Retired battery health degree calculation method based on cycle life test data

Abstract

The invention discloses a retired battery health degree calculation method based on cycle life test data, and the method comprises the steps: comparing short-time discharge data with a prior database of batteries of the same model, reversely deducing the health degree of the batteries through the corresponding relation between voltage and capacity, and calculating the health degree of the batteries only through the comparison of the short-time discharge data. The full-full and full-discharge process required by the ampere-hour integral method is omitted, and the calculation speed is higher; compared with the prior data of the batteries of the same model, the SOH calculation has the basis and is not a pure prediction and equivalent calculation method, so that the precision is high.

Description

Technical Field

[0001] The present invention relates to the technical field of battery status evaluation, and in particular to a method for calculating the health of retired batteries based on cycle life test data. Background Art

[0002] According to the national standard for new energy vehicles, power batteries can be "retired" when the remaining capacity is 80%. In the energy storage industry, the cascade utilization of these batteries can greatly save costs. However, the historical operation data of retired batteries is difficult to obtain, and the original management system of the battery cannot be used to estimate the health. Therefore, a retired battery health calculation method based on cycle test data is proposed for health management of reused retired batteries.

[0003] However, due to the complex battery aging mechanism, retired batteries from different batches or even the same batch have poor consistency, and the SOH estimation method applicable to the first life stage is often not feasible for retired batteries. For the SOH estimation problem of different retired power batteries, a lot of manpower is needed to carry out feature design and build prediction models. This is one of the difficulties in the current SOH estimation of retired batteries.

[0004] Chinese patent document CN105322536A discloses a "retired battery health status assessment method based on Bayesian deep learning". The typical multi-terminal flexible DC distribution system is divided into AC system protection zone, main converter protection zone, DC line protection zone and access converter protection zone; on the basis of the protection area division, combined with the operating requirements of the DC distribution system and its fault electrical characteristics and other factors, a protection scheme is configured for each protection zone. The scheme should include several different protection algorithms to protect the safety of the equipment in this area and serve as backup protection for other areas at the same time. The sampling process of the above technical scheme is complex and prone to errors. Summary of the invention

[0005] The present invention mainly solves the technical problem that the sampling process of the original technical solution is complicated and prone to errors. It provides a retired battery health calculation method based on cycle life test data, which uses short-time discharge data to compare with a priori database of batteries of the same model, and uses the correspondence between voltage and capacity to infer the battery health. Only short-time discharge data needs to be compared, eliminating the full charge and discharge process required by the ampere-hour integration method, and the calculation speed is faster; by comparing with the priori data of batteries of the same model, the calculation of SOH has a basis to follow, and it is not a simple prediction or equivalent calculation method, so the accuracy is higher.

[0006] The above technical problem of the present invention is mainly solved by the following technical solution: The present invention comprises the following steps: S1 before the retired battery is reused, a full charge and discharge cycle life test is performed at a specified rate X; S2 records the data during each discharge as a priori data set and constructs a three-dimensional curve; When the retired battery is reused, S3 performs short-term external discharge at the specified rate X; S4 takes the average temperature of the short-time external discharge process, and selects data within the range before and after the average temperature from the prior data set as the cycle test data; S5 calculates the cycle test data to obtain the maximum likelihood life curve as the battery health at a certain moment.

[0007] Preferably, the step S1 specifically includes, before reusing the retired battery, taking one of the batteries, performing a full charge and discharge cycle life test at a specified rate X, recording the number of each cycle, selecting a time granularity, and recording the current I, voltage U and temperature T data during each discharge process. The time granularity is "1s".

[0008] Preferably, the step S1 further comprises, for each cycle, calculating the current state of health SOH, and stopping the cycle life test when the battery SOH drops to 60% or below.

[0009] Preferably, the step S2 specifically includes taking the temperature-voltage-capacity data in the priori data set to construct a three-dimensional curve.

[0010] Preferably, for each timestamp, the current cumulative capacity Q is calculated as follows: Where Δt is the time granularity, and t0 is the starting time of the cycle corresponding to time t.

[0011] Preferably, for each cycle, the current health status SOH is calculated as follows: Where Q n Q is the capacity released by the complete discharge in the nth cycle, x is the rated capacity of the battery.

