A method for estimating the state of health of lithium-ion batteries by integrating multi-dimensional characteristic parameters of the constant voltage charging stage.

Abstract

The application discloses a lithium ion battery health state estimation method fusing constant voltage charging stage multi-dimensional feature parameters, comprising: offline determination of battery model specific structure; calculation and extraction of current curve explicit candidate feature parameters and implicit candidate feature parameters corresponding to different equivalent cutoff currents under battery constant voltage working conditions; spirlman correlation analysis is carried out to determine the feature parameter set representing battery aging and the corresponding effective current interval; the battery health state estimation model corresponding to different equivalent constant voltage cutoff currents is constructed and stored; the battery constant voltage charging time-current sequence is recorded and stored, and the explicit feature parameters are calculated in real time; the battery health state estimation model corresponding to the actual constant voltage charging cutoff current value is obtained through inquiry; and the actual health state of the battery is estimated. Compared with the battery health state estimation method based on dynamic discharge data and constant current charging data, the estimation method provided by the application has higher robustness.

Description

Technical Field

[0001] This invention belongs to the field of battery technology, and in particular relates to a method for estimating the health status of lithium-ion batteries by integrating multi-dimensional characteristic parameters of the constant voltage charging stage. Background Technology

[0002] Lithium-ion batteries have been widely used in electric vehicle power systems in recent years due to their advantages such as high energy density, low self-discharge rate, and environmental friendliness. However, as the driving range of electric vehicles increases, the state-of-health (SOH) of lithium-ion batteries gradually declines, thereby reducing vehicle performance and increasing safety risks. Therefore, accurately estimating the battery's SOH is crucial for the efficient and safe operation of vehicles.

[0003] Compared to discharge conditions, battery charging conditions provide more stable data. Therefore, battery SoH estimation methods based on charging condition data have received widespread attention. However, for constant current charging, the randomness of the initial charging state can affect the accuracy of SoH estimation based on this data. In contrast, constant voltage charging is less sensitive to the initial charging state and contains more information reflecting battery aging, leading to its increasing application in SoH estimation in recent years. However, existing SoH estimation methods based on constant voltage characteristics primarily rely on complete constant voltage charging data and use parameters from traditional RC network equivalent circuit models or dominant charging current parameters for SoH estimation. They lack comprehensive analysis and application of high-precision battery models and multi-dimensional battery characteristic parameters under incomplete constant voltage charging conditions. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides a method for estimating the health status of lithium-ion batteries by integrating multi-dimensional feature parameters from the constant-voltage charging stage. This method accurately estimates the battery's health status online by integrating explicit and implicit feature parameters extracted from the constant-voltage charging data after the battery's constant-voltage charging condition has ended.

[0005] The present invention is achieved through the following technical solution.

[0006] This invention provides a method for estimating the state of health of a lithium-ion battery by integrating multi-dimensional characteristic parameters of the constant voltage charging stage. The method includes the following steps:

[0007] S1. Based on the battery open-circuit voltage characteristics under constant voltage conditions and the general model structure, determine the specific structure of the battery model.

[0008] S2, based on the constant voltage operating condition data of the battery throughout its entire life cycle, calculate and extract the explicit candidate characteristic parameters and implicit candidate characteristic parameters of the current curves corresponding to different equivalent cutoff currents under the constant voltage operating condition of the battery.

[0009] S3. Perform Spearman correlation analysis on the dominant and latent characteristic parameters obtained in step S2 to determine the set of characteristic parameters characterizing battery aging and the corresponding effective current range.

[0010] S4 combines intelligent optimization algorithms and machine learning algorithms to construct and store battery health state estimation models corresponding to different equivalent constant voltage cutoff currents.

[0011] S5. After the battery enters the constant voltage charging mode, record and store the battery constant voltage charging time-current sequence, and calculate the dominant characteristic parameters determined in step S3 in real time.

[0012] S6. When the charging process is over, record the actual cut-off current value corresponding to the end of the actual constant voltage charging, and calculate and extract the corresponding effective implicit feature parameters according to the effective current range in which it is located. Based on the explicit feature parameters calculated in step S5, select and extract the explicit feature parameters according to the effective current range to form a feature parameter set.

