CN115524625A - Lithium battery overcharge and overdischarge detection method based on characteristic impedance - Google Patents

Abstract

The present disclosure discloses a lithium battery overcharge and overdischarge detection method based on characteristic impedance, which includes: s100: acquiring an electrochemical impedance spectrum of a lithium battery to be detected; s200: extracting the characteristic impedance of the lithium battery to be tested based on the electrochemical impedance spectrum; s300: and inputting the characteristic impedance into an over-charge and over-discharge detection model to realize over-charge and over-discharge detection of the lithium battery to be detected. The method has the advantages that the information is represented by using the electrochemical impedance spectroscopy, the representation is comprehensive, the battery does not need to be disassembled, the electrochemical impedance spectroscopy is sensitive to the change of the battery state and the change of the internal structure, and the battery state can be effectively represented.

Description

Lithium battery overcharge and overdischarge detection method based on characteristic impedance

Technical Field

The disclosure belongs to the field of battery detection, and particularly relates to a lithium battery overcharge and overdischarge detection method based on characteristic impedance.

Background

With the rapid development of high-efficiency utilization of clean energy, lithium batteries are widely applied to various industries such as electric vehicles and electrochemical energy storage. However, due to poor battery consistency and imperfect current balancing strategy, the single battery with poor performance in the battery module has overcharge and overdischarge faults, so that the health state of the battery is accelerated to attenuate, the temperature rise is increased, and potential safety hazards are brought to the application of the battery module. And the overcharge and overdischarge faults can cause micro short circuit, structural phase change and the like in the battery, and the broadband alternating current internal resistance can effectively represent the electrochemical reaction and the structural phase change process in the battery. Therefore, it is very important for the safety application of the lithium battery to analyze the impedance characteristics of overcharge and overdischarge faults in the charging and discharging processes of the lithium battery and research on a fault detection method.

The existing method for detecting overcharge and overdischarge of the battery mainly comprises a structural form-based detection method and an electrochemical performance detection method, wherein the structural form-based detection method is characterized by using microscopic observation means such as SEM, XRD, TEM and the like to realize the characterization of the battery characteristics, and the method can accurately and cleanly react the internal structure change of the battery, but has a narrow observation range and needs to disassemble the battery. The electrochemical performance detection mainly aims at detecting the impedance, voltage, capacity, gas and the like of the battery, and mainly utilizes the macroscopic characteristic quantity of the battery to carry out analysis and detection, so that the method is easy to realize and does not need to disassemble the battery, but the method cannot clearly represent the weak change and the reaction lag in the battery.

Disclosure of Invention

In view of the defects in the prior art, the present disclosure provides a method for detecting overcharge and overdischarge of a lithium battery based on characteristic impedance, which uses an electrochemical impedance spectrum to represent charge and discharge information of the lithium battery, and not only has comprehensive representation without disassembling the battery, but also is sensitive to changes in the battery state and changes in the internal structure, and can effectively represent the battery state.

In order to achieve the above object, the present disclosure provides the following technical solutions:

a lithium battery overcharge and overdischarge detection method based on characteristic impedance comprises the following steps:

s100: acquiring an electrochemical impedance spectrum of a lithium battery to be detected;

s200: extracting the characteristic impedance of the lithium battery to be detected based on the electrochemical impedance spectrum;

s300: and inputting the characteristic impedance into an over-charge and over-discharge detection model to realize over-charge and over-discharge detection of the lithium battery to be detected.

Preferably, the characteristic impedance includes ohmic internal resistance, SEI film internal resistance, charge transfer resistance, and low frequency diffusion resistance.

Preferably, in step S300, the overcharge-overdischarge detection model is a support vector machine detection model based on particle swarm optimization.