[0012] Preferably, the step S3 obtains the current I, voltage U and temperature T data during the short-time external discharge process, and calculates the cumulative discharge capacity Q at each moment to construct a short-time external discharge data set. avg , select T in the prior data set avg Cyclic test data within ±3℃, that is, three-dimensional curve collection, select T according to the temperature coordinate axis avg The curve set is within the range of ±3° C. The preferred embodiment of the present invention is 60 seconds, i.e. 60 time points.

[0013] As a preference, the step S4 uses a gradient descent method to calculate the minimum root mean square error (RMSE) of each curve in the short-time external discharge data set and the cycle test data. min .

[0014] Preferably, the step S5 includes interpolating the target curve using the voltage data of 60 points in the short-time external discharge data set as the y-axis coordinate to obtain 60 target capacities Q ′ data; Construct the objective function: Among them, Q n is the cumulative discharge capacity at the nth moment in the short-term external discharge data set; Q ′ n is the target capacity corresponding to the nth moment obtained by interpolation; Δx is the value of the short-time external discharge dataset being shifted to the right along the x-axis as a whole; RMSE(Δx) is the root mean square error between the short-time external discharge dataset and the corresponding point on the selected curve under the current shift amount.

[0015] Preferably, the step S5 further comprises calculating the gradient of RMSE with respect to Δx: Where, lim(ε)=0; Iteratively update Δx: Where Δx (k) is the translation amount of the kth iteration, α is the learning rate; Set the convergence conditions: Compare the minimum root mean square error (RMSE) of each curve min ), take the curve with the smallest minimum RMS deviation as the maximum likelihood life curve, and use the SOH corresponding to this curve as the battery health at time t1.

[0016] The beneficial effects of the present invention are as follows: short-time discharge data is compared with a priori database of batteries of the same model, and the corresponding relationship between voltage and capacity is used to infer the battery health. Only short-time discharge data needs to be compared, eliminating the full charge and discharge process required by the ampere-hour integration method, and the calculation speed is faster; by comparing with the priori data of batteries of the same model, the calculation of SOH has a basis to follow, and it is not a simple prediction or equivalent calculation method, so the accuracy is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the present invention.

[0018] Figure 2It is a three-dimensional schematic diagram of cycle life test data of the present invention.

[0019] Figure 3 This invention is a T avg Cyclic test data curve within ±3℃.

[0020] Figure 4 It is a voltage-capacity curve diagram under different health levels of the present invention.

[0021] Figure 5 It is a curve translation diagram of data set A of embodiment 2 of the present invention. DETAILED DESCRIPTION

[0022] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the technical solutions of the present application are further described in detail below through embodiments and in combination with the accompanying drawings. It should be understood that the specific implementation method described here is only an optimal embodiment of the present application, which is only used to explain the present application and does not limit the protection scope of the present application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0023] The cascade utilization of retired power batteries has attracted more and more attention. Power batteries used in new energy vehicles can be recycled for other purposes when the available capacity degrades to 80% of the initial capacity. In order to ensure the safety of the cascade utilization of retired power batteries, it is necessary to monitor the SOH of retired batteries during secondary use. At present, the SOH estimation method for the first life stage of batteries has been widely studied. However, due to the complex battery aging mechanism, the consistency of retired batteries from different batches or even the same batch is poor, and the SOH estimation method suitable for the first life stage is often not feasible for retired batteries. For the SOH estimation problem of different retired power batteries, a lot of manpower is required for feature design and the construction of prediction models. This is one of the difficulties in the current SOH estimation of retired batteries.

[0024] According to data, the existing battery health calculation technology generally monitors the health of new batteries throughout their life cycle. The more commonly used methods are: 1) Ampere-hour integration method: Perform a complete charge and discharge operation on the battery under test, and integrate the current data in the process by time to obtain the maximum capacity that can be discharged at present.

[0025] 2) Battery operating parameters + data-driven method: Currently, many literatures and patents have collected various parameters of batteries under different health levels, such as internal resistance, electrochemical impedance, electrochemical capacitance, equivalent capacitance, self-discharge resistance, etc., and used deep learning and other technologies to train health assessment models.

[0026] 3) Life curve method: Use the "cycle count-life" curve of the same model battery leaving the factory to estimate the current health status based on the current number of cycles.

[0027] Disadvantages of existing technology: 1. Ampere-hour integration method: Since this method performs a complete charge and discharge process, its calculation of health is the most accurate. However, it takes a long time to fully charge and discharge the power battery, so it is not suitable for real-time health calculation of working batteries.

