Lithium battery health state evaluation method considering internal resistance and temperature

Abstract

The invention belongs to the technical field of battery state-of-health prediction, and particularly relates to a lithium battery state-of-health (SOH) evaluation method considering internal resistance and temperature. The invention aims to solve the problems of large internal resistance estimation error and low precision during traditional SOH estimation. According to the specific scheme, when voltage, current and temperature are known, a lithium battery second-order equivalent circuit model is firstly established, and the current ohmic internal resistance is identified according to an FFRLS optimization algorithm; an Adam optimization algorithm and FFRLS are fused for improvement, and forgetting factors are adaptively adjusted by correcting errors so as to optimize the accuracy of the model; and a coupling evaluation model is obtained by considering the influence of the temperature on the internal resistance, and the SOH of the battery can be obtained through calculation.

Description

Technical field:

[0001] The present invention belongs to the technical field of battery health state prediction, and in particular relates to a lithium battery health state (State of Health, SOH) evaluation method taking internal resistance and temperature into consideration. Background technology:

[0002] With the widespread application of new energy, lithium batteries are used in the field of electric vehicles. However, as lithium batteries are continuously charged and discharged, their internal structure begins to gradually age, the active substances decrease, the internal resistance increases, and the battery health status becomes worse and worse.

[0003] Battery health status is an important indicator for evaluating the degree of lithium battery aging. According to the current industry standards for power batteries, when the battery SOH drops below 80%, it means that the battery needs maintenance and repair. Therefore, predicting and evaluating the battery health status can help people detect the health status of lithium batteries. Accurately estimating the battery health status can effectively avoid safety hazards caused by battery aging and promptly remind staff to replace the battery when it needs to be replaced.

[0004] Currently, SOH calculation can be obtained by the following formula:

[0005]

[0006] In the formula, R e is the internal resistance value when the battery reaches the end of its service life, R n is the current internal resistance of the battery, R i It is the internal resistance value of the battery when it leaves the factory. According to the internal resistance expression of SOH, the SOH of the battery is characterized by the internal resistance, but in actual use, the internal resistance of the battery is often unknown. It is related to the complex chemical structure inside the battery, and the temperature will affect the discharge characteristics of the battery, which is manifested as an increase or decrease in resistance in the equivalent model.

[0007] The internal resistance of a battery can indicate the health status of the battery. As it increases with the use time, the health status tends to decline. However, the change of internal resistance during actual use is also affected by temperature changes. The internal resistance of batteries at different temperatures will change when used. Therefore, considering the relationship between temperature and internal resistance, coupling the relationship between the two when evaluating the health status of the battery can theoretically achieve a more accurate SOH estimation.

[0008] Patent CN 106353687 A discloses a method for predicting the health status of lithium batteries, which uses the change rate of the terminal voltage and the internal resistance of the lithium battery within a fixed time as health assessment factors, and then calculates the health status of the lithium battery over the entire life cycle according to the different attenuation levels of the lithium battery. The problem is that the temperature of the battery will change with charging and discharging and specific usage scenarios, and the temperature environment in this method is set, and the coupling effect of temperature on internal resistance is not considered. When the temperature changes, the model cannot be updated in real time.

[0009] Patent CN 116679211 A discloses a method for predicting the health status of lithium batteries. By establishing a second-order equivalent circuit model of lithium batteries, the SDO algorithm, Gaussian regression machine learning model, and Transformer deep learning model are used to calculate three SOH values, and then the three methods are weighted and fused to obtain the final health status. The problems are: machine learning and deep learning have high requirements for the data sets that need to be trained, require a large amount of data, have a large amount of calculation, and have a strong dependence on data. When the training data is unbalanced, the prediction results are not very interpretable. Summary of the invention:

[0010] In order to solve the current deficiencies, the purpose of the present invention is to provide a lithium battery health status assessment method taking into account the coupling effect between temperature and internal resistance, and to achieve lithium battery SOH assessment when the battery parameters (current, voltage, temperature) are known.