Preferably, the construction process of the support vector machine detection model based on particle swarm optimization is as follows:

s301: inputting characteristic quantity impedance values of the lithium battery in different circulation states;

s302: selecting a Gaussian function as a kernel function;

s303: initializing a particle swarm, and setting the population scale of the particles, penalty values, maximum iteration times and initial parameter values of a kernel function to establish a support vector machine detection model;

s304: and calculating the fitness function of each particle, judging whether the fitness function of the particle meets the precision requirement when the iteration times reach the maximum, if so, obtaining the optimal support vector machine detection model, otherwise, updating the position and the speed of the particle, returning to the step S303, and reconstructing a new support vector machine diagnosis model.

Preferably, in step S303, the particle position and velocity are set by the following formula:

v _t+1 ＝wv _t +r ₁ ·rand()·(P _t -x _t )+r ₂ ·rand()·(G _t -x _t )

x _t+1 ＝x _t +v _t

wherein x is _t Is the position of the particle at the current time, v _t Is the velocity of the particle at the current time, w is the inertial weight, r ₁ And r ₂ Is constant, rand () is a generated random number, P _t 、G _t The current time t is the self optimal position and the global optimal position of the particle.

The present disclosure also provides a lithium battery overcharge and overdischarge detection apparatus based on characteristic impedance, including:

the acquisition unit is used for acquiring an electrochemical impedance spectrum of the lithium battery to be tested;

the extraction unit is used for extracting the characteristic impedance of the lithium battery to be detected based on the electrochemical impedance spectrum;

and the detection unit is used for inputting the characteristic impedance into the over-charge and over-discharge detection model so as to realize over-charge and over-discharge detection of the lithium battery to be detected.

Compared with the prior art, the beneficial effect that this disclosure brought does:

1. according to the method, the charging and discharging information of the lithium battery is represented by utilizing the electrochemical impedance spectrum, so that the representation is comprehensive, the battery does not need to be disassembled, and meanwhile, the battery state is sensitive to the change of the battery state and the change of the internal structure, and the battery state can be effectively represented.

2. The support vector machine has obvious advantages in small sample and non-linear problems. The electrochemical impedance spectrum and the support vector machine algorithm are simultaneously applied, the overcharge and overdischarge diagnosis can be rapidly carried out through the characteristic impedance, the diagnosis method can be effectively applied to the overcharge and overdischarge fault diagnosis in the whole life cycle of the battery, the reference is provided for the battery management strategy, the battery safety is favorably improved, and meanwhile, the diagnosis method has important significance for screening and recycling of the retired battery.

Drawings

Fig. 1 is a flowchart of a method for detecting overcharge and overdischarge of a lithium battery based on characteristic impedance according to an embodiment of the present disclosure;

FIG. 2 is a schematic diagram of an optimal classification plane of a support vector machine according to another embodiment of the present disclosure;

FIG. 3 is a flow chart of a fault diagnosis model based on a particle swarm optimization support vector machine according to another embodiment of the disclosure;

FIG. 4 is a diagram of a final test set prediction result based on a particle swarm optimization support vector machine model according to another embodiment of the disclosure;

fig. 5 is an electrochemical impedance spectrum of a lithium battery to be tested according to another embodiment of the present disclosure;

FIG. 6 shows the result of an overcharge-overdischarge test on a lithium battery to be tested according to another embodiment of the present disclosure;

FIG. 7 (a) is a diagram illustrating the result of a decision tree detection confusion matrix provided by another embodiment of the present disclosure;

fig. 7 (b) is a KNN detection confusion matrix result diagram provided by another embodiment of the present disclosure;

fig. 7 (c) is a graph of a bayesian classifier detection confusion matrix result provided by another embodiment of the present disclosure;

FIG. 7 (d) is a diagram illustrating the result of detecting the confusion matrix by the artificial neural network according to another embodiment of the present disclosure;

fig. 7 (e) is a diagram illustrating a result of detecting a confusion matrix by a basic SVM according to another embodiment of the present disclosure.

