Dynamic threshold setting method for battery health grading management of battery swap station

Abstract

The application relates to the technical field of battery health grading, and discloses a dynamic threshold setting method for battery health grading management of a battery swap station, which comprises the following steps: acquiring state parameters and working condition parameters of a target battery in each charging period, and constructing a health evolution track of the target battery; setting a health grading threshold for the target battery; in each charging period, calculating a health index of the target battery based on the state parameters, and combining the health grading threshold to divide a health grade for the target battery; if the health grade of the target battery deteriorates in any charging period, predicting deviation confidence and rebound confidence of the health index of the target battery, and adjusting the health grading threshold of the target battery; and dividing the health grade for the target battery again based on the adjusted health grading threshold. The application can identify and correct the deviation in the existing health grade judgment process, and realizes robust and self-adaptive adjustment of the battery health grade.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of battery health grading, and in particular to a dynamic threshold setting method for battery health grading management of a battery swap station. BACKGROUND

[0002] Current industry health grading management of the battery of the battery swap station mainly focuses on a static threshold system, that is, a unified health grading standard is set based on the factory parameters or fixed experience values of the battery. Most battery swap stations take the SOH value as the core judgment index; some schemes supplement the voltage, internal resistance and other basic parameters as auxiliary judgment basis, but the overall health grading framework still cannot get rid of the fixed threshold and single parameter. With the expansion of the coverage range of the battery swap network and the diversification of user use scenarios, the limitations of the existing health grading management method gradually become prominent, and it has been difficult to adapt to the dynamic changes of the battery health state under complex working conditions.

[0003] The instant decision of the traditional battery health evaluation and grade division is irreversible. When the instantaneous SOH of the battery decreases or is degraded due to an abnormal behavior, the grade is usually fixed and cannot be subsequently corrected, resulting in that some temporarily deteriorated batteries are prematurely eliminated and the available resources are wasted. The existing health grading model has a one-way explanation ability for the behavior of the battery, that is, the current state and historical trend determine the health grade. However, in actual operation, there is a cognitive bias of the model itself, for example, some abnormal charging behaviors of the battery are mistakenly considered as aging, or short-term fluctuations caused by the environment are considered as health decline, resulting in distortion of the threshold division. The current grade division is usually based on a certain instantaneous or periodic data point, ignoring the subsequent self-recovery ability and state fluctuation trend of the battery, thereby causing the misjudgment frequency to rise.

[0004] A kind of quantitative judgment method of battery health state is disclosed in Chinese patent with authorization announcement No.CN111856309B, including battery health state judgment model, battery online monitoring system and quantitative evaluation system, specifically by constructing battery health state judgment model, collecting battery internal resistance, voltage, temperature and battery SOH value, battery health state is divided into three kinds of health state, sub-health state and unhealthy state, uploaded to the online monitoring system of battery, different battery data is set threshold level, quantitative evaluation system carries out battery data weight distribution, quantitative evaluation and battery health state judgment according to the set threshold level, and battery health state is displayed. The scheme judges the battery health state through multi-dimensional monitoring of the battery state, improves the accuracy of the judgment result, also facilitates technicians to master the state information of the battery, and provides a good judgment method for users.

[0005] A battery pack health state diagnosis system and method are disclosed in Chinese patent application CN104297691A, which comprises: a battery management information input device for receiving and / or collecting battery management information of a battery pack in a charging and discharging state, and sending the battery management information to an information processing device; the information processing device is used for comparing the battery management information received from the battery management information input device with a pre-set threshold value, and if any signal in the battery management information continuously exceeds the threshold value for a set time, it is determined that the battery pack has failed. This scheme considers battery SOC, battery internal resistance, battery internal resistance change rate, temperature change rate and other data for battery health state estimation, has the characteristics of more perfect and more accurate battery health state estimation, and is beneficial to safe operation of the battery pack.

[0006] The above patents all have the problem pointed out in the background art: the subsequent self-recovery ability and state fluctuation trend of the battery are ignored, resulting in an increase in the misjudgment frequency.