[0011] To achieve the purpose of the present invention, the technical solution adopted by the present invention is: a method for evaluating the health status of a lithium battery taking into account internal resistance and temperature: when the voltage, current, and temperature are known, firstly, a second-order equivalent circuit model of the lithium battery is established, and the current ohmic internal resistance is identified according to the FFRLS optimization algorithm; the Adam optimization algorithm is integrated with the FFRLS for improvement, and the forgetting factor is adaptively adjusted by correcting the error to optimize the accuracy of the model; then, the influence of temperature on the internal resistance is considered to obtain a coupled evaluation model, and the battery health status SOH can be calculated.

[0012] Further, the specific steps are:

[0013] The following steps are involved:

[0014] Step 1: Collect voltage, current, and temperature data;

[0015] For the charging and discharging process of the same type of lithium battery at different temperatures, the current and voltage data are collected, saved as a database file, and the data is preprocessed;

[0016] Step 2: Construct a second-order RC equivalent circuit model;

[0017] Based on the Thevenin model, a resistor-capacitor parallel circuit is connected in series to form a second-order RC equivalent circuit model.

[0018] Step 3: Use FFRLS combined with the Adam optimization algorithm to identify the parameters of the internal resistance and identify the current internal resistance.

[0019] Step 4: Obtain the relationship between internal resistance and temperature based on comparative experiments at different temperatures, and identify the temperature internal resistance correlation coefficient;

[0020] The relationship between internal resistance and temperature is expressed according to the Arrhenius equation and temperature coefficient as follows:

[0021] R(T) = R0 * exp(α * (T - T0)) (18)

[0022] Where R(T) is the internal resistance at temperature T, R0 is the internal resistance at reference temperature T0 (usually 25 degrees Celsius), and α is the temperature coefficient, which indicates the sensitivity of the internal resistance to temperature changes;

[0023] The internal resistance at different temperatures is identified through the fusion algorithm, and the temperature coefficient between temperature and internal resistance is obtained by fitting.

[0024] Step 5: Establish a coupled health status assessment model for internal resistance and temperature to predict the health status SOH of lithium batteries:

[0025]

[0026] When the SOH value is greater than 80%, it means that the battery is in a safe period; when it reaches 80%, it reaches the safety threshold; when the SOH value is less than 80%, the battery needs to be replaced.

[0027] Furthermore, the second-order RC equivalent circuit model in step 2) includes an ohmic internal resistance R0, an open circuit voltage U oc 、Load current I L , two equivalent polarized resistors R1 and R2, two equivalent polarized capacitors C1 and C2, the circuit model is expressed as:

[0028]

[0029] Further, the specific method of step 3) is:

[0030] 1) First, the mathematical relationship between battery voltage, current and model parameters is derived based on the second-order RC equivalent circuit model; and the discrete mathematical formulas between battery voltage, current and model parameters are obtained, and the current internal resistance is identified through FFRLS;

[0031] 2) Secondly, the voltage error between the model output and the real data is calculated, and the voltage value U of the current model is calculated according to the currently calculated internal resistance parameter value. model ;

[0032] 3) Finally, the voltage value inferred by the model calculation is calculated and the known output voltage error is calculated. When the error is within the allowable range, the current forgetting factor is used to continue the internal resistance calculation at the next moment; when the error is large, the Adam optimization algorithm is used to correct the error of the parameter identification result, and the forgetting factor is automatically adjusted according to the corrected data, and then the internal resistance is re-identified by the FFRLS algorithm.

[0033] Compared with the prior art, the advantages of the present invention are:

[0034] 1. The present invention adopts a second-order RC equivalent circuit model: on the basis of the Thevenin model, a RC parallel circuit is connected in series, and two RC parallel circuits are used to simulate concentration polarization and electrochemical polarization respectively. While describing the nonlinear characteristics of the lithium battery, it more accurately describes the polarization effect during the charging and discharging process of the lithium battery.

[0035] 2. The present invention adopts a fusion algorithm to realize the estimation of internal resistance: by using the Adam optimization algorithm to fuse the FFRLS algorithm, the forgetting factor can dynamically change with the change of the error to achieve a more accurate estimation of the internal resistance, thereby achieving an accurate estimation of the SOH.

