.

[0095] (2) According to the balance factor B f Determine whether the algorithm is in the exploration stage or the development stage. f When it is greater than 0.5, it is in the exploration stage, and when it is less than or equal to 0.5, it is in the development stage. Different White Whale position update strategies are implemented in the two stages.

[0096] (3) Further judgement of B f Is it less than the whale fall probability W? f If the condition is met, the white whale position will be further updated by whale fall.

[0097] (4) Check whether the updated white whale position exceeds the parameter boundary. For each parameter in θ, its normal upper and lower parameters are set. If it exceeds, the random reset strategy is used to adjust the parameter position. The adjustment formula is:

[0098]

[0099] in, represents the new position of the i-th individual in the j-th dimension obtained by the White Whale optimization algorithm, T represents the current number of iterations, It represents the position after random reset strategy adjustment. lb and ub represent the lower and upper bounds of the parameter respectively. rand() is a function that generates random numbers between the interval [0,1]. d is an adjustment factor, which is set to 0.3 by default.

[0100] In this embodiment, the number of iterations is set to 1000, the population size is set to 200, and the four A i The parameter optimization results on the fragment are:

[0101] Snippet <![CDATA[R o (mΩ)]]> <![CDATA[R p (mΩ)]]> a μ(V) λ Temperature(℃) <![CDATA[A1]]> 28.6 1.5 0.98 3.31 1401.1 24.16 <![CDATA[A2]]> 26.1 0.8 0.91 3.61 11944.4 23.53 <![CDATA[A3]]> 25.7 1.2 0.94 3.81 9925.7 23.94 <![CDATA[A4]]> 25.5 0.8 0.94 4.01 16046.1 23.16

[0102] In a preferred embodiment of the present invention, step S5 comprises the following steps:

[0103] S5.1. According to the first-order RC equivalent current model U ocv The linear model is used to solve the open circuit voltage curves of all segments in A;

[0104] S5.2. Solve the open circuit voltage curves of all segments in B. The calculation formula is:

[0105] U OCV =U t -I(R o +R p )

[0106] Where I is the current value, which is positive during charging. If the maximum temperature difference during charging does not exceed 10°C, then R o and R p Take the average of all fragment identification results in A. If the maximum temperature difference during charging exceeds 10°C, use a linear model to fit the temperature and R o and R p The relationship between R and o and R p The value of .

[0107] In a preferred embodiment of the present invention, step S6 comprises the following steps:

[0108] S6.1. Connect the open circuit voltage curves corresponding to all the segments in A and B according to the positional relationship, and use a Savitzky-Gol ay filter to smooth the combined open circuit voltage curve;

[0109] S6.2. According to the SOC data recorded by the battery management system, a curve of the open circuit voltage with respect to the SOC is obtained.

[0110] S6.3. Use a p-order polynomial model to fit the relationship between open circuit voltage and SOC, where p is at least 6th order and can be adjusted based on the fitting effect.

[0111] In this embodiment, the implementation object is a ternary lithium-ion battery, and an 8th-order polynomial model is used to fit the relationship between open circuit voltage and SOC. The comparison between the reconstructed OCV curve and the real curve is as follows: Figure 6 As shown in the figure, it can be seen that the open circuit voltage curve reconstructed by the present invention is very close to the real open circuit voltage curve, with a maximum error of no more than 15mV, and has a very high accuracy. i R in the fragment o and R p Mean, determine R o 26.5mΩ, R p is 1.1 mΩ, which is used to calculate the open circuit voltage curve of all segments in B. The calculation formula is:

[0112] U ocv =U t -I(R o +R p )

[0113] Where U t and I are voltage and current respectively, which are known in the charging data.

[0114] Now we have obtained the OCV of all A1-A4 and B1-B4 segments, which are spliced ​​according to their positional relationship, and then the 8th-order polynomial fitting is used to obtain the attached Figure 6 The open circuit voltage reconstruction results in .