[0007] The information disclosed in this BACKGROUND section is only for the purpose of increasing the understanding of the general background of the application, and should not be considered as admitting or implying that the information constitutes prior art that is known to those of ordinary skill in the art. SUMMARY

[0008] The technical problem to be solved by the present application is to overcome the defects of the prior art, and to provide a dynamic threshold setting method for battery health grading management of a battery swap station, which identifies and corrects the deviation in the existing health level judgment process, and realizes stable and adaptive adjustment of the battery health level.

[0009] To solve the above technical problems, the present application provides the following technical solutions:

[0010] A dynamic threshold setting method for battery health grading management of a battery swap station, comprising the following steps:

[0011] Obtaining state parameters and working condition parameters of a target battery in each charging cycle, and constructing a health evolution trajectory of the target battery; setting a health grading threshold for the target battery;

[0012] In each charging cycle, calculating a health index of the target battery based on the state parameters, and dividing the health level of the target battery in combination with the health grading threshold;

[0013] If the health level of the target battery deteriorates in any charging cycle, predicting a reference evolution trajectory of the health index of the target battery and a rebound confidence based on the health evolution trajectory;

[0014] Calculating a deviation confidence of the health index of the target battery based on the reference evolution trajectory, and adjusting the health grading threshold of the target battery based on the deviation confidence and the rebound confidence.

[0015] reclassifying the health grade of the target battery based on the adjusted health grading threshold.

[0016] As a preferred solution of the dynamic threshold setting method for battery health grading management of the battery swap station described in the present application, wherein: the state parameters at least include SOH, SOH decay amount, average temperature, voltage variation range, internal resistance of the target battery; the working condition parameters at least include charging rate, SOC variation range, ambient temperature, cycle number, charging cycle interval;

[0017] The health evolution trajectory is represented by a health evolution sequence; and the health evolution trajectory of the target battery is constructed, specifically including: for any charging cycle, each state parameter is normalized with each working condition parameter; each state parameter is arranged into a state vector corresponding to the charging cycle, and each working condition parameter is arranged into a working condition vector corresponding to the charging cycle; the state vector and the working condition vector of any charging cycle form a health evolution factor of the corresponding charging cycle; and the health evolution factors of each charging cycle are arranged into a health evolution sequence of the target battery in chronological order.

[0018] As a preferred solution of the dynamic threshold setting method for battery health grading management of the battery swap station described in the present application, wherein: the method for calculating the health index is as follows: each state parameter is used to assign a value to the health index, and SOH is positively correlated with the health index, and SOH decay amount, average temperature, voltage variation range, and internal resistance are all negatively correlated with the health index;

[0019] The method for grading the health of the target battery is as follows: if the health index of the target battery is greater than the health grading threshold, the health grade remains unchanged, otherwise, the health grade of the target battery is downgraded;

[0020] The degradation of the health grade indicates that the health grade of the target battery is downgraded in the current charging cycle.

[0021] As a preferred solution of the dynamic threshold setting method for battery health grading management of the battery swap station described in the present application, wherein: the reference evolution trajectory contains reference state vectors of the last M charging cycles; and the method for predicting the reference evolution trajectory is as follows:

[0022] Based on the health evolution trajectory, health evolution factors of the last M charging cycles are extracted and marked as actual evolution factors; M is a positive integer;

[0023] Before extracting the M actual evolution factors, the state vectors of the last N charging cycles are extracted and marked as benchmark state vectors; N is a positive integer; each working condition parameter is assigned a standard value to obtain counterfactual condition vectors of each charging cycle;

[0024] inputting the continuous N reference state vectors and the corresponding N counterfactual condition vectors of the charging cycles into the trained first prediction model; the first prediction model outputs the predicted values of the state vectors of the latest M charging cycles; arranging the predicted values as the reference state vectors in chronological order to form the reference evolution trajectory.