Detailed Description

Specific embodiments of the present disclosure will be described in detail below with reference to fig. 1 to 7 (e). While specific embodiments of the disclosure are shown in the drawings, it should be understood that the disclosure can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

It should be noted that certain terms are used throughout the description and following claims to refer to particular components. As one skilled in the art will appreciate, various names may be used to refer to a component. This specification and claims do not intend to distinguish between components that differ in name but not function. In the following description and in the claims, the terms "include" and "comprise" are used in an open-ended fashion, and thus should be interpreted to mean "include, but not limited to. The description which follows is a preferred embodiment of the disclosure, but is made for the purpose of illustrating the general principles of the disclosure and not for the purpose of limiting the scope of the disclosure. The scope of the present disclosure is to be determined by the terms of the appended claims.

To facilitate an understanding of the embodiments of the present disclosure, the following detailed description is to be considered in conjunction with the accompanying drawings, and the drawings are not to be construed as limiting the embodiments of the present disclosure.

In one embodiment, as shown in fig. 1, the present disclosure provides a method for detecting overcharge and overdischarge of a lithium battery based on characteristic impedance, comprising the steps of:

s100: acquiring an electrochemical impedance spectrum of a lithium battery to be detected;

s200: extracting the characteristic impedance of the lithium battery to be tested based on the electrochemical impedance spectrum;

s300: and inputting the characteristic impedance into an over-charge and over-discharge detection model to realize over-charge and over-discharge detection of the lithium battery to be detected.

The above embodiments constitute a complete technical solution of the present disclosure. The method has the advantages that the information is represented by using the electrochemical impedance spectroscopy, the representation is comprehensive, the battery does not need to be disassembled, and meanwhile, the battery state can be effectively represented by being sensitive to the change of the battery state and the change of the internal structure.

In another embodiment, the characteristic impedances include ohmic internal resistance, SEI film internal resistance, charge transfer resistance, and low frequency diffusion impedance.

In this embodiment, in an actual working condition, the battery may be in an overcharge fault cycle, an overdischarge fault cycle, and a normal cycle aging, and different structural changes may be generated inside the battery in different fault states, where the normal cycle may cause loss of active lithium inside the lithium battery; overcharge failure cycles can result in significant loss of active lithium from the lithium battery and the generation of lithium dendrites at the negative electrode material; overdischarge failure cycles do not result in loss of the battery negative electrode current collector and precipitation of copper dendrites of the positive electrode material, but only in loss of active lithium and reduction of SEI film thickness. Besides, the SOC, SOH and temperature of the battery can affect the electrochemical impedance spectrum test of the battery, resulting in non-negligible change of the ac impedance value. In order to detect and classify the fault state of the lithium battery, characteristic parameters capable of effectively representing the fault state need to be extracted. This example tests the alternating current impedance of overcharge, overdischarge and normal cycles at 25 deg.C and 100% SOC of a lithium battery over its life cycle, and characterizes the electrochemical reaction inside the battery during charging and discharging simultaneously, with the results shown in tables 1 to 3 in the following order:

TABLE 1.35V overcharge faulty Battery characterization parameters values (label = 1)

Table 2.50-4.20V normal cycle battery characteristic parameter values (label = 2)

TABLE 3.00V overdischarge fault cell characterization parameter values (label = 3)

It can be readily seen from the above table that the ohmic internal resistance R under overcharge failure cycles is higher than that under normal cycles ₀ SEI film internal resistance R _SEI Charge transfer resistance Re and low frequency diffusion resistance R _0.1 All exhibit an increase in ohmic internal resistance R under over-discharge fault cycles ₀ Charge transfer resistance Re and low frequency diffusion resistance R _0.1 Increased SEI film internal resistance R due to over-discharge failure _SEI A drop occurs. The charge transfer resistance has no obvious change under normal, overcharge fault and overdischarge fault circulation and is maintained to fluctuate within a certain interval.

Therefore, in order to effectively distinguish the battery cycling state, the ohmic internal resistance R is used in the embodiment ₀ SEI film internal resistance R _SEI A charge transfer resistor R _e And low frequency diffusion impedance magnitude R _0.1 As a characteristic impedance for detecting overcharge and overdischarge failures of a lithium battery.

