Lithium battery health state evaluation method of EKF (Extended Kalman Filter) optimization algorithm considering capacity and cycle index

Abstract

The invention relates to the field of battery health state prediction, in particular to a lithium battery health state evaluation method based on an EKF optimization algorithm considering capacity and cycle index. According to the technical scheme, when voltage, current and time are known, a second-order equivalent circuit model of the lithium battery is firstly established, the influence of the number of cycles on the maximum available capacity is considered, and a model between SOC-OCV-Cycs is established; the capacity is dynamically adjusted after one-time circulation is finished; when the Kalman filtering algorithm is used for correcting errors, the influence of aging is considered, and the estimation precision of the SOH can be achieved by comprehensively considering the influence of the number of cycles.

Description

Technical field:

[0001] The present invention relates to the field of battery health state prediction, and in particular to a lithium battery health state evaluation method using an EKF optimization algorithm that takes capacity and cycle number 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, they will inevitably age, and their maximum available capacity and power performance will decrease. Battery health status is an important indicator for evaluating the degree of lithium battery aging. Predicting and evaluating the health status of batteries can help people detect the health status of lithium batteries. More accurate estimation of battery health status can effectively avoid safety hazards caused by battery aging, and promptly remind staff to replace batteries when they need to be replaced.

[0003] Since the battery health state cannot be measured directly, but can be indirectly characterized by the current maximum available capacity of the battery, current research often uses a combined method of SOC (State of Charge) and SOH for estimation. Therefore, the estimation accuracy of SOC has a significant impact on the final result of SOH.

[0004] Current issues:

[0005] The patent (CN 116381507 A) establishes a sixth-order mapping relationship between OCV and SOC, uses the adaptive forgetting factor recursive least squares algorithm (AFFRLS) to estimate model parameters, and predicts SOC by integrating the unscented Kalman filter (UKF) with the genetic algorithm. However, it does not consider the impact of the number of cycles on the maximum available capacity during actual use.

[0006] The SOH defined according to capacity is shown in formula (2):

[0007]

[0008] Q max is the maximum available capacity of the current battery, Q i It is the initial capacity rating of the battery when it leaves the factory. IEEE Standard 1188-1996 stipulates that when the current capacity of a lithium battery reaches 80% of the rated capacity, the battery is considered to have failed. max When the value drops to 80%, it can no longer meet the performance requirements of the power battery. max It is difficult to measure directly, so Q needs to be calculated indirectly through measurable physical quantities. max , and Q max It will decrease with the increase of usage time. Summary of the invention:

[0009] In order to solve the current deficiencies, the purpose of the present invention is to provide a lithium battery health status assessment method based on an EKF optimization algorithm that takes capacity and cycle number into consideration.

[0010] To solve the above problems, the technical solution adopted by the present invention is: a lithium battery health status assessment method based on an EKF optimization algorithm that considers capacity and cycle number. When the voltage, current and time are known, a second-order equivalent circuit model of the lithium battery is first established, considering the influence of the number of cycles on the maximum available capacity, and constructing a model between SOC-OCV-Cycs; the capacity is dynamically adjusted at the end of one cycle; when correcting the error through the Kalman filter algorithm, the influence of aging is considered, and the influence of the number of cycles is comprehensively considered to achieve the estimation accuracy of SOH.

[0011] Further, the following steps are included:

[0012] Step 1): Collect voltage, current and time data and perform data preprocessing;

[0013] For the charging and discharging process of the same type of lithium battery at different temperatures, the voltage, current, and charging and discharging time data are collected, saved as a database file, and data preprocessing is performed;

[0014] Step 2): Building a SOC estimation model based on an equivalent second-order circuit model;

[0015] First, the dynamic relationship model between SOC and voltage and current is established as follows:

[0016]

[0017] V d =V oc (SOC)-V c1 -V c2 -I·R0

[0018] Where: Q N is the rated capacity of the battery; η is the discharge efficiency; SOC is the actual capacity of the battery; I is the actual discharge current;

[0019] U0, R0, R1, R2, C1, C2 in the above dynamic equations are identified by identification method;

[0020] Based on the above SOC dynamic relationship model and the parameter values ​​identified by parameters, the state space equation and observation equation are established, and x = [SOC V c1 V c2 ] T , then it is expressed as:

[0021]

[0022] Step 3): Calculate the SOC at different times based on the state equation in step 2, and calculate the reference SOC value using the ampere-hour integration method;

[0023] The percentage of the changed capacity is calculated using the ampere-hour integration method, and the difference between the initial SOC and the changed SOC is finally obtained, that is, the remaining capacity, which is expressed as:

[0024]

[0025] Step 4): Perform preliminary error correction using the EKF algorithm;

[0026] The specific process is:

[0027] 1) System initialization;

[0028] 2) Prior estimation and error covariance matrix generation;

[0029] 3) Calculate the Kalman gain matrix;

[0030] 4) Correcting the prior state estimate according to the observed value and the Kalman gain to obtain the posterior state estimate and a new error covariance matrix;

[0031] Step 5): Considering cycle aging and capacity correction, the EKF algorithm in step 4 is improved;

[0032] Based on step 4), the capacity calibration link and the coupling effect of cycle aging on SOC are added, and the improved process is as follows:

[0033] 1) comparing the reference SOC output value in step 3) with the SOC threshold;

[0034] 2) If it is lower than the threshold, determine again whether it is in the charging state;

[0035] 3) If it is in the charging state, the capacity calibration is performed and the cycle number is updated, and the OCV model is corrected;

[0036] During capacity calibration, the upper and lower calibration thresholds are set, which is completed using the following capacity calibration formula:

[0037]

[0038] Where: Q N is the total capacity after calibration; Q is the charge value obtained by the ampere-hour integration method during the charging process; SOC β is the estimated SOC value of the model at the end of calibration; SOC α is the estimated SOC value of the model at the beginning of calibration;

[0039] Considering the influence of cycle number on battery aging, the ninth-order polynomial fitting method is used to fit the mapping relationship to construct the lithium battery cycle aging OCV-SOC estimation model:

[0040]

[0041] Where U OCV (SOC, T) is the cycle-number derivative matrix of the OCV-SOC model coefficients; K i is the parameter to be solved; T is the cycle number matrix; SOC is the remaining capacity of the battery; the calculation formula for the current cycle number is:

[0042]

[0043] Where T is one cycle, ΔT1 is the assumed charging time, and ΔT2 is the discharge time;

[0044] 4) Otherwise, perform the preliminary error correction in step 4

[0045] Step 6): Establish a coupled health status assessment model considering cycle aging factors to predict the health status SOH of lithium batteries:

[0046] From SOH's definition of capacity, we can see that:

[0047] According to the calculation formula of SOC:

[0048] Combining the above two formulas, we can get:

[0049] Where: Q i is the initial capacity rating of the battery when it leaves the factory, Qn is the capacity value calculated according to the ampere-hour integration method in step 3, Q max is the maximum available capacity of the current battery, obtained through the relationship between SOC and Qn after correction in step 5;

[0050] 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.

[0051] Furthermore, in step 5), the capacity correction and consideration of cycle aging are performed only when the remaining battery capacity is less than 20% and charging begins, and the calibration is terminated when the remaining battery capacity is greater than 90% and charging ends. The cycle aging model is updated during each charge and discharge process.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] (1) The present invention takes into account the influence of the number of cycles on the remaining maximum capacity of the battery, and considers the influence of the number of cycles by constructing a model between OCV-SOC-Cycs, so as to achieve SOH estimation; by fitting and verifying the ninth-order relationship of the OCV-SOC relationship, it is proved that the ninth-order relationship can well describe the OCV-SOC relationship; in addition, the relationships under different number of cycles are fitted and verified respectively, and the model relationship between OCV-SOC-Cycs is obtained.

[0054] (2) The present invention performs a dynamic capacity calibration on the total capacity before estimating the remaining capacity of the battery to ensure a more accurate total capacity value when the battery is used next time, and uses an aging priori formula to correct the OCV model in the filter model while calibrating the total capacity.

[0055] (3) The present invention uses an extended Kalman filter algorithm to combine the open circuit voltage method with the ampere-hour integration method, uses a priori estimation and a posteriori estimation depending on the accuracy of the model, and uses the relationship between the open circuit voltage and SOC to continuously correct the SOC obtained by the ampere-hour method, thereby obtaining an SOH estimate.