[0036] 3. The present invention realizes the identification of internal resistance at different temperatures, and fits the functional relationship between internal resistance and temperature in the normal temperature range; taking into account the fact that the internal resistance will change when the temperature changes, the correlation coefficient between temperature and internal resistance is obtained by exploring the change of internal resistance at different temperatures, so that the internal resistance prediction at different temperatures can be realized.

[0037] 4. The Adam optimization algorithm used in the present invention can achieve a smooth and stable transition of the gradient and can adapt well to the changing objective function. In addition, the algorithm records the first-order moment and second-order moment of the gradient during the optimization process and can generate an adaptive learning rate for the forgetting factor to match the dynamic response of the battery. Description of the drawings:

[0038] Figure 1 It is a flow chart of lithium battery health status assessment.

[0039] Figure 2 It is the equivalent second-order RC circuit model diagram.

[0040] Figure 3 It is the flow chart of FFRLS-Adam fusion algorithm. Specific implementation method:

[0041] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only 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. The technical solutions provided by the embodiments of the present invention are described in detail below in conjunction with the drawings.

[0042] The present invention considers constructing an equivalent circuit model to obtain the equivalent internal resistance, and uses the voltage and current of the battery to identify the battery by designing an identification algorithm to obtain the ohmic internal resistance and polarization internal resistance of the battery. Since Rn at different temperatures shows differences, the results of parameter identification at different temperatures are used to describe this difference and obtain the relationship between the two. Then, Rn is generalized by a fitting function, that is, the temperature coefficient between temperature and internal resistance is obtained, and then SOH is obtained.

[0043] This embodiment provides a lithium battery health status assessment method, which realizes lithium battery SOH estimation when battery parameters (current, voltage, temperature) are known, and mainly includes the following steps: Figure 1 As shown:

[0044] Step 1: Collect voltage, current and temperature data;

[0045] For the charging and discharging process of the same type of lithium battery at different temperatures, the current and voltage data are collected, or can be monitored by the BMS system, then saved as a database file, and the data is pre-processed;

[0046] Step 2: Construct a second-order RC equivalent circuit model;

[0047] The second-order RC equivalent circuit model is based on the Thevenin model and adds a resistor-capacitor parallel circuit in series; Figure 2 As shown: The second-order RC circuit model includes the ohmic internal resistance R0, the open circuit voltage U oc 、Load current I L (negative when charging, positive when discharging), two equivalent polarized resistors R1 and R2, and two equivalent polarized capacitors C1 and C2. The circuit model can be expressed as follows:

[0048]

[0049] Step 3: Use FFRLS combined with the Adam optimization algorithm to identify the parameters of the internal resistance and identify the current internal resistance; since the fixed forgetting factor will affect the dynamic characteristics of the model, the Adam optimization algorithm is integrated and improved. The main purpose is to optimize the accuracy of the model by adaptively adjusting the forgetting factor. The specific process is as follows Figure 3 shown.

[0050] 1) FFRLS (Forgetting Factor Recursive Least Squares) derives the mathematical relationship between battery voltage, current and model parameters based on the second-order RC equivalent circuit model, and obtains the discrete mathematical formulas between battery voltage, current and model parameters to identify the current internal resistance.

[0051] First, we can generalize the above formula (2) to get

[0052]

[0053] Among them, τ1=R1C1, τ2=R2C2.

[0054] Substituting formula (3) into the above formula, we get

[0055] a(U oc -U L )s 2 +b(U oc -U L )s+U oc -U L =aI L R0S 2 +dI L s+cI L (4)

[0056] Transforming equation (4) into a differential equation, we can obtain

[0057]

[0058] Then, the above formula is discretized using the difference transformation method so that it can be applied online. It is known that:

[0059] The discretization result is shown in formula (6):

[0060]

[0061] In the above formula

[0062]

[0063] The final mathematical model of parameter identification is shown in formula (8):

[0064]

[0065] According to the above formula, the expressions of a, b, c, d and R0 are as follows:

[0066]

[0067] Then, other RC parameters can be deduced according to formula (9):

[0068]

[0069] At this point, the derivation of the parameter identification model is completed.