[0025] As a preferred scheme of the dynamic threshold setting method for battery health grading management of the battery swap station described in the present application, the method for predicting the rebound confidence is as follows:

[0026] extracting the health evolution factors of the latest m continuous charging cycles and inputting them into the trained second prediction model, the second prediction model calculates and outputs the rebound confidence of the target battery; m is a positive integer;

[0027] The rebound confidence represents the confidence of the health indicator of the target battery appearing elastic rebound in the future n charging cycles, and n is a positive integer; the elastic rebound means that the health indicator of the target battery gradually increases to be greater than the health grading threshold.

[0028] As a preferred scheme of the dynamic threshold setting method for battery health grading management of the battery swap station described in the present application, the method for calculating the deviation confidence is as follows:

[0029] extracting the reference state vectors of the latest M charging cycles, and respectively combining them with the counterfactual condition vectors of the latest M charging cycles to form the reference evolution factor of each charging cycle;

[0030] Based on the reference evolution factor and the actual evolution factor, the weighted deviation degree of each charging cycle is calculated respectively;

[0031] The cumulative sum of the weighted deviation degrees of the latest M charging cycles is calculated and normalized to obtain the deviation confidence.

[0032] As a preferred scheme of the dynamic threshold setting method for battery health grading management of the battery swap station described in the present application, the health grading threshold of the target battery is adjusted based on the deviation confidence and the rebound confidence, specifically including: setting the minimum value of the health grading threshold; adjusting the health grading threshold based on the deviation confidence and the rebound confidence; the deviation confidence and the rebound confidence are negatively correlated with the health grading threshold; the adjusted health grading threshold is greater than or equal to the minimum value.

[0033] As a preferred scheme of the dynamic threshold setting method for battery health grading management of the battery swap station described in the present application, the method for calculating the weighted deviation degree of any charging cycle is as follows:

[0034] calculate the Euclidean distance between the state vector in the actual evolution factor and the reference state vector in the reference evolution factor as a first deviation indicator of the corresponding charging cycle;

[0035] calculate the Euclidean distance between the working condition vector in the actual evolution factor and the counterfactual condition vector in the reference evolution factor as a second deviation indicator of the corresponding charging cycle;

[0036] assign a deviation weight to the corresponding charging cycle based on the second deviation indicator, and the second deviation indicator is positively correlated with the deviation weight;

[0037] multiply the first deviation indicator by the deviation weight to obtain a weighted deviation degree of the corresponding charging cycle.

[0038] As a preferred scheme of the dynamic threshold setting method for battery health grading management of the battery swap station, if the health level of the target battery does not deteriorate after the health level is re-divided based on the adjusted health grading threshold, the health evolution trajectory of the target battery is continuously observed and updated, and the health grading threshold and the health level of the target battery are adjusted.

[0039] The continuous observation and update of the health evolution trajectory of the target battery specifically include:

[0040] The state parameters and working condition parameters of the target battery in the next n charging cycles are continuously observed, and the state vector and working condition vector of each charging cycle are constructed;

[0041] The counterfactual condition vector of the next n charging cycles is constructed, and the reference state vector in the next n charging cycles is re-predicted based on the first prediction model;

[0042] The weighted deviation degrees of the next n charging cycles are calculated based on the state vector, the working condition vector, the reference state vector and the counterfactual condition vector respectively, and the average deviation indicator of the target battery is obtained by averaging.

[0043] The health indicators of the target battery in the next n charging cycles are calculated based on the state vector.

[0044] As a preferred scheme of the dynamic threshold setting method for battery health grading management of the battery swap station, the adjustment of the health grading threshold and the health level of the target battery specifically includes:

[0045] a deviation index threshold value is set; if the average deviation index of the target battery is less than the deviation index threshold value, and the health index of the target battery gradually increases to be greater than the health classification threshold value before adjustment in the next n charging cycles, the health classification threshold value after adjustment and the health level are kept unchanged; otherwise, the health classification threshold value of the target battery is reset to the health classification threshold value before adjustment, and the health level of the target battery is lowered.