[0056] (4) The present invention combines the charge and discharge time and the change of SOC during each cycle, and calculates the actual total number of cycles of the battery during use, which can provide higher accuracy for considering the cycle aging factor and then evaluating the battery health status. Description of the drawings:

[0057] Figure 1 It is a flow chart of lithium battery health status assessment.

[0058] Figure 2 It is a cycle number description diagram.

[0059] Figure 3 This is the flow chart of the Kalman filter algorithm.

[0060] Figure 4 It is a flow chart of the improved EKF algorithm based on cyclic aging prior correction.

[0061] Figure 5 This is a comparison chart of SOC error estimation using different methods. Specific implementation method:

[0062] 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.

[0063] The present invention provides a battery health status assessment method considering the cycle number and battery capacity coupling state, which mainly includes the following steps: Figure 1 As shown:

[0064] Step 1: Collect voltage, current, and time data and perform data preprocessing;

[0065] For the charging and discharging process of the same type of lithium battery at different temperatures, the current, voltage, and charging and discharging time data are collected, or can be monitored by the BMS system, and then saved as a database file and pre-processed for later analysis;

[0066] Step 2: Construct a SOC estimation model based on an equivalent second-order circuit model;

[0067] First, the dynamic relationship model between SOC and voltage and current is established as follows:

[0068]

[0069] V d =V oc (SOC)-V c1 -V c2 -I·R0

[0070] Where: Q N is the rated capacity of the battery; η is the discharge efficiency; SOC is the actual capacity of the battery; I is the actual discharge current;

[0071] Then, U0, R0, R1, R2, C1, and C2 in the above dynamic equation are identified by least squares method or maximum likelihood estimation;

[0072] Finally, based on the above SOC dynamic relationship model and the parameter values ​​identified by parameters, the state space equation and observation equation are established, and x = [SOC V c1 V c2 ] T , then it is expressed as:

[0073]

[0074] The above constructs the SOC estimation model, including the dynamic model, state space equation and observation equation.

[0075] Step 3: Calculate the reference SOC value using the ampere-hour integration method;

[0076] The ampere-hour integration method is used to calculate the integral of the charge and discharge current and the corresponding time within a certain period of time, thereby calculating the percentage of the changed capacity, and finally finding the difference between the initial SOC and the changed SOC, that is, the remaining capacity, which is expressed as:

[0077]

[0078] Step 4: Perform preliminary error correction through the EKF algorithm;

[0079] The core contents of EKF include the state equation and the measurement equation:

[0080] x k =f(x k-1 ,u k-1 )+w k

[0081] y k =g(x k )+v k

[0082] Where: w k is the process error; v k is the measurement error; x k is the state quantity obtained by the SOC estimation model, y k is the SOC calculated by the ampere-hour integration method.

[0083] Both errors obey normal distribution, and obey covariance matrices Q and R respectively. The above equation is linearized to define the Jacobian matrix:

[0084]

[0085] Combined with the state space equation in step 2, discretization is obtained:

[0086]

[0087] In summary, the state estimation value of SOC can be obtained by combining the spatial state equation with the EKF algorithm.

[0088] In order to make the accuracy of the model closer to the actual data, the battery capacity calculated by the ampere-hour integration method in step 3 above is used as a reference value for the actual capacity and compared with the state estimation value of the SOC above. When there is a large error between the two, the EKF algorithm is used to correct the error. The advantage of the EKF algorithm is that the state quantity at the current moment takes into account the influence of the previous moment, and can correct the state quantity at the current moment according to the error value.