[0028] 2. Battery equivalent model + data-driven method: Different operating parameters have different difficulties in obtaining and different prediction accuracy.

[0029] 1) For example, although the “battery internal resistance” is easy to measure and the accuracy of health calculation using internal resistance is high under fixed working conditions, it is greatly affected by temperature and charge and discharge rate, which in turn affects the model progress; 2) For example, the "electrochemical impedance spectroscopy method" can reflect the condition of the electrode and interface structure during the aging process of the battery. In theory, it can accurately calculate the health of the battery. However, first of all, the measurement of electrochemical impedance spectroscopy is extremely difficult and requires the use of high-precision professional equipment and has very high requirements for the test environment. Secondly, this method is derived based on the equivalent circuit model. Generally, the equivalent model cannot accurately reflect the actual operation of the battery, so the accuracy of the health calculation cannot be very high.

[0030] 3. Life curve method: Since this method uses the current number of cycles to compare with the life curve to get the health, it is necessary to know all the operating data of the battery since it left the factory, which is unrealistic for retired batteries. In addition, the actual health is affected by historical operating conditions, which may cause the health of the battery in the middle and late stages to deviate from the "cycle number-life" curve. Therefore, there is a certain error in estimating the health only by the number of cycles.

[0031] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the drawings; the process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0032] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0033] Embodiment: This embodiment provides a method for calculating the health of retired batteries based on cycle life test data, such as Figure 1 As shown, the following steps are included: S1 Before the retired battery is reused, a full charge and discharge cycle life test is performed at a specified rate X. Specifically, before the retired battery is reused, one of the batteries is taken out and a full charge and discharge cycle life test is performed at a specified rate X. The number of each cycle is recorded, the time granularity is selected, and the current I, voltage U and temperature T data during each discharge process are recorded. The time granularity is "1s". For each cycle, the current health SOH is calculated, and the cycle life test is stopped when the battery SOH drops to 60% or below.

[0034] For each timestamp, calculate the current cumulative capacity Q, the calculation method is as follows: Where Δt is the time granularity, and t0 is the starting time of the cycle corresponding to time t.

[0035] For each cycle, the current health status SOH is calculated as follows: Where Q n Q is the capacity released by the complete discharge in the nth cycle, x is the rated capacity of the battery.

[0036] S2 records the data of each discharge process as a priori data set and constructs a three-dimensional curve, specifically including taking the temperature-voltage-capacity data in the priori data set to construct the three-dimensional curve.

[0037] S3 When the retired battery is reused, it is discharged to the outside for a short time at the specified rate X. The current I, voltage U and temperature T data during the short-time external discharge are obtained, and the cumulative discharge capacity Q at each moment is calculated to construct a short-time external discharge data set. The average temperature T of the discharge process is taken avg , select T in the prior data set avg Cyclic test data within ±3℃, that is, three-dimensional curve collection, select T according to the temperature coordinate axis avg The curve set is within the range of ±3° C. The preferred embodiment of the present invention is 60 seconds, i.e. 60 time points.

[0038] S4 takes the average temperature of the short-time external discharge process, and selects the data before and after the average temperature from the prior data set as the cycle test data. Use the gradient descent method to calculate the minimum root mean square error (RMSE) of each curve in the short-time external discharge data set and the cycle test data. min .

[0039] S5 calculates the cycle test data to obtain the maximum likelihood life curve as the battery health at a certain moment. The voltage data of 60 points in the short-term external discharge data set is used as the y-axis coordinate, and the target curve is interpolated to obtain 60 target capacities Q ′ data; Construct the objective function: Among them, Q n is the cumulative discharge capacity at the nth moment in the short-term external discharge data set; Q ′ n is the target capacity corresponding to the nth moment obtained by interpolation; Δx is the value of the short-time external discharge dataset being shifted to the right along the x-axis as a whole; RMSE(Δx) is the root mean square error between the short-time external discharge dataset and the corresponding point on the selected curve under the current shift amount.

[0040] Find the gradient of RMSE with respect to Δx: Where, lim(ε)=0; Iteratively update Δx: Where Δx (k) is the translation amount of the kth iteration, α is the learning rate; Set the convergence conditions: Compare the minimum root mean square error (RMSE) of each curve min ), take the curve with the smallest minimum RMS deviation as the maximum likelihood life curve, and use the SOH corresponding to this curve as the battery health at time t1.