[0046] Compared with the prior art, the application has the following beneficial effects:

[0047] The application breaks through the limitation of traditional single parameter judgment, and multiple dimension parameters can accurately capture the health status of the battery surface and identify potential performance abnormalities. At the same time, the rebound confidence distinguishes between reversible fluctuations and irreversible degradation of the health index, avoiding the problems of misgrading healthy batteries and not timely processing degraded batteries.

[0048] Based on the deviation confidence, the application can quantify the difference between the actual working condition and the ideal working condition, offset the influence of environmental interference such as low temperature and high frequency fast charging on classification; combined with the health evolution track and continuous observation, the threshold value can also be corrected according to the attenuation characteristics of different use stages of the battery, solving the problem of rigid and lagging of traditional static threshold value. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor. Among them:

[0050] Figure 1 A flow chart of a dynamic threshold setting method for battery health classification management of a battery swap station provided by the application;

[0051] Figure 2 A method flow chart for calculating the weighted deviation degree of the charging cycle provided by the application. DETAILED DESCRIPTION

[0052] The technical solutions of the application will be described in detail below with the help of the drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solutions of the application, rather than limitations of the technical solutions of the application. In the case of no conflict, the technical features in the embodiments and the embodiments can be combined with each other.

[0053] This embodiment introduces a dynamic threshold setting method for battery health classification management of a battery swap station, which refers to Figure 1 The method comprises the following steps:

[0054] Acquire state parameters and working condition parameters of the target battery in each charging cycle, and construct a health evolution track of the target battery; set a health grading threshold for the target battery;

[0055] The charging cycle is a time period from the beginning of charging to the completion of charging of the target battery in each charging cycle;

[0056] The state parameters at least include SOH, SOH decay amount, average temperature, voltage variation range, and internal resistance of the target battery;

[0057] Among them, SOH is a battery health state parameter, which is a core parameter for describing the state of the battery; SOH decay amount is the change amount of SOH compared with the adjacent previous charging cycle, which is used to describe the degradation or rebound trend of the health state; the voltage variation range is the difference between the maximum voltage and the minimum voltage in the charging cycle; the average temperature, the voltage variation range, and the internal resistance are used to assist in describing the health state of the battery, reflecting the voltage stability, battery aging, and other abnormal conditions.

[0058] The working condition parameters at least include charging rate, SOC variation range, environmental temperature, cycle number, and charging cycle interval;

[0059] Among them, the charging rate can reflect whether there is a high-stress working condition such as fast charging; the SOC variation range is the difference between the SOC at the completion of charging and the SOC at the beginning of charging, which can reflect the shallow charging or deep charging mode; the charging cycle interval is the time difference between the starting time of the current charging cycle and the ending time of the adjacent previous charging cycle; the charging cycle interval, the environmental temperature, and the cycle number can reflect the external use environment and time dimension characteristics of the battery, and are used to represent the influence of the environment and the use of the battery on the evolution of the health state.

[0060] The health evolution track is represented by a health evolution sequence; the health evolution track of the target battery is constructed, specifically including:

[0061] For any charging cycle, each state parameter and each working condition parameter are normalized; each state parameter is arranged into a state vector corresponding to the charging cycle, and each working condition parameter is arranged into a working condition vector corresponding to the charging cycle; the state vector and the working condition vector of any charging cycle form a health evolution factor corresponding to the charging cycle; the health evolution factors of each charging cycle are arranged in time sequence into a health evolution sequence of the target battery.

[0062] After the above processing, the data of each charging cycle is abstracted into two vectors, wherein the state vector directly reflects the battery health of each charging cycle, and the working condition vector describes the external and self-working condition of the target battery in each charging cycle, and is used to analyze the interference strength of the health state evolution. The health evolution trajectory reflects the change of the health state of the battery over time under the influence of the corresponding working condition parameters.

[0063] In each charging cycle, the health index of the target battery is calculated based on the state parameters, and the health level of the target battery is divided in combination with the health grading threshold;

[0064] The method for calculating the health index is as follows: each state parameter is assigned a value to the health index, and the SOH is positively correlated with the health index, and the SOH decay amount, average temperature, voltage variation range, and internal resistance are negatively correlated with the health index.