[0089] The specific process of using the EKF algorithm to calculate the error is as follows Figure 3 , including the following steps:

[0090] 1) System initial settings:

[0091]

[0092] 2) Prior estimation and error covariance matrix generation:

[0093]

[0094] 3) Calculate the Kalman gain matrix:

[0095]

[0096] 4) According to the observation value and Kalman gain, the prior state estimate is corrected to obtain the posterior state estimate and the new error covariance matrix:

[0097]

[0098] Step 5: Considering cycle aging and capacity correction, the EKF algorithm in step 4 is improved, such as Figure 5 As shown;

[0099] This step adds the capacity calibration link and the coupling effect of cycle aging on SOC on the basis of step 4. The improved process is as follows:

[0100] 1) Compare the reference SOC output value in step 3 above with the SOC threshold;

[0101] 2) If it is lower than the threshold, determine again whether it is in the charging state;

[0102] 3) If in charging state, perform capacity calibration and update the number of cycles, and perform OCV-SOC model correction;

[0103] As the battery ages, the discharge time will become shorter under the same discharge rate conditions, which directly reflects the continuous attenuation of the total capacity of the battery. Therefore, the total capacity should be calibrated before estimating the remaining capacity of the battery. When calibrating the capacity, set the upper and lower thresholds of the calibration, which is completed through the following capacity calibration formula:

[0104]

[0105] Where: Q N is the total capacity after calibration; Q is the charge value obtained by the ampere-hour integration method during the charging process; SOC β is the estimated SOC value of the model at the end of calibration; SOC α is the estimated SOC value of the model at the beginning of calibration;

[0106] During calibration, set the upper and lower thresholds for calibration, set SOCmin to 20%, and SOCmax to 90%. The system will perform capacity calibration only when the remaining battery capacity is lower than 20% and charging begins, and the calibration will end when the remaining battery capacity is higher than 90% and charging ends.

[0107] Considering the influence of cycle number on battery aging, the ninth-order polynomial fitting method is used to fit the mapping relationship to construct the lithium battery cycle aging OCV-SOC estimation model:

[0108]

[0109] Where U OCV (SOC, T) is the cycle-number derivative matrix of the OCV-SOC model coefficients; K i is the parameter to be solved; T is the cycle number matrix; SOC is the remaining capacity of the battery;

[0110] By collecting the charge and discharge time of the battery, considering that it will not be fully charged and discharged every time during actual use, the number of cycles in actual use is defined, and the number of battery cycles is divided into quantities related to SOC. SOC will increase with charging time and decrease with discharge time. Therefore, the current number of cycles can be calculated in any charge and discharge cycle. The number of cycles within a period of time is reflected as the sum of the number of cycles each time. In this way, the changes in the health status of the battery as the number of cycles increases can be comprehensively considered. The calculation formula for the current number of cycles is:

[0111]

[0112] Among them, T is one cycle, ΔT1 is the assumed charging time, and ΔT2 is the discharge time. Figure 2 As shown;

[0113] 4) Otherwise, perform the preliminary error correction in step 4

[0114] The total capacity correction link and the OCV model correction link are carried out simultaneously. In actual use, every time the battery completes a cycle, the EKF algorithm is corrected according to the capacity calibration formula and the cycle aging prior formula. The correction is performed after the battery has completed a cycle.

[0115] Step 6: Establish a coupled health status assessment model considering cycle aging factors to predict the health status SOH of lithium batteries:

[0116] From SOH's definition of capacity, we can see that:

[0117] According to the calculation formula of SOC:

[0118] Combining the above two formulas, we can get:

[0119] Where: Q i is the initial capacity rating of the battery when it leaves the factory, Qn represents the capacity value calculated by the time-based integration method in step 3, Q max is the maximum available capacity of the current battery, which can be obtained through the relationship between SOC and Qn after correction in step 5.

[0120] 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.

[0121] Based on the above-mentioned improvement method, CALCE battery data was selected for simulation verification. The SOC estimation accuracy of lithium-ion batteries was compared under the scenarios of the cycle aging adaptive OCV estimation model and the traditional extended Fahlman filter model. The reference SOC was obtained by the ampere-hour integration method. The comparison results are shown in Figure 2. Figure 5 As shown in the figure, it can be found that as the model matching degree of the traditional EKF model gradually decreases, the difference between the observed value and the model output value will show a trend of gradually increasing, and the SOC estimation value will deviate more and more from the reference value in the continuous cycle. However, the model output value of the self-correcting EKF model is also continuously corrected as the battery is continuously cycled, which can suppress the deviation of the SOC estimation value from the reference value as the cycle continues.