[0013] S7. Based on the relationship between the actual constant voltage charging cutoff current value and the equivalent cutoff current of constant voltage charging stored in step S4, the battery health status estimation model corresponding to the actual constant voltage charging cutoff current value is obtained by querying.

[0014] S8. Substitute the feature variables obtained in step S6 into the battery actual health state estimation model corresponding to the actual constant voltage charging cutoff current obtained in step S7 to estimate the actual health state of the battery.

[0015] In step S1, the general battery model structure applied to constant voltage conditions includes multiple open-circuit voltage equivalent voltage sources V connected in parallel and in series. eoc,k Resistance R k and inductor L k Multiple open-circuit voltage equivalent voltage sources V connected in parallel and series are used. eoc,k Resistance R k and inductor L k The input terminal is connected to the positive terminal of the voltage output terminal of the model, and multiple open-circuit voltage equivalent voltage sources V are connected in parallel and in series. eoc,k Resistance R k and inductor L k The output terminal is connected to the negative terminal of the model's voltage output terminal. The analytical mathematical expression for the general battery model applied under constant voltage conditions is:

[0016]

[0017] Where n is the open-circuit voltage equivalent voltage source V connected in parallel and series. eoc,k Resistance R kand inductor L k The number of branches, i.e., the order of the equivalent circuit model; I k (t) represents the equivalent voltage V flowing through the series-connected open-circuit voltage source at time t. eoc,k Resistance R k and inductor L k The current; I k (0) is the equivalent voltage source V flowing through the series-connected open-circuit voltage. eoc,k Resistance R k and inductor L k The initial current; the k-th time constant τ k =L k / R k .

[0018] In step S1, the steps for determining the battery model structure are as follows:

[0019] S1.1 Extract the OCV-SoC relationship curve for the SoC range corresponding to the constant voltage stage;

[0020] S1.2, perform piecewise linear approximation on the extracted OCV-SoC relationship curve;

[0021] S1.3, determine the order of the battery model based on the number of sub-function segments contained in the piecewise function.

[0022] In step S2, the method for extracting constant voltage operating condition datasets corresponding to different equivalent cutoff currents is as follows: Define the equivalent cutoff current dataset as I. cut,eq,n =I cut +n*ΔI, where I cut A cutoff current is set for the battery under constant voltage conditions, where ΔI is the step size of the equivalent cutoff current change, n is the number of intervals, and the relationship n < [I(0) - I cut ] / ΔI; Extract data segments (t, I(t)) starting from the current and corresponding time at the beginning of constant voltage charging and ending at each equivalent cutoff current and corresponding time, where I cut,eq,n <I(t)<I(0)。

[0023] In step S2, the dominant candidate characteristic parameters of the current curve under constant voltage conditions mainly include: the constant voltage charging time T of the battery. CV The current slope k at the end of constant voltage charging I and constant voltage stage charging capacity Q CV The expression for the current slope at the end of constant voltage charging is:

[0024]

[0025] Where Δt represents the time interval, satisfying the relationship Δt=m*T s Ts The sampling period is represented by m, where m is a positive integer. The expression for the charging capacity during the constant voltage stage is:

[0026]

[0027] In step S2, the latent candidate characteristic parameters under constant voltage conditions mainly include: battery model parameters, the difference characterization coefficient between the initial current curve and the cycle current curve, and the battery capacity-current difference value during the constant voltage stage; wherein, the battery model parameter set is [R k ,I k (0),τ k The set of coefficients characterizing the difference between the initial current curve and the circulating current curve is [FoI]. SSE FoI MAE FoI MAPE FoI RMSE The specific expression is:

[0028]

[0029] Among them, I ini and I cyc These represent the initial and cyclic current curves, respectively, with N representing the data length of the initial current curve; the expression for the battery capacity current difference value during the constant voltage stage is:

[0030]

[0031] In step S3, the expression for the correlation coefficient in Spearman's correlation analysis is:

[0032]

[0033] Where, ρ FoI,SoH (I cut,eq,n ) represents the corresponding equivalent cutoff current I cut,eq,n Spearman correlation coefficient between characteristic parameters and battery health state, n cycle d represents the number of battery RPT tests in offline testing. i The rank difference between the characteristic parameters and the observed values ​​corresponding to the battery health status;

[0034] In step S3, extract the values ​​that satisfy |ρ FoI,SoH (I cut,eq,n The characteristic parameters and corresponding equivalent cutoff currents corresponding to the condition )|>0.8 are used to determine the characteristic parameters and corresponding equivalent cutoff currents that satisfy |ρ FoI,SoH (I cut,eq,n The equivalent cutoff current under the condition > 0.8 constitutes the effective current range [I]. cut,eq,min ,I cut,eq,maxUltimately, this forms a set of characteristic variables representing the battery's SoH and the corresponding effective current range.