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for reconstructing an open circuit voltage curve of a vehicle-mounted lithium-ion battery, characterized in that ,include: S1. The vehicle battery management system collects and stores the battery status data of the lithium-ion battery during the multi-stage constant current charging process, including current, voltage, state of charge and temperature data; S2. According to the current change during the multi-stage constant current charging process, the data is processed in segments; S3. Establish a first-order RC equivalent circuit model of the battery for each of the segments in S2; S4. Use the White Whale optimization algorithm enhanced by random reset strategy to optimize the model parameters; S5. solving the open circuit voltage curve of each segment according to the optimal model parameters; S6. Combine all segmented open circuit voltage curves, smooth the curves, use a polynomial model to fit the open circuit voltage-state of charge curve, and the reconstruction is completed.

2. The method according to claim 1, characterized in that: In step S1, the multi-stage constant current charging process should include at least 3 constant current stages, and the entire charging process starts when the state of charge is lower than 10% and ends when the state of charge is higher than 90%.

3. The method according to claim 1, characterized in that include: S2.

1. Intercept all the segments where the current suddenly changes when charging starts and switching between the constant current phase, and record them as A, A = [A0, A1, ..., A n ], where A i represents the i-th segment, each segment length is m, m is set to different time lengths; S2.

2. After the entire charging process is cut off from segment A, all the remaining segments are recorded as B, where B = [B0, B1, ..., B n ], where B j represents the jth segment.

4. The method according to claim 1, characterized in that: In step S3, the first-order RC equivalent circuit model is only for segment A. i Modeling is performed, and the open circuit voltage source U in the model ocv Using a linear model instead, Where k is the sampling time point, k = 0, 1, 2, ..., Δt is the sampling time interval, μ and λ are parameters, and the parameter set to be solved for the entire model is [μ, λ, R o , R p , C p ], where R o ,R p and C p They represent ohmic internal resistance, polarization internal resistance and polarization capacitance respectively.

5. The method according to claim 1, characterized in that In step S4, the White Whale optimization algorithm enhanced by the random reset strategy is specifically as follows: after the White Whale optimization algorithm updates the parameter position, check whether the new parameter position exceeds the set parameter boundary. If it exceeds, use the random reset strategy to adjust the parameter position. The adjustment formula is: in, represents the new position of the i-th individual in the j-th dimension obtained by the white whale algorithm, T represents the current number of iterations, It represents the position after random reset strategy adjustment, lb and ub represent the lower and upper bounds of the parameter respectively, rand() is a function that generates random numbers between the interval [0,1], and d is an adjustment factor.

6. The method according to claim 5, characterized in that In step S4, when optimizing the model parameters, the loss function is defined as the root mean square error between the actual terminal voltage of the battery and the terminal voltage estimated by the model, and the calculation formula is: Where n is the number of data points, U t is the real terminal voltage, U est Estimate the terminal voltages for the model.

7. The method according to claim 1, characterized in that The step S5 comprises the following steps: S5.

1. According to the first-order RC equivalent circuit model, U ocv The linear model is used to solve the open circuit voltage curves of all segments in A; S5.

2. Solve the open circuit voltage curves of all segments in B. The calculation formula is: U ocv =U t -I(R o +R p ) Where I is the current value, which is positive during charging. If the maximum temperature difference during charging does not exceed 10°C, then R o and R p Take the average of all fragment identification results in A. If the maximum temperature difference during charging exceeds 10°C, use a linear model to fit the temperature and R o and R p The relationship between R and o and R p The value of .

8. The method according to claim 1, characterized in that The step S6 comprises the following steps: S6.

1. Connect the open circuit voltage curves corresponding to all the segments in A and B according to their positional relationship, and use a Savitzky-Golay filter to smooth the combined open circuit voltage curve; S6.

2. Obtain a curve of the open circuit voltage with respect to the state of charge according to the state of charge data recorded by the battery management system; S6.

3. Use a p-order polynomial model to fit the relationship between open circuit voltage and state of charge, where p is at least 6th order.

Images