[0041] Example 1 1. Loop test data acquisition and prior database construction 1.1 Before reusing a retired battery of a certain model, take one of the batteries and perform a full charge and discharge cycle life test at a specified rate X. Record the number of each cycle, and use "1s" as the time granularity to record the current (I), voltage (U), and temperature (T) data during each discharge process.

[0042] 1.2 For each timestamp, calculate the current cumulative capacity (Q). The calculation method is as follows: Where Δt is the time granularity, and t0 is the starting time of the cycle corresponding to time t.

[0043] 1.3 For each cycle, calculate the current health status (SOH). The calculation method is as follows: Where Q n Q is the capacity released by the complete discharge in the nth cycle, x is the rated capacity of the battery.

[0044] 1.4 When the battery SOH drops to 60% or below, stop the cycle life test.

[0045] 1.5 Take the temperature-voltage-capacity data in the above data set to construct a three-dimensional curve, such as Figure 2 shown.

[0046] 2 Health calculation based on cycle life test data 2.1 When a retired battery is reused, a health calculation request is received at time t1, and the battery performs a short-term external discharge operation (preferably 60 seconds, i.e. 60 time points) at a specified rate X. The current (I), voltage (U), and temperature (T) data during the discharge process are obtained, and the cumulative discharge capacity (Q) at each moment is calculated according to the method described in 1.2.

[0047] 2.2 Take the average temperature of the discharge process (T avg ), select T from the prior data set described in step 1 avg Cyclic test data within ±3℃, i.e. Figure 2 In the curve shown, select T by clicking the temperature axis box. avg The curve set within the range of ±3℃, the box selection result is as follows Figure 3 shown.

[0048] 2.3 Take the 60 sets of voltage and cumulative discharge capacity data obtained in 2.1 as data set A. Under the premise that it can be translated as a whole along the x-axis (capacity axis), use the gradient descent method to find its relative Figure 2 The minimum root mean square error (RMSE) of each curve in min ), the specific calculation method is as follows: 2.3.1 Using the voltage data of data set A as the y-axis coordinate, interpolate the target curve to obtain 60 target capacities (Q ′ )data.

[0049] 2.3.2 Constructing the objective function: Among them, Q n is the cumulative discharge capacity at the nth moment in data set A; Q ′ nis the target capacity corresponding to the nth moment obtained by interpolation in 2.3.1; Δx is the value of the entire data set taken in 2.3 translated to the right along the x-axis; RMSE(Δx) is the root mean square error between data set A and the corresponding point on the selected curve under the current translation amount.

[0050] 2.3.3 Find the gradient of RMSE with respect to Δx: Among them, lim(ε)=0.

[0051] 2.3.4 Iterative update Δx: Where Δx (k) is the translation amount at the kth iteration and α is the learning rate.

[0052] 2.3.5 Set convergence conditions: 2.4 Compare the minimum root mean square error (RMSE) of each curve min ), take the curve with the smallest minimum RMS deviation as the maximum likelihood life curve, and use the SOH corresponding to this curve as the battery health at time t1.

[0053] The overall route diagram is as follows Figure 1 shown.

[0054] Battery health is defined as The starting voltage and ending voltage of a battery are not affected by its health. Therefore, when the SOH is lower, the voltage changes more when the same capacity is released. Figure 4 The main idea of ​​the present invention is to compare the short-time discharge data with the prior database of the same type of battery, so that the battery health can be inferred from the corresponding relationship between voltage and capacity.

[0055] The advantages of this method are: 1) The calculation speed is faster, because only short-time discharge data is needed for comparison, eliminating the full charge and discharge process required by the ampere-hour integration method.

[0056] 2) Comparing with the prior data of the same model battery makes the calculation of SOH have a basis to follow. It is not a prediction or equivalent calculation method, so it has higher accuracy.

[0057] Example 2 1. Loop test data acquisition and prior database construction 1.1 A battery with good health, model FC2007A155, is tested for cycle life. Each cycle includes five steps: a. constant current charging, b. constant voltage charging, c. resting, d. constant current discharging, and e. resting. a Constant current charging: charging at a constant current rate of 3C (60A current) causes the battery voltage to rise from the starting voltage (2.495V) to the cut-off voltage (3.65V) b. Constant voltage charging: Immediately after step a, start constant voltage charging with a starting current of 100A, and end when the current drops to 1A. c. Wait: Wait for 10 minutes. d Constant current discharge: discharge at a constant current rate of 3C (60A current), and end when the battery voltage drops to the starting voltage (2.495V) e Standby: Standby for 10 minutes After one cycle is completed, the next cycle is performed. During the test, the number of cycles, current, voltage, temperature, and timestamp are recorded during each constant current discharge, with a recording interval of 1 second.