[0035] In step S4, a battery health state estimation model corresponding to different equivalent cutoff currents is constructed using the support vector regression algorithm, and the parameters of the support vector regression algorithm are optimized using the gray wolf optimization algorithm to obtain a more accurate battery health state estimation model corresponding to different equivalent cutoff currents. The method for constructing the battery health state estimation model combining the gray wolf optimization algorithm and the support vector regression algorithm is as follows:

[0036] S4.1, Algorithm parameter initialization: Gray wolf population size N GW Maximum number of iterations T GW ;

[0037] S4.2, Population Initialization: Randomly generate N GW Individual gray wolf x j j = 1, 2, ..., N GW The model is trained using the support vector regression algorithm, and the fitness value of each individual is calculated.

[0038] S4.3 Calculate and record the position vectors x of the leader wolf σ, the second-in-command wolf β, and the advisor wolf δ. σ x β and x δ ;

[0039] S4.4, Update the position of gray wolves in the gray wolf pack;

[0040] S4.5, determine whether the maximum number of iterations T set in S4.1 has been reached. GW Or whether the termination condition is met; if T is not reached. GW If the termination condition is not met, proceed to step S4.2; otherwise, proceed to step S4.6.

[0041] S4.6 outputs the optimal gray wolf position and obtains the optimal parameters for the support vector regression algorithm.

[0042] In step S7, the actual constant voltage charging cutoff current value I cut,eq,act The process for determining and querying the constant voltage charging equivalent cutoff current relationship stored in step S4 is as follows: Obtain I cut,eq,act , when I cut,eq,act Greater than or equal to I cut,eq,max At that time, according to I cut,eq,max Select the corresponding battery health state estimation model; otherwise, when I cut,eq,act Less than or equal to I cut,eq,min At that time, according to I cut,eq,min Select the corresponding battery health state estimation model; otherwise, i.e., when I cut,eq,act Greater than I cut,eq,min And when Icut,eq,act Less than I cut,eq,max At that time, according to I cut,eq,act Select the appropriate battery health status estimation model.

[0043] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0044] 1. This invention comprehensively considers the explicit and implicit characteristic parameters of the battery under constant voltage conditions. The fusion of the above-mentioned multi-dimensional characteristic parameters can more comprehensively and accurately characterize the degradation of the battery's health status.

[0045] 2. Compared with the battery health state estimation method based on data length under incomplete constant voltage conditions, the battery health state estimation method based on equivalent cutoff current provided by this invention is not sensitive to the preset cutoff conditions of constant voltage conditions, and has higher flexibility and robustness.

[0046] 3. The equivalent circuit model and structure determination method of battery resistor-inductor network applied to constant voltage conditions constructed in this invention establishes the relationship between the current characteristics and the open-circuit voltage characteristics of the battery during constant voltage charging. While ensuring the accuracy of the model, it has low complexity and a certain degree of interpretability.

[0047] 4. Compared with battery health status estimation methods based on dynamic discharge data and constant current charging data, the estimation method provided by this invention has higher robustness. Attached Figure Description

[0048] Figure 1 This is a flowchart of the battery health state estimation algorithm of the present invention.

[0049] Figure 2 This is a structural diagram of the equivalent circuit model of the battery multi-stage resistor-inductor network applied to constant voltage conditions according to the present invention.

[0050] Figure 3 This is the open-circuit voltage-state-of-charge relationship curve for the constant voltage stage of the ternary lithium-ion battery of this invention.

[0051] Figure 4 This is a comparison and verification diagram of the battery model of the present invention.

[0052] Figure 5 A comparison chart showing the evolution of the root mean square error in current estimation for the first-order and second-order models with different cycle numbers.