In another embodiment, the overcharge-overdischarge detection model employs a support vector machine detection model based on particle swarm optimization.

In this embodiment, the support vector machine is a classification model of two classes, and its basic principle is to establish an optimal classification plane in the space where the samples are located, so that the plane can effectively distinguish the samples, and ensure that the classification interval between the samples is maximized, and its two-dimensional schematic diagram is shown in fig. 2, corresponding to the fault diagnosis in this embodiment, i.e. to find a hyperplane to effectively distinguish the battery overcharge and overdischarge faults from the normal cycle.

In fig. 2, the solid dots and the hollow dots correspond to the overcharge and overdischarge failure detection problems of the lithium battery according to the present embodimentFor different fault states of the battery, H ₀ To enable accurate detection of battery faults, classification model H ₁ And H ₂ Is the line closest to the different kinds of samples, and H ₀ In parallel, wherein H ₁ And H ₂ The distance between is called the classification interval and falls within H ₁ And H ₂ The upper sample point is called the support vector. The diagnosis of the lithium battery faults is a multi-dimensional data model, when two-dimensional data is expanded into a multi-dimensional space, a straight line separating the two faults is evolved into a hyperplane, and the optimal classification plane is determined by a support vector. The following two conditions need to be satisfied in the multidimensional space: the two types of faults should be divided on one side of the hyperplane, respectively, and the distance from the point closest to the plane in the two types of faults reaches the maximum. While the equation for any hyperplane in an N-dimensional space can be represented by the following equation, w is an N-dimensional vector:

w ^T x+b＝0 (1)

in this embodiment, after the lithium battery fault data extends the model to the multidimensional space, the impedance data point R (R) in different fault states ₀ ，R _SEI ，R _e ，R _0.1 ) The distance to any plane can be expressed as:

wherein w is a hyperplane normal vector, b is a parameter,

as can be seen from FIG. 2, the distance d from the point of each fault state to the classification hyperplane is not less than d, from which it can be deduced that

If let | | w | | | d =1, merging the equations can be abbreviated as:

y(w ^T R+b)≥1 ( ₄ )

and the distance of the support vector to the hyperplane is:

from the foregoing y (w) ^T R + b) is more than or equal to 1, and y (w) can be obtained ^T R+b)＝|w ^T R + b |, and for the support vector, y (w) ^T R + b) =1, so the classification pitch maximum can be scaled as:

to calculate | | w | | non-calculation ² The minimum value of (d) can be calculated and solved by a lagrangian method, and the optimization condition of the corresponding lagrangian function is solved, so that the optimization function can be obtained as follows:

wherein alpha is _i Is Lagrange multiplier.

Furthermore, when such a non-linear classification problem is detected for faults of the present disclosure, an optimal classification hyperplane in a low dimensional space does not exist, and thus a mapping function φ (R) may be utilized _i ) Non-linear samples of a low dimensional space are mapped to a high dimension and a relaxation variable ξ is introduced on each sample of the original linear classification _i (ξ _i Not less than 0), the learning of the support vector machine model is realized by utilizing a linear classification method in a high-dimensional space. The optimal classification hyperplane under high dimension is:

y _i (w ^T φ(R _i )+b)-1+ξ _i ≥0

introducing kernel function K (R) at the same time _i ·x)＝(φ(R _i )·φ(R _j ) To avoid "dimensional disasterHard ", one can get the optimization function:

the gaussian kernel function in the common kernel functions is most widely applied due to its strong mapping capability, so the present embodiment constructs a classification model of a support vector machine using the gaussian kernel function as the kernel function.