[0070] 2) Calculate the voltage error between the model output and the real data. According to the calculated internal resistance value, the voltage value Umodel of the current model can be calculated. Let the terminal voltage error be e(λ).

[0071] e(λ)=U L -U model (11)

[0072] When the error is within the allowable range, the current internal resistance calculation result is correct, and the current forgetting factor is used to continue the internal resistance calculation at the next moment. When the error is large, the Adam optimization algorithm is used to correct the error of the parameter identification result. The steps of the Adam optimization algorithm are as follows:

[0073] (1) Initialize η, β1, β2, λ, p0, q0

[0074] (2) Let error e(λ) = f t (λ) is the objective function, and the error function gradient is calculated

[0075] g t =▽ λ f t (λ)(12)

[0076] (3) Update the values ​​of p0 and q0

[0077] p t =β1p t-1 +(1-β1)*g t (13)

[0078] q t =β2q t-1 +(1-β2)*(g t ) 2 (14)

[0079] (4) Correction deviation

[0080]

[0081] (5) Update the forgetting factor

[0082]

[0083] 3) Finally, the voltage value inferred by the model calculation is calculated and the known output voltage error is calculated. When the error is within the allowable range, the current forgetting factor is used to continue the internal resistance calculation at the next moment; when the error is large, the Adam optimization algorithm is used to correct the error of the parameter identification result, and the forgetting factor is automatically adjusted according to the corrected data, and then the internal resistance is re-identified by the FFRLS algorithm;

[0084] Step 4: Obtain the relationship between internal resistance and temperature based on comparative experiments at different temperatures, and identify the temperature internal resistance correlation coefficient;

[0085] The relationship between internal resistance and temperature is expressed according to the Arrhenius equation and temperature coefficient as follows:

[0086] R(T)=R0*exp(α*(T-T0))(18)

[0087] Where R(T) is the internal resistance at temperature T, R0 is the internal resistance at reference temperature T0 (usually 25 degrees Celsius), and α is the temperature coefficient, which indicates the sensitivity of the internal resistance to temperature changes.

[0088] The internal resistance at different temperatures is identified through a fusion algorithm, and the temperature coefficient between temperature and internal resistance is obtained by fitting.

[0089] Step 5: Establish a coupled health status assessment model for internal resistance and temperature to predict the health status SOH of lithium batteries.

[0090] Based on the above parameters to identify the comprehensive influence of internal resistance and the coupling between temperature and internal resistance, the SOH evaluation model is constructed as follows:

[0091]

[0092] When the SOH value is greater than 80%, it means that the battery is in a safe period; when it reaches 80%, it reaches the safety threshold; when the SOH value is less than 80%, the battery needs to be replaced.

[0093] Explanation of symbols:

[0094] SOH (State of Health) indicates the battery health status

[0095] FFRLS stands for Forgetting Factor Recursive Least Squares

[0096] Adam (Adaptive Moment Estimation) represents the adaptive moment estimation algorithm

[0097] U L Indicates the terminal voltage of the battery in real time

[0098] U OCV Indicates the open circuit voltage during battery discharge

[0099] R0 represents the ohmic resistance of the battery

[0100] R1, R2 represent the polarization resistance of the battery

[0101] C1 and C2 represent the polarization capacitance of the battery.

[0102] I L Indicates the load current, which is positive during discharge and negative during charge

[0103] R e Expressed as the internal resistance value when the battery reaches the end of its service life

[0104] R n Indicated as the current internal resistance of the battery

[0105] R i Indicates the internal resistance of the battery when it leaves the factory

[0106] α represents the temperature coefficient

[0107] T represents the current temperature

[0108] T0 indicates the mark temperature, which is usually 25℃

[0109] λ represents the forgetting factor

[0110] U model Voltage value representing the model

[0111] e(k) represents the error of terminal voltage

[0112] η is the learning rate

[0113] β1 represents the first-order decay rate

[0114] β2 represents the second-order decay rate

[0115] ε represents a constant, which is 10 -8 , the purpose is to ensure numerical stability and prevent the divisor from being 0