[0122] 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 lithium battery health status assessment method based on an EKF optimization algorithm considering capacity and cycle number, characterized in that: When the voltage, current and time are known, the second-order equivalent circuit model of the lithium battery is first established, the influence of the number of cycles on the maximum available capacity is considered, and the model between SOC-OCV-Cycs is constructed; the capacity is dynamically adjusted after one cycle; when correcting the error through the Kalman filter algorithm, the influence of aging is considered, and the influence of the number of cycles is comprehensively considered to achieve the estimation accuracy of SOH.

2. The lithium battery health status assessment method of the EKF optimization algorithm considering capacity and cycle number according to claim 1 is characterized in that: The specific steps include the following: Step 1): Collect voltage, current and time data and perform data preprocessing; For the charging and discharging process of the same type of lithium battery at different temperatures, the voltage, current, and charging and discharging time data are collected, saved as a database file, and data preprocessing is performed; Step 2): Building a SOC estimation model based on an equivalent second-order circuit model; First, the dynamic relationship model between SOC and voltage and current is established as follows: V d =V oc (SOC)-V c1 -V c2 -I R0 Where: Q N is the rated capacity of the battery; η is the discharge efficiency; SOC is the actual capacity of the battery; I is the actual discharge current; U0, R0, R1, R2, C1, C2 in the above dynamic equations are identified by identification method; Based on the above SOC dynamic relationship model and the parameter values ​​identified by parameters, the state space equation and observation equation are established, and x = [SOC V c1 V c2 ] T , then it is expressed as: Step 3): Calculate the SOC at different times based on the state equation in step 2, and calculate the reference SOC value using the ampere-hour integration method; The percentage of the changed capacity is calculated using the ampere-hour integration method, and the difference between the initial SOC and the changed SOC is finally obtained, that is, the remaining capacity, which is expressed as: Step 4): Perform preliminary error correction using the EKF algorithm; The specific process is: 1) System initialization; 2) Prior estimation and error covariance matrix generation; 3) Calculate the Kalman gain matrix; 4) Correcting the prior state estimate according to the observed value and the Kalman gain to obtain the posterior state estimate and a new error covariance matrix; Step 5): Considering cycle aging and capacity correction, the EKF algorithm in step 4 is improved; Based on step 4), the capacity calibration link and the coupling effect of cycle aging on SOC are added, and the improved process is as follows: 1) comparing the reference SOC output value in step 3) with the SOC threshold; 2) If it is lower than the threshold, determine again whether it is in the charging state; 3) If it is in the charging state, the capacity calibration is performed and the cycle number is updated, and the OCV model is corrected; During capacity calibration, the upper and lower calibration thresholds are set, which is completed using the following capacity calibration formula: Where: Q N is the total capacity after calibration; Q is the charge value obtained by the ampere-hour integration method during the charging process; SOC β is the estimated SOC value of the model at the end of calibration; SOC α is the estimated SOC value of the model at the beginning of calibration; Considering the influence of cycle number on battery aging, the ninth-order polynomial fitting method is used to fit the mapping relationship to construct the lithium battery cycle aging OCV-SOC estimation model: Where U OCV (SOC, T) is the cycle-number derivative matrix of the OCV-SOC model coefficients; K i is the parameter to be solved; T is the cycle number matrix; SOC is the remaining capacity of the battery; the calculation formula for the current cycle number is: Where T is one cycle, ΔT1 is the assumed charging time, and ΔT2 is the discharge time; 4) Otherwise, perform the preliminary error correction in step 4 Step 6): Establish a coupled health status assessment model considering cycle aging factors to predict the health status SOH of lithium batteries: From SOH's definition of capacity, we can see that: According to the calculation formula of SOC: Combining the above two formulas, we can get: Where: Q i is the initial capacity rating of the battery when it leaves the factory, Qn is the capacity value calculated according to the ampere-hour integration method in step 3, Q max is the maximum available capacity of the current battery, obtained through the relationship between SOC and Qn after correction in step 5; 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. The lithium battery health status assessment method of the EKF optimization algorithm considering capacity and cycle number according to claim 1 is characterized in that: In the step 5), the capacity correction and consideration of cycle aging are performed only when the remaining battery capacity is less than 20% and charging begins, and the calibration is terminated when the remaining battery capacity is greater than 90% and charging ends. The cycle aging model is updated during each charge and discharge process.

Images