[0065] For example, each state parameter is normalized, and then the weight value of each state parameter is set according to actual needs and weighted summation is performed to obtain a health index that comprehensively reflects the multi-dimensional state parameters, wherein the weight coefficients of the SOH decay amount, average temperature, voltage variation range, and internal resistance are multiplied by -1.

[0066] The method for dividing the health level of the target battery is as follows: if the health index of the target battery is greater than the health grading threshold, the health level remains unchanged, otherwise, the health level of the target battery is downgraded.

[0067] Alternatively, the health grading threshold is set based on the current health level of the target battery; a person skilled in the art can set the value of the health grading threshold based on industry experience or actual needs; for example, for the power battery used in an electric vehicle, when it is first put into use after leaving the factory, the health state is at the best level, and the health level can be set to the first health level. At this level, the battery can meet the high-load use scenarios such as full-load starting, continuous climbing, and high-frequency battery replacement, and has strong output capacity and thermal stability. According to experience, the health grading threshold is set for the first health level, and when it is detected that the health index of the battery is lower than the threshold, it means that it no longer has the above key capabilities, and its health level is downgraded to the second health level. The second health level can correspond to the situation that the battery performance has partially degraded but can still be used for general working conditions, for example, supporting medium-speed riding on flat roads, intermittent battery replacement operation, etc., and is suitable for use in low-power output modes. At the second health level, the health grading threshold is reset according to experience to judge whether the battery can continue to be used or needs to be further degraded. When it is detected that the health index is lower than the health grading threshold corresponding to the health level, the battery is no longer suitable for regular operation and needs to be downgraded to the third health level or marked for recycling and evaluation.

[0068] If the health level of the target battery deteriorates in any charging cycle, a reference evolution trajectory of the target battery health indicator is predicted based on the health evolution trajectory and a rebound confidence level;

[0069] The deterioration of the health level means that the health level of the target battery is reduced in the current charging cycle;

[0070] The reference evolution trajectory includes reference state vectors of the last M charging cycles; the reference evolution trajectory is predicted as follows:

[0071] Based on the health evolution trajectory, health evolution factors of the last M charging cycles are extracted and marked as actual evolution factors; M is a positive integer;

[0072] The state vectors of the last N charging cycles before the extraction of the M actual evolution factors are extracted and marked as reference state vectors; N is a positive integer; each working condition parameter is standardized and assigned to obtain an counterfactual condition vector of each charging cycle;

[0073] A person skilled in the art can set the standard value of the working condition parameter according to experience or actual needs. For example, an optional way of standardizing and assigning each working condition parameter is as follows: the charging rate is assigned as 1C; the SOC change range is assigned as 75%; the ambient temperature is assigned as 25℃; the cycle number remains unchanged corresponding to the cycle number of the charging cycle; the charging cycle interval is assigned as 20 hours; each standardized and assigned working condition parameter is normalized and arranged into a counterfactual condition vector corresponding to the charging cycle.

[0074] The continuous N reference state vectors and the counterfactual condition vectors corresponding to the N charging cycles are input into the trained first prediction model; the first prediction model outputs the predicted values of the state vectors of the last M charging cycles; the predicted values are arranged in time sequence as the reference state vectors to form the reference evolution trajectory.

[0075] In this embodiment, the reference evolution trajectory is a standard evolution trajectory of the state vector of the target battery under ideal use conditions predicted by the first prediction model, which can be compared with the measured state vector to assist in determining whether the system deviation of the health indicator calculation is caused by the difference in working conditions. Alternatively, the first prediction model is any one of a Transformer time series prediction model and a sequence-to-sequence long short-term memory network.

[0076] The rebound confidence level is predicted as follows:

[0077] The health evolution factors of the last m consecutive charging cycles are extracted and input into the trained second prediction model, and the second prediction model calculates and outputs the rebound confidence level of the target battery; m is a positive integer;

[0078] The rebound confidence represents a confidence of the health index of the target battery rebounding in the future n charging cycles, n being a positive integer; the rebounding represents that the health index of the target battery gradually increases to be greater than the health classification threshold.