[0053] Figure 6 The graph shows the evolution trend of the battery model parameter τ1 identified under different health conditions with respect to the constant voltage charging cutoff current.

[0054] Figure 7The graph shows the evolution trend of the battery model parameter τ2 identified under different health conditions with respect to the constant voltage charging cutoff current.

[0055] Figure 8 Compare the estimated battery health status with the actual measured value for different constant voltage charging cutoff current values. Detailed Implementation

[0056] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0057] This invention provides a method for estimating the health status of lithium-ion batteries by integrating multi-dimensional characteristic parameters of the constant voltage charging stage, thereby enabling efficient and safe management of on-board power batteries and power storage systems.

[0058] A method for estimating the state of health of lithium-ion batteries by integrating multi-dimensional characteristic parameters of the constant voltage charging stage is as follows: Figure 1 As shown, this method is mainly divided into two parts: the first part is the offline model construction part, and the second part is the online state estimation part; the two parts will be further explained below.

[0059] The offline model construction part includes the following steps:

[0060] 1) Determine the specific structure of the battery model based on the battery open-circuit voltage characteristics and general model structure under constant voltage conditions;

[0061] Battery model structure as follows Figure 2 As shown, the general battery model structure applied to constant voltage conditions includes multiple open-circuit voltage equivalent voltage sources V connected in parallel and in series. eoc,k Resistance R k and inductor L k Multiple open-circuit voltage equivalent voltage sources V connected in parallel and series are used. eoc,k Resistance R k and inductor L k The input terminal is connected to the positive terminal of the voltage output terminal of the model, and multiple open-circuit voltage equivalent voltage sources V are connected in parallel and in series. eoc,k Resistance R k and inductor L k The output terminal is connected to the negative terminal of the model's voltage output terminal. The analytical mathematical expression for the general battery model applied under constant voltage conditions is:

[0062]

[0063] Where n is the open-circuit voltage equivalent voltage source V connected in parallel and series. eoc,k Resistance R k and inductor L k The number of branches, i.e., the order of the equivalent circuit model; I k (t) represents the equivalent voltage V flowing through the series-connected open-circuit voltage source at time t. eoc,k Resistance R k and inductor L k The current; I k (0) is the equivalent voltage source V flowing through the series-connected open-circuit voltage. eoc,k Resistance R k and inductor L k The initial current; the k-th time constant τ k =L k / R k ;

[0064] The steps for determining the battery model structure are as follows:

[0065] 1.1) Extract the OCV-SoC relationship curve for the SoC range corresponding to the constant voltage stage;

[0066] 1.2) The extracted OCV-SoC relationship curve is approximated by piecewise linear approximation;

[0067] 1.3) Determine the order of the battery model based on the number of sub-function segments contained in the piecewise function;

[0068] 2) Based on the constant voltage operating condition data of the battery throughout its entire life cycle, calculate and extract the explicit and implicit candidate characteristic parameters of the current curves corresponding to different equivalent cutoff currents under the constant voltage operating condition of the battery.

[0069] The method for extracting constant voltage operating condition datasets corresponding to different equivalent cutoff currents is as follows: Define the equivalent cutoff current dataset as I. cut,eq,n =I cut +n*ΔI, where I cut A cutoff current is set for the battery under constant voltage conditions, where ΔI is the step size of the equivalent cutoff current change, n is the number of intervals, and the relationship n < [I(0) - I cut ] / ΔI; Extract data segments (t, I(t)) starting from the current and corresponding time at the beginning of constant voltage charging and ending at each equivalent cutoff current and corresponding time, where I cut,eq,n <I(t)<I(0);

[0070] The dominant candidate characteristic parameters of the current curve under constant voltage conditions mainly include: the constant voltage charging time T of the battery. CV The current slope k at the end of constant voltage charging I and constant voltage stage charging capacity QCV The expression for the current slope at the end of constant voltage charging is:

[0071]

[0072] Where Δt represents the time interval, satisfying the relationship Δt=m*T s T s The sampling period is represented by m, where m is a positive integer. The expression for the charging capacity during the constant voltage stage is:

[0073]