There are many parameters in the classification algorithm of the support vector machine, wherein the main influencing factors are a penalty factor C and a kernel function parameter O. The penalty parameter represents the penalty strength for misjudgment of the sample, and the smaller the penalty parameter is, the higher the tolerance of the model to the misjudgment rate is, the lower the complexity is, otherwise, the maximum complexity is. The kernel function parameter indicates the generalization ability of the model, the kernel function used in the embodiment is a gaussian kernel function, the kernel function parameter O controls the action range of the kernel function, the larger the O value is, the better the O value is, the larger the O value reaches the peak value, the larger the O value is, the larger the O value continues to be, and the generalization ability of the model becomes worse. In the conventional calculation, the two parameters are manually given or are subjected to gridding search by setting a given interval, the calculation capability and the calculation complexity of the model are poor, and with the development of an algorithm technology, an algorithm is adopted to optimize the selection of the parameters, such as a genetic algorithm, a particle swarm algorithm and the like, and the particle swarm algorithm is adopted to optimize the selection of the parameters of the support vector machine model.

The particle swarm optimization is mainly based on the biological evolution theory in nature and bird feeding behavior, and comprises the steps that all particles are searched in a solution space, each particle has own position and speed, the particles move independently to search for an optimal solution, the particles are communicated with one another, the optimal solution can be exchanged, the optimal solution in the current state is finally found, then all particles adjust own positions and speeds, the searching is repeated, and the overall optimal solution is finally found. The method is applied to the lithium battery fault detection, the model diagnosis accuracy is taken as an adaptive value, the local optimal correspondence is taken as a current optimal classification model, and the global optimal correspondence is taken as a model with the highest diagnosis accuracy. In the particle swarm algorithm, the position and velocity of a particle are set by:

v _t+1 ＝wv _t +r ₁ ·rand()·(P _t -x _t )+r ₂ ·rand()·(G _t -x _t ) (10)

x _t+1 ＝x _t +v _t (11)

wherein x is _t ＝(x _t1 ，...，x _tn ) And v _t ＝(v _t1 ，...，v _tn ) Is the position and velocity of the particle at the current time t, w is the inertial weight, r ₁ And r ₂ As a constant, rand () is a generated random number, and P, G is the self optimal position and the global optimal position of the particle at the current time t.

Based on this, the process for constructing the support vector machine detection model based on particle swarm optimization in this embodiment is as shown in fig. 3:

1. inputting characteristic impedance of the lithium battery in different cycle states;

2. selecting a kernel function (in the embodiment, a Gaussian function is used as the kernel function), initializing a particle swarm, setting a particle swarm scale, a penalty value, the maximum iteration number and an initial parameter value of the kernel function, and then establishing a classification model of the support vector machine according to the parameters;

3. calculating a fitness function of each particle, and then updating the position and the speed of the particle according to the formulas (10) and (11) so as to determine new parameters and construct a new support vector machine diagnosis model;

4. and (3) judging whether the fitness function of the particles meets the precision requirement when the iteration times reach the maximum (when the maximum value of the fitness function calculated for n times is equal to the maximum value of the fitness function for n-1 times, the precision requirement is met), if so, obtaining the optimal support vector machine detection model, otherwise, updating the positions and the speeds of the particles, returning to the step 3, and reconstructing a new support vector machine diagnosis model.

In the embodiment, a support vector machine algorithm is adopted to construct a fault diagnosis model, and polarization internal resistances corresponding to different electrochemical processes and electrochemical impedance amplitudes of the battery in a low frequency region are taken as input characteristic parameters. The classification model of the support vector machine can learn the characteristic parameters and the battery overcharge and overcharge during trainingAfter the model training is completed, the characteristic parameters of the battery in an unknown state are input into the model, and the overcharge and overdischarge faults are diagnosed. In the model, the initial value of the learning factor is set to c ₁ ＝1.5，c ₂ =1.6, maximum iteration number maxgen =50, population size setting sizepop =10.00, and penalty factor c is in the value interval of [0.1, 1000]The value interval of the kernel function parameter g is [0.01, 1000%]. As can be seen from tables 1 to 3, as the state of health decreases, the characteristic parameter values of the battery impedance increase and decrease, 234 sets of overcharge failure, overdischarge failure and normal cycle data for different cycle numbers of the battery at 100% SOC, 174 sets were randomly extracted as the training data set of the SVM classification model and input, and the remaining 60 sets were used as the model test set to verify the classification effect of the model. A final test set prediction result graph based on the particle swarm optimization support vector machine model is shown in fig. 4. In fig. 4, tag 1 represents overcharge, tag 2 represents normal battery, tag 3 represents overdischarge, and over-charge, normal and over-discharge battery test sets 20 groups of data, and finally, only one group of overcharged and normal batteries are classified according to the classification model, and the accuracy reaches 96.67%.