[0079] In this embodiment, the second prediction model is a sliding window time series classification model based on a Transformer encoder or a prediction model combining a bidirectional long short-term memory network and a multilayer perceptron. The second prediction model is trained by observing whether the health index rebounds after the health level deteriorates in the historical data, and constructing a training data set containing continuous health evolution factors and a label of whether the rebounding occurs.

[0080] The deviation confidence of the health index of the target battery is calculated based on the reference evolution trajectory; and the health classification threshold of the target battery is adjusted based on the deviation confidence and the rebound confidence;

[0081] The method for calculating the deviation confidence is as follows:

[0082] The reference state vectors of the last M charging cycles are extracted, and each charging cycle is composed of a reference evolution factor of the last M charging cycles and an counterfactual condition vector of the last M charging cycles;

[0083] The weighted deviation degree of each charging cycle is calculated based on the reference evolution factor and the actual evolution factor;

[0084] The weighted deviation degrees of the last M charging cycles are calculated and normalized to obtain the deviation confidence;

[0085] The health classification threshold of the target battery is adjusted based on the deviation confidence and the rebound confidence, and the adjustment specifically includes:

[0086] The minimum value of the health classification threshold is set; the health classification threshold is adjusted based on the deviation confidence and the rebound confidence; the deviation confidence and the rebound confidence are negatively correlated with the health classification threshold; and the adjusted health classification threshold is greater than or equal to the minimum value. For example, the minimum value of the health classification threshold is set according to experience or actual demand, or the minimum health index of a battery of the same model as the target battery at the current health level is statistically obtained from historical data, and is used as the minimum value of the health classification threshold. Alternatively, a deviation confidence threshold and a rebound confidence threshold are set respectively; if the deviation confidence is greater than the deviation confidence threshold, the health classification threshold is reduced by a first preset proportion according to the size of the deviation confidence; and if the rebound confidence is greater than the rebound confidence threshold, the health classification threshold is reduced by a second specified proportion according to the size of the rebound confidence. The specific values of the first proportion and the second proportion can be set by experience or historical data analysis.

[0087] The embodiment can realize active correction of the deviation of the health grade deterioration judgment, and inhibit the influence of the working environment interference on the health grade classification result by adjusting the health classification threshold of the target battery based on the deviation confidence. The reversibility of the health index of the target battery can be distinguished from the irreversible deterioration by adjusting the health classification threshold of the target battery based on the rebound confidence, and the waste of battery resources caused by the over-conservative health grade classification can be avoided.

[0088] Referring to Figure 2 The method for calculating the weighted deviation degree of any charging cycle is as follows:

[0089] The Euclidean distance between the state vector in the actual evolution factor and the reference state vector in the reference evolution factor is calculated as the first deviation index of the corresponding charging cycle;

[0090] The Euclidean distance between the working condition vector in the actual evolution factor and the counterfactual condition vector in the reference evolution factor is calculated as the second deviation index of the corresponding charging cycle;

[0091] The second deviation index is used to assign a deviation weight to the corresponding charging cycle, and the second deviation index is positively correlated with the deviation weight;

[0092] The first deviation index is multiplied by the deviation weight to obtain the weighted deviation degree of the corresponding charging cycle.

[0093] In the embodiment, the deviation confidence represents the difference between the actual observed evolution trend of the state vector and the theoretical evolution trend under the standard working condition. The greater the difference, the greater the deviation confidence, that is, the higher the confidence that the state vector has a system error due to the use condition. When the second deviation index of a charging cycle is large, it indicates that the working condition corresponding to the cycle is more different from the standard working condition, and the confidence that the state vector has a system error due to the use condition is higher, and the corresponding deviation weight is assigned higher.

[0094] The health grade of the target battery is reclassified based on the adjusted health classification threshold.

[0095] If the health grade of the target battery does not deteriorate after the health grade is reclassified based on the adjusted health classification threshold, the health evolution trajectory of the target battery is continuously observed and updated, and the health classification threshold and the health grade of the target battery are adjusted.