[0074] The latent candidate characteristic parameters of the battery under constant voltage conditions mainly include: battery model parameters, the characterization coefficient of the difference between the initial current curve and the cycle current curve, and the battery capacity-current difference value during the constant voltage stage; among them, the battery model parameter set is [R k ,I k (0),τ k The model parameters are identified using an intelligent optimization algorithm; the set of coefficients representing the differences between the initial current curve and the circulating current curve is [FoI]. SSE FoI MAE FoI MAPE FoI RMSE The specific expression is:

[0075]

[0076] Among them, I ini and I cyc These represent the initial and cyclic current curves, respectively, with N representing the data length of the initial current curve; the expression for the battery capacity current difference value during the constant voltage stage is:

[0077]

[0078] The intelligent optimization algorithm includes parameter identification algorithms such as nonlinear least squares method and genetic algorithm.

[0079] 3) Perform Spearman correlation analysis on the dominant and latent characteristic parameters obtained in step 2) to determine the set of characteristic parameters characterizing battery aging and the corresponding effective current range.

[0080] The expression for the correlation coefficient in Spearman's correlation analysis is as follows:

[0081]

[0082] Where, ρ FoI,SoH (I cut,eq,n ) represents the corresponding equivalent cutoff current I cut,eq,n Spearman correlation coefficient between characteristic parameters and battery health state, ncycle d represents the number of battery RPT tests in offline testing. i The rank difference between the characteristic parameters and the observed values ​​corresponding to the battery health status;

[0083] In step 3), extract the values ​​that satisfy |ρ FoI,SoH (I cut,eq,n The characteristic parameters and corresponding equivalent cutoff currents corresponding to the condition )|>0.8 are used to determine the characteristic parameters and corresponding equivalent cutoff currents that satisfy |ρ FoI,SoH (I cut,eq,n The equivalent cutoff current under the condition > 0.8 constitutes the effective current range [I]. cut,eq,min ,I cut,eq,max Ultimately, this forms a set of characteristic variables representing the battery's SoH and the corresponding effective current range;

[0084] 4) Combining intelligent optimization algorithms and machine learning algorithms, construct and store battery health state estimation models corresponding to different equivalent constant voltage cutoff currents;

[0085] The intelligent optimization algorithm uses the Grey Wolf Optimization Algorithm, and the machine learning algorithm uses the Support Vector Regression Algorithm. These algorithms can be directly implemented using the corresponding algorithm modules in MATLAB. Specifically, the method for constructing the battery health state estimation model combining the Grey Wolf Optimization Algorithm and Support Vector Regression Algorithm is as follows:

[0086] S4.1, Algorithm parameter initialization: Gray wolf population size N GW Maximum number of iterations T GW ;

[0087] S4.2, Population Initialization: Randomly generate N GW Individual gray wolf x j j = 1, 2, ..., N GW The model is trained using the support vector regression algorithm, and the fitness value of each individual is calculated.

[0088] S4.3 Calculate and record the position vectors x of the leader wolf σ, the second-in-command wolf β, and the advisor wolf δ. σ x β and x δ ;

[0089] S4.4, Update the position of gray wolves in the gray wolf pack;

[0090] S4.5, determine whether the maximum number of iterations T set in S4.1 has been reached. GW Or whether the termination condition is met; if T is not reached. GW If the termination condition is not met, proceed to step S4.2; otherwise, proceed to step S4.6.

[0091] S4.6 outputs the optimal gray wolf position and obtains the optimal parameters for the support vector regression algorithm.

[0092] The online status estimation section includes the following steps:

[0093] 1) After the battery enters the constant voltage charging mode, record and store the battery constant voltage charging time-current sequence, and calculate the explicit characteristic parameters determined in step 3) of the offline model construction part in real time.

[0094] 2) After the charging process is completed, record the actual cut-off current value corresponding to the end of the actual constant voltage charging, and calculate and extract the corresponding effective implicit feature parameters according to the effective current range in which it is located. Based on the explicit feature parameters calculated in step 1) of the online state estimation part, select and extract the effective explicit feature parameters according to the effective current range to form a feature parameter set.

[0095] 3) Based on the actual constant voltage charging cutoff current value and the relationship between the constant voltage charging equivalent cutoff current stored in step 4) of the offline model construction section, query the battery health status estimation model corresponding to the actual constant voltage charging cutoff current value.