[0058] 1.2 Calculate the cumulative capacity released at each moment in each cycle, The first second is 59.9702*1 / 3600=0.16658Ah The second second is 59.9764*1 / 3600+0.16658=0.033319Ah... Table 1 Cumulative discharge capacity at each moment in each cycle 1.3 Calculate the health degree corresponding to each cycle. It is known that the total capacity discharged in the first cycle test is 21.1517Ah, which is used as the standard. The total capacity discharged in the second cycle test is 21.0946Ah, so the health degree in the second cycle test is 21.0946 / 21.1517=99.62%.

[0059] 1.4 When the health degree obtained after a certain cycle is less than 60%, the cycle test is ended and the database construction is completed.

[0060] 2 Health calculation based on cycle life test data 2.1 Select another retired battery of the same model. To measure its current health, discharge it at a constant current of 3C rate (60A current) for 60 seconds from the current moment, with 1 second as the time granularity, and record the current, voltage, and temperature data during this period (hereinafter referred to as data set A). If the current in the first second is -59.9826A (negative numbers indicate discharge), the cumulative discharge capacity in the first second is 59.9826*1 / 3600=0.01667Ah, the current in the second second is -59.995A, the cumulative discharge capacity in the second second is 59.995*1 / 3600+0.01667=0.03332Ah, the current in the third second is -59.9826A, the corresponding cumulative discharge capacity is 59.9826*1 / 3600+0.3332=0.04998Ah, and so on to complete the calculation of the cumulative capacity of 60 data.

[0061] 2.2 The average temperature measured in 60 seconds is 26.81°C. Test data within the range of ±3°C (23.81-29.81°C) are selected from the prior database as the data to be compared (hereinafter referred to as data set B).

[0062] 2.3 In the data set A obtained in the above steps, the capacity gradually increases from 0.01667Ah to 0.5998Ah, and the voltage gradually decreases from 2.97725V to 2.95795V, which is a monotonically decreasing curve. First, compare this curve with the "capacity-voltage" curve recorded in the first cycle in data set B. The principle of comparison is: under the same voltage drop, the closer the battery discharge capacity is, the closer its health is. Therefore, when comparing, the curve of data set A can be translated left and right along the "capacity axis".

[0063] 2.3.1 The coordinates of the first point of data set A are (0.01667, 2.97725). According to the voltage value (2.97725V), the mapping value is found on the first cycle data curve. It is found that 2.97725 is between (17.015394, 2.97795) and (17.148627, 2.97686). According to the interpolation method, the coordinates of the mapping point on the curve are (17.10096, 2.97725); the second point of data set A is (0.03332, 2.97693). The same method is used to calculate the corresponding point on the first cycle data curve as (17.14007, 2.97693)... and so on. Find the points along the horizontal x-axis mapped on the first cycle data curve for the 60 points in data set A (hereinafter referred to as data set C).

[0064] 2.3.2 Find the root mean square error (rmse) between data set A and data set C. For the above curve, 2.3.3 Shift the data set A to the right along the “capacity axis” by a small value to find the gradient of rmse with respect to “capacity”. Here, the shift value ε=0.0001 is taken, then Find the gradient 2.3.4 Shift the dataset A to the right by Δx. Set the learning rate α = 0.0001, then 2.3.5 Repeat 2.3.2-2.3.4 and continue to calculate RMSE. Continue to update Δx. Until When the curve of data set A moves to this point, it is considered that the fit with data set C is the highest. In this case, after the curve of data set A is shifted to the right by 17.0932, At this moment, record the RMSE between data sets A and C = 0.000311.2.4 According to step 2.3, calculate the RMSE when the fitting degree between each cycle data curve in data set A and data set B is the highest. In this case, the RMSE of the 153rd cycle data in data set A and data set B is the smallest, which is 0.0000941821. The corresponding health degree of the 153rd cycle test is 98.13388%, so it is considered that the health degree of the tested battery is also 98.13388%.