[0116] g t Represents the gradient

[0117] p0 represents the initial value of the first-order matrix estimator

[0118] q0 represents the initial value of the second-order matrix estimator

[0119] p t Represents the value of the first-order matrix estimator at a certain moment

[0120] q t Represents the value of the second-order matrix estimator at a certain moment

[0121] p t-1 Represents the value of the first-order matrix estimator at the previous moment

[0122] q t-1 Represents the value of the second-order matrix estimator at the previous moment

[0123] Represents the corrected value of the first-order matrix estimator

[0124] Represents the modified value of the second-order matrix estimator

[0125] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any simple modification, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention are still within the protection scope of the technical solution of the present invention.

Claims

1. A method for evaluating the health status of a lithium battery taking into account internal resistance and temperature, characterized in that: The method adopted is: when the voltage, current and temperature are known, the second-order equivalent circuit model of the lithium battery is first established, and the current ohmic internal resistance is identified according to the FFRLS optimization algorithm; the Adam optimization algorithm is integrated with FFRLS for improvement, and the forgetting factor is adaptively adjusted by correcting the error to optimize the accuracy of the model; then the effect of temperature on the internal resistance is considered to obtain a coupled evaluation model, and the battery health status SOH can be calculated.

2. A lithium battery health status assessment method considering internal resistance and temperature according to claim 1, characterized in that: The specific steps are: The following steps are involved: Step 1: Collect voltage, current, and temperature data; For the charging and discharging process of the same type of lithium battery at different temperatures, the current and voltage data are collected, saved as a database file, and the data is preprocessed; Step 2: Construct a second-order RC equivalent circuit model; Based on the Thevenin model, a resistor-capacitor parallel circuit is connected in series to form a second-order RC equivalent circuit model. Step 3: Use FFRLS combined with the Adam optimization algorithm to identify the parameters of the internal resistance and identify the current internal resistance. Step 4: Obtain the relationship between internal resistance and temperature based on comparative experiments at different temperatures, and identify the temperature internal resistance correlation coefficient; The relationship between internal resistance and temperature is expressed according to the Arrhenius equation and temperature coefficient as follows: R(T) = R0 * exp(α * (T - T0)) (18) Where R(T) is the internal resistance at temperature T, R0 is the internal resistance at reference temperature T0 (usually 25 degrees Celsius), and α is the temperature coefficient, which indicates the sensitivity of the internal resistance to temperature changes; The internal resistance at different temperatures is identified through the fusion algorithm, and the temperature coefficient between temperature and internal resistance is obtained by fitting. Step 5: Establish a coupled health status assessment model for internal resistance and temperature to predict the health status SOH of lithium batteries: When the SOH value is greater than 80%, it means that the battery is in a safe period; when it reaches 80%, it reaches the safety threshold; when the SOH value is less than 80%, the battery needs to be replaced.

3. A lithium battery health status assessment method considering internal resistance and temperature according to claim 2, characterized in that: The second-order RC equivalent circuit model in step 2) includes an ohmic internal resistance R0, an open circuit voltage U oc 、Load current I L , two equivalent polarized resistors R1 and R2, two equivalent polarized capacitors C1 and C2, the circuit model is expressed as:

4. A lithium battery health status assessment method considering internal resistance and temperature according to claim 2 or 3, characterized in that: The specific method of step 3) is: 1) First, the mathematical relationship between battery voltage, current and model parameters is derived based on the second-order RC equivalent circuit model; and the discrete mathematical formulas between battery voltage, current and model parameters are obtained, and the current internal resistance is identified through FFRLS; 2) Secondly, the voltage error between the model output and the real data is calculated, and the voltage value U of the current model is calculated according to the currently calculated internal resistance parameter value. model ; 3) Finally, the voltage value inferred by the model calculation is calculated and the known output voltage error is calculated. When the error is within the allowable range, the current forgetting factor is used to continue the internal resistance calculation at the next moment; when the error is large, the Adam optimization algorithm is used to correct the error of the parameter identification result, and the forgetting factor is automatically adjusted according to the corrected data, and then the internal resistance is re-identified by the FFRLS algorithm.

Images