[0096] Among them, the actual constant voltage charging cutoff current value I cut,eq,act The process for determining and querying the constant voltage charging equivalent cutoff current relationship stored in step 4) of the offline model construction section is as follows: Obtain I cut,eq,act , when I cut,eq,act Greater than or equal to I cut,eq,max At that time, according to I cut,eq,max Select the corresponding battery health state estimation model; otherwise, when I cut,eq,act Less than or equal to I cut,eq,min At that time, according to I cut,eq,min Select the corresponding battery health state estimation model; otherwise, i.e., when I cut,eq,act Greater than I cut,eq,min And when I cut,eq,act Less than I cut,eq,max At that time, according to I cut,eq,act Select the appropriate battery health status estimation model;

[0097] 4) Substitute the feature variables obtained in step 2) of the online state estimation section into the battery actual health state estimation model corresponding to the actual constant voltage charging cutoff current obtained in step 3) of the online state estimation section to estimate the actual health state of the battery.

[0098] In this embodiment, the application is a ternary lithium-ion battery with a nominal capacity of 3.4Ah. However, it is not limited to this in practical applications. The open-circuit voltage-state-of-charge relationship curve for the constant-voltage stage of a ternary lithium-ion battery is shown below. Figure 3As shown, the relationship curve can be represented as two linear functions with slopes of tanθ1 and tanθ2, respectively. Therefore, for the battery being implemented, the model order is set to 2. The comparison and verification diagrams of different structural models and the evolution of the root mean square error of current estimation are shown below. Figure 4 and Figure 5 As shown in the figure, compared with the first-order model, the current estimation accuracy of the second-order model is significantly improved across different cycle numbers. The evolution trends of the battery model parameters τ1 and τ2 identified under different health states with respect to the constant-voltage charging cutoff current are as follows: Figure 6 and Figure 7 As shown, τ1 and τ2 have a one-to-one correspondence with the constant voltage charging cutoff current value under different health states. The estimated battery health state values ​​corresponding to different constant voltage charging cutoff current values ​​obtained using this invention are compared with the actual measured values. Figure 8 As shown, the estimated battery health status can be well tracked by the measured values. Therefore, this method can effectively achieve online battery health status estimation.

[0099] In summary, the present invention provides a method for estimating the health status of lithium-ion batteries by integrating multi-dimensional characteristic parameters of the constant-voltage charging stage, comprising: offline determination of the specific structure of the battery model; calculation and extraction of explicit and implicit candidate characteristic parameters of the current curves corresponding to different equivalent cutoff currents under constant-voltage conditions; Spearman correlation analysis to determine the set of characteristic parameters characterizing battery aging and the corresponding effective current range; construction and storage of battery health status estimation models corresponding to different equivalent constant-voltage cutoff currents; recording and storing the battery constant-voltage charging time-current sequence and calculating explicit characteristic parameters in real time; recording the actual cutoff current value corresponding to the end of actual constant-voltage charging, calculating and extracting the corresponding effective implicit characteristic parameters, filtering and extracting effective explicit characteristic parameters to form a set of characteristic parameters; querying the battery health status estimation model corresponding to the actual constant-voltage charging cutoff current value; and estimating the actual health status of the battery. The estimation method proposed in this invention has the following four advantages: (1) This invention comprehensively considers the explicit and implicit characteristic parameters of the battery under constant voltage conditions. The fusion of the above multi-dimensional characteristic parameters can more comprehensively and accurately characterize the decay of the battery health status; (2) Compared with the battery health status estimation method based on the incomplete constant voltage conditions based on data length, the battery health status estimation method based on the equivalent cutoff current provided by this invention is not sensitive to the preset constant voltage cutoff conditions and has higher flexibility and robustness; (3) The battery resistor-inductor network equivalent circuit model and structure determination method applied to constant voltage conditions constructed in this invention establishes the relationship between the current characteristics of the battery in the constant voltage charging stage and the open circuit voltage characteristics of the battery. While ensuring the accuracy of the model, it has lower complexity and a certain degree of interpretability; (4) Compared with the battery health status estimation method based on dynamic discharge data and constant current charging data, the estimation method provided by this invention has higher robustness.