In order to facilitate further understanding of the method of the present disclosure, a lithium battery to be tested is taken as an example for description.

1. Acquiring an electrochemical impedance spectrum of the lithium battery to be tested, as shown in fig. 5;

2. analysis of FIG. 5 using the relaxation time Distribution (DRT) method (which is a conventional method in the art and will not be described in detail herein) extracts the ohmic internal resistance R ₀ SEI film internal resistance R _SEI A charge transfer resistor R _e And low frequency diffusion resistance R _0.1 4 feature quantities are taken as the characteristic impedance, and the four feature quantities are respectively R in FIG. 5 ₀ ＝24.67，R _SEI ＝2.80，R _e ＝1.41，R _0.1 ＝36.08；

3. The characteristic impedance is input into an overcharge and overdischarge detection model, an output result shows that the lithium battery to be tested is in an overcharge state, the state of the lithium battery to be tested is actually tested by the method, the test result is shown in fig. 6, and the model output result is consistent with the actual test result when the lithium battery to be tested is in the overcharge state.

Note that the over-charge and over-discharge detection model detects the characteristic impedance by the following procedure, which is exemplified as follows:

% results analysis, showing whether the test results and the actual results match

toc

bestc＝global_x(1)

bestg＝global_x(2)

bestCVaccuarcy＝-fit_gen(maxgen)

cmd＝[′-c′，num2str(bestc)，′-g′，num2str(bestg)]；

model＝libsvmtrain(Y_train，X_train，cmd)；

[predict_train，train_accuracy，dec_value1]＝libsvmpredict(Y_train，X_train，model)；

train_accuracy

Y_train_part＝Y_test；

X_train_part＝X_test；

[predict_train，test_accuracy，dec_value1]＝libsvmpredict(Y_train_part，X_train_part，model)；

test_accuracy

figure；

hold on；

plot(Y_train_part，′o′)；

plot(predict_train，′r*′)；

xlabel ('test set sample', 'FontSize', 12);

ylabel ('class label', 'FontSize', 12);

legend ('actual test set classification', 'predictive test set classification');

title ('actual classification and prediction classification of test set', 'FontSize', 12);

grid minor

box on

set(gca，′xtick′，[0：1：1]，′ytick′，[1 2 3])

next, in order to effectively screen out the classification algorithm adapted to the application scenario of the present disclosure, a classification diagnosis model is constructed by using different classification algorithms, and the accuracy of the data shown in the present disclosure is compared with the accuracy of the commonly used different algorithms. Algorithm verification ten-fold cross-validation was selected, and the Positive Predictive Value (PPV) and false hair occurrence rate (FDR) for the different algorithms are shown in fig. 7 (a) to 7 (e).

According to the actual classification and prediction classification results of the final test set, the positive prediction value and the false occurrence rate, different algorithms and the accuracy rates of the support vector machine classification models are obtained as shown in table 4:

TABLE 4

As can be seen from Table 4, in the classification learning time, the decision tree and the K-NN are relatively long in classification learning time, and the other algorithm time is about 1 s; in the aspect of classification accuracy, the accuracy of the support vector machine is the highest of several algorithms, wherein the classification accuracy of the support vector machine optimized by the particle swarm can reach 96.67%, and the model classification learning time is 1.91s. And as can be seen from the sample number, the present disclosure is directed to the diagnosis of overcharge and overdischarge failures of lithium batteries under study, which is a classification problem for a class of small samples. The simultaneous comparison of other classification algorithms shows that the support vector machine algorithm has good adaptability to the small sample problem. The support vector machine model optimized by the particle swarm is proved to have higher adaptability to the detection scene of the overcharge and overdischarge fault cycle of the lithium battery. And the characteristic quantity of the relaxation time distribution of the lithium battery can effectively detect and distinguish the overcharge and overdischarge fault cycle states of the battery, and the branch vector machine classification algorithm can be applied to the detection of the overcharge and overdischarge fault cycles of the battery.

In another embodiment, the present disclosure further provides a lithium battery overcharge and overdischarge detection apparatus based on characteristic impedance, including:

the acquisition unit is used for acquiring an electrochemical impedance spectrum of the lithium battery to be tested;

the extraction unit is used for extracting the characteristic impedance of the lithium battery to be detected based on the electrochemical impedance spectrum;

and the detection unit is used for inputting the characteristic impedance into the over-charge and over-discharge detection model so as to realize over-charge and over-discharge detection of the lithium battery to be detected.

The above embodiments are merely illustrative of the technical concepts and features of the present disclosure, and are intended to enable those skilled in the art to understand the present disclosure and implement the present disclosure, and not to limit the scope of the present disclosure. All equivalent changes and modifications made according to the spirit of the present disclosure should be covered within the scope of the present disclosure.

Claims (6)

1. A lithium battery overcharge and overdischarge detection method based on characteristic impedance comprises the following steps:

s100: acquiring an electrochemical impedance spectrum of a lithium battery to be detected;

s200: extracting the characteristic impedance of the lithium battery to be tested based on the electrochemical impedance spectrum;

s300: and inputting the characteristic impedance into an over-charge and over-discharge detection model to realize over-charge and over-discharge detection of the lithium battery to be detected.

2. The method according to claim 1, wherein in step S200, the characteristic impedance comprises ohmic internal resistance, SEI film internal resistance, charge transfer resistance and low frequency diffusion resistance.

3. The method of claim 1, wherein said overcharge-overdischarge detection model employs a particle swarm optimization-based support vector machine detection model in step S300.

4. The method according to claim 3, wherein the particle swarm optimization-based support vector machine detection model is constructed by the following steps:

s301: inputting characteristic quantity impedance values of the lithium battery in different circulation states;

s302: selecting a Gaussian function as a kernel function;

s303: initializing a particle swarm, and setting the population scale of the particles, penalty values, maximum iteration times, particle positions and speeds and initial parameter values of a kernel function to construct a support vector machine detection model;

s304: calculating the fitness function of each particle, judging whether the fitness function of the particle meets the precision requirement when the iteration times reach the maximum, and if so, obtaining an optimal support vector machine detection model; otherwise, the position and the speed of the particle are updated and the step S303 is returned to reconstruct a new diagnostic model of the support vector machine.

5. The method of claim 4, wherein in step S303, the particle position and velocity are set by:

v _t+1 ＝wv _t +r ₁ ·rand()·(P _t -x _t )+r ₂ ·rand()·(G _t -x _t )

x _t+1 ＝x _t +v _t

wherein x is _t Is the position of the particle at the current time, v _t Is the velocity of the particle at the current time, w is the inertial weight, r ₁ And r ₂ Is constant, rand () is a generated random number, P _t 、G _t The current time t is the self optimal position and the global optimal position of the particle.

6. A lithium battery overcharge and overdischarge detection device based on characteristic impedance comprises:

the acquisition unit is used for acquiring an electrochemical impedance spectrum of the lithium battery to be tested;

the extraction unit is used for extracting the characteristic impedance of the lithium battery to be detected based on the electrochemical impedance spectrum;

and the detection unit is used for inputting the characteristic impedance into the over-charge and over-discharge detection model so as to realize over-charge and over-discharge detection of the lithium battery to be detected.

Images