[0065] The short-time discharge data is compared with the prior database of batteries of the same model, and the battery health is inferred from the corresponding relationship between voltage and capacity. Only short-time discharge data needs to be compared, eliminating the full charge and discharge process required by the ampere-hour integration method, and the calculation speed is faster. Comparing with the prior data of batteries of the same model makes the calculation of SOH have a basis to follow, rather than a simple prediction or equivalent calculation method, so the accuracy is higher.

[0066] The specific embodiments described herein are merely examples of the spirit of the present invention. The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that a technician in the technical field to which the present application belongs may make various modifications or supplements to the described specific embodiments or replace them in a similar manner, but will not deviate from the spirit of the present application or exceed the scope defined in the attached claims. For a person of ordinary skill in the art, multiple variations and improvements may also be made without departing from the concept of the present application. Therefore, the scope of protection of this patent shall be subject to the attached claims.

Claims

1. A method for calculating the health of retired batteries based on cycle life test data, characterized in that: The following steps are involved: S1 performs a full charge and discharge cycle life test at a specified rate X before reusing the retired battery; S2 records the data during each discharge process as a priori data set and constructs a three-dimensional curve; When the retired battery is reused, S3 performs short-term external discharge at the specified rate X; S4 takes the average temperature of the short-time external discharge process, and selects data within the range before and after the average temperature from the prior data set as the cycle test data; S5 calculates the cycle test data to obtain the maximum likelihood life curve as the battery health at a certain moment.

2. A method for calculating the health of retired batteries based on cycle life test data according to claim 1, characterized in that: The step S1 specifically includes, before the retired batteries are reused, taking one of the batteries, performing a full charge and discharge cycle life test at a specified rate X, recording the number of each cycle, selecting a time granularity, and recording the current I, voltage U and temperature T data during each discharge process.

3. The method for calculating the health of retired batteries based on cycle life test data according to claim 1, characterized in that: The step S1 also includes, for each cycle, calculating the current state of health SOH, and stopping the cycle life test when the battery SOH drops to 60% or below.

4. The method for calculating the health of retired batteries based on cycle life test data according to claim 1, characterized in that: The step S2 specifically includes taking the temperature-voltage-capacity data in the priori data set to construct a three-dimensional curve.

5. A method for calculating the health of retired batteries based on cycle life test data according to claim 1, 2 or 3, characterized in that: For each timestamp, calculate the current cumulative capacity Q, the calculation method is as follows: Where Δt is the time granularity, and t0 is the starting time of the cycle corresponding to time t.

6. The method for calculating the health of retired batteries based on cycle life test data according to claim 3, characterized in that: For each cycle, the current health status SOH is calculated as follows: Where Q n Q is the capacity released by the complete discharge in the nth cycle, x is the rated capacity of the battery.

7. The method for calculating the health of retired batteries based on cycle life test data according to claim 1, characterized in that: The step S3 obtains the current I, voltage U and temperature T data during the short-time external discharge process, calculates the cumulative discharge capacity Q at each moment, and constructs a short-time external discharge data set.

8. The method for calculating the health of retired batteries based on cycle life test data according to claim 7, characterized in that: The step S4 uses the gradient descent method to calculate the minimum root mean square error (RMSE) of each curve in the short-time external discharge data set and the cycle test data. min .

9. A method for calculating the health of retired batteries based on cycle life test data according to claim 8, characterized in that: The step S5 includes taking the voltage data of 60 points in the short-time external discharge data set as the y-axis coordinate, interpolating the target curve, and obtaining 60 target capacities Q ′ data; Construct the objective function: Among them, Q n is the cumulative discharge capacity at the nth moment in the short-term external discharge data set; Q ′ n is the target capacity corresponding to the nth moment obtained by interpolation; Δx is the value of the short-time external discharge dataset being shifted to the right along the x-axis as a whole; RMSE(Δx) is the root mean square error between the short-time external discharge dataset and the corresponding point on the selected curve under the current shift amount.

10. A method for calculating the health of retired batteries based on cycle life test data according to claim 9, characterized in that: The step S5 further includes calculating the gradient of RMSE with respect to Δx: Where, lim(ε)=0; Iterative update Δx: Where Δx (k) is the translation amount of the kth iteration, α is the learning rate; Set the convergence conditions: Compare the minimum root mean square error (RMSE) of each curve min ), take the curve with the smallest minimum RMS deviation as the maximum likelihood life curve, and use the SOH corresponding to this curve as the battery health at time t1.

Images