[0100] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0101] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0102] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for estimating the state of health of a lithium-ion battery by fusing multi-dimensional characteristic parameters in the constant-voltage charging phase, characterized in that, The method comprises the following steps: S1, according to the open-circuit voltage characteristics of the battery in the constant voltage condition and the general model structure, the specific structure of the battery model is determined; S2, according to the battery constant voltage condition data in the whole life cycle, the current curve explicit candidate feature parameters and implicit candidate feature parameters corresponding to different equivalent cutoff currents under the battery constant voltage condition are calculated and extracted; S3, the explicit and implicit feature parameters obtained in step S2 are subjected to Spearman correlation analysis, and a feature parameter set representing the battery aging and a corresponding effective current interval are determined; S4, combining intelligent optimization algorithm and machine learning algorithm, a battery state of health estimation model corresponding to different equivalent constant voltage cutoff currents is constructed and stored; S5, when the battery enters the constant voltage charging mode, the constant voltage charging time-current sequence of the battery is recorded and stored, and the explicit feature parameters determined in step S3 are calculated in real time; S6, when the charging process is completed, the actual cutoff current value corresponding to the actual constant voltage charging end is recorded, and the corresponding effective implicit feature parameters are calculated and extracted according to the effective current interval, and the effective explicit feature parameters are extracted according to the effective current interval, and the feature parameter set is composed; S7, according to the relationship between the actual constant voltage charging cutoff current value and the equivalent constant voltage charging cutoff current stored in step S4, the battery state of health estimation model corresponding to the actual constant voltage charging cutoff current value is obtained; S8, the feature variables obtained in step S6 are substituted into the battery actual state of health estimation model corresponding to the actual constant voltage charging cutoff current obtained in step S7, and the actual state of health of the battery is estimated.

2. A method according to claim 1, characterized in that In step S1, the general battery model structure applied to constant voltage conditions includes multiple open-circuit voltage equivalent voltage sources V connected in parallel and in series. eoc,k Resistance R k and inductor L k Multiple open-circuit voltage equivalent voltage sources V connected in parallel and series are used. eoc,k Resistance R k and inductor L k The input terminal is connected to the positive terminal of the voltage output terminal of the model, and multiple open-circuit voltage equivalent voltage sources V are connected in parallel and in series. eoc,k Resistance R k and inductor L k The output terminal is connected to the negative terminal of the model's voltage output terminal. The analytical mathematical expression for the general battery model applied under constant voltage conditions is: where V t and I(t) are the battery terminal voltage and the charging current at time t, respectively; n is the number of branches of the series connection of open-circuit voltage equivalent voltage sources V eoc,k , resistors R k , and inductors L k , i.e., the order of the equivalent circuit model; I k (t) is the current at time t through the series connection of open-circuit voltage equivalent voltage sources V eoc,k , resistors R k , and inductors L k ; I k (0) is the initial current through the series connection of open-circuit voltage equivalent voltage sources V eoc,k , resistors R k , and inductors L k ; and the kth time constant τ k = L k / R k .

3. A method according to claim 1, characterized in that In step S1, the determination step of the battery model structure is: S1.1, extracting the OCV-SoC relationship curve corresponding to the SoC range in the constant voltage stage; S1.2, the extracted OCV-SoC relationship curve is subjected to piecewise linear approximation; S1.3, the order of the battery model is determined according to the number of sub-functions contained in the piecewise function.

4. The method of claim 1, wherein, The constant voltage working condition data set corresponding to different equivalent cut-off currents in the step S2 is extracted by defining an equivalent cut-off current data set as I cut,eq,n =I cut +n*ΔI, wherein I cut is a cut-off current of the battery in the constant voltage working condition, ΔI is a change step of the equivalent cut-off current, n is an interval number, and the relationship n<[I(0)-I cut ] / ΔI is satisfied; and a data segment (t, I(t)) is extracted, wherein a current at a constant voltage charging starting moment and a corresponding time are taken as a starting point, and each equivalent cut-off current and a corresponding time are taken as an ending point, and I cut,eq,n <I(t)<I(0).

5. The method of claim 1, wherein, The current curve dominant candidate feature parameters in the step S2 under the constant voltage condition of the battery mainly include: the constant voltage charging time T of the battery CV , the current slope k at the end of constant voltage charging I , and the constant voltage charging capacity Q CV ; wherein, the current slope at the end of constant voltage charging is expressed as: where I is the battery charging current, Δt represents a time interval, and satisfies the relationship Δt = m*T s , T s represents a sampling period, and m is a positive integer; and the charging capacity expression in the constant-voltage phase is 。 6. The method of claim 1, wherein, The implicit candidate feature parameters in the constant voltage condition of the battery mainly include: battery model parameters, initial current curve and cycle current curve difference representation coefficients, and constant voltage stage battery capacity current difference values; wherein the battery model parameter set is [R k ,I k (0),τ k ]; the initial current curve and cycle current curve difference representation coefficient set is [FoI SSE ,FoI MAE ,FoI MAPE ,FoI RMSE ], and the specific expression is: wherein FoI SSE , FoI RMSE , FoI MAE and FoI MAPE represent characteristic parameters based on residual sum of squares, root mean square error, mean absolute error and mean absolute percentage error, respectively; ini and I cyc represent initial and cyclic current curves, respectively, and N represents initial current curve data length; and the expression for the constant voltage stage battery capacity current difference value is: where Q CV and T s represent the battery charge capacity at constant voltage phase and the sampling period, respectively.

7. The method of claim 1, wherein, In step S3, the expression of the correlation coefficient in the Spearman correlation analysis is: wherein, ρ FoI,SoH (I cut,eq,n ) is the characteristic parameter corresponding to the equivalent cut-off current I cut,eq,n Spearman correlation coefficient between the characteristic parameter and the battery state of health, n cycle is the number of RPT tests in offline testing, d i is the difference in rank of the corresponding observation values of the characteristic parameter and the battery state of health. The characteristic parameters satisfying the condition of |p FoI,SoH (I cut,eq,n )|>0.8 and the corresponding equivalent cutoff current are extracted in step S3. FoI,SoH (I cut,eq,n )|>0.8 corresponding to the same characteristic parameter form an effective current interval [I cut,eq,min ,I cut,eq,max ], and finally a characteristic variable set representing the battery SoH and the corresponding effective current interval are formed.

8. The method of claim 1, wherein, In step S4, the support vector regression algorithm is used to construct the battery state of health estimation model corresponding to different equivalent cutoff currents, and the grey wolf optimization algorithm is used to optimize the support vector regression algorithm parameters, so that a more accurate battery state of health estimation model corresponding to different equivalent cutoff currents is obtained.

9. The method of claim 1, wherein , the step S7, the actual constant voltage charging cutoff current value I cut,eq,act and the judgment and query process of the constant voltage charging equivalent cutoff current relationship stored in step S4 are: obtaining I cut,eq,act When I cut,eq,act is greater than or equal to I cut,eq,max , the corresponding battery health state estimation model is selected according to I cut,eq,max ; otherwise when I cut,eq,act is less than or equal to I cut,eq,min , a corresponding battery state of health estimation model is selected according to I cut,eq,min ; otherwise, i.e., when I cut,eq,act is greater than I cut,eq,min and when I cut,eq,act is less than I cut,eq,max , a corresponding battery state of health estimation model is selected according to I cut,eq,act .

10. A method according to claim 8, wherein, The battery state of health estimation model construction method combining the grey wolf optimization algorithm and the support vector regression algorithm is: S4.1, algorithm parameter initialization: gray wolf population size N GW , maximum iteration number T GW ; S4.2, population initialization: randomly generate N GW gray wolf individuals x j , j = 1, 2,..., N GW ; train the model by support vector regression algorithm, calculate the fitness value of each individual; S4.3, calculate and record the position vectors x of the alpha wolf, beta wolf and delta wolf σ , x β and x δ ; S4.4, updating the position of the grey wolf in the grey wolf group; S4.5, determine whether the maximum number of iterations T set in S4.1 has been reached. GW Or whether the termination condition is met; if T is not reached. GW And since the termination condition is not met, proceed to step S4.2; Otherwise, step S4.6 is executed; S4.6, output the optimal grey wolf position and obtain the optimal parameters of the support vector regression algorithm.

Images