[0096] The continuous observation and update of the health evolution trajectory of the target battery specifically include:

[0097] The state parameters and working condition parameters of the target battery in the next n charging cycles are continuously observed, and the state vector and working condition vector of each charging cycle are constructed;

[0098] constructing a counterfactual condition vector of the next n charging cycles, and re-forecasting the reference state vector in the next n charging cycles based on the first prediction model;

[0099] calculating a weighted deviation degree of the target battery in the next n charging cycles based on the state vector, the working condition vector, the reference state vector and the counterfactual condition vector respectively, and averaging to obtain an average deviation index of the target battery;

[0100] calculating a health index of the target battery in the next n charging cycles based on the state vector.

[0101] The adjusting the health classification threshold and the health level of the target battery specifically comprises:

[0102] setting a deviation index threshold; if the average deviation index of the target battery is less than the deviation index threshold, and the health index of the target battery gradually increases to be greater than the health classification threshold before adjustment in the next n charging cycles, the health classification threshold and the health level after adjustment are kept unchanged; otherwise, the health classification threshold of the target battery is reset to the health classification threshold before adjustment, and the health level of the target battery is downgraded.

[0103] In the embodiment, if the average deviation index of the target battery is less than the deviation index threshold, and the health index of the target battery gradually increases to be greater than the health classification threshold before adjustment in the next n charging cycles, it means that the health index of the target battery rebounds effectively, and the health level of the target battery does not need to be downgraded, which also means that the original health classification threshold is too aggressive, which is easy to cause misjudgment of the degradation of the health level, and the health classification threshold after adjustment needs to be kept; otherwise, the health level of the target battery actually degrades, the health classification threshold after adjustment is too loose, which is easy to cause missed detection of the degradation of the health level, and the health classification threshold after adjustment needs to be reset to the original value before adjustment.

[0104] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] The embodiments of the present application are described above in combination with the drawings, but the present application is not limited to the above specific embodiments, and the above specific embodiments are only illustrative, not restrictive. Those skilled in the art can make many forms without departing from the purpose and the scope of protection of the present application under the inspiration of the present application, which are all within the protection of the present application.

Claims

1.A method for dynamic threshold setting of battery health grading management of a battery swap station, characterized in that: The method comprises the following steps: obtaining state parameters and working condition parameters of the target battery in each charging cycle, and constructing a health evolution track of the target battery; setting a health grading threshold for the target battery; in each charging cycle, calculating a health index of the target battery based on the state parameters, and dividing a health grade for the target battery in combination with the health grading threshold; if the health grade of the target battery deteriorates in any charging cycle, predicting a reference evolution track of the health index of the target battery and a rebound confidence based on the health evolution track; calculating a deviation confidence of the health index of the target battery based on the reference evolution track, and adjusting the health grading threshold of the target battery based on the deviation confidence and the rebound confidence; redividing the health grade for the target battery based on the adjusted health grading threshold; the method for predicting the rebound confidence is as follows: extracting health evolution factors of the last m continuous charging cycles and inputting the trained second prediction model, and the second prediction model calculates and outputs the rebound confidence of the target battery; m is a positive integer; the rebound confidence represents a confidence that the health index of the target battery will rebound elastically in the future n charging cycles, and n is a positive integer; the elastic rebound means that the health index of the target battery gradually increases to be greater than the health grading threshold; the method for calculating the deviation confidence is as follows: extracting reference state vectors of the last M charging cycles, and respectively combining the reference state vectors with counterfactual condition vectors of the last M charging cycles to form reference evolution factors of each charging cycle; based on the reference evolution factors and actual evolution factors, calculating a weighted deviation degree of each charging cycle; calculating an accumulated sum of the weighted deviation degrees of the last M charging cycles and normalizing to obtain the deviation confidence; adjusting the health grading threshold of the target battery based on the deviation confidence and the rebound confidence, specifically including: setting a minimum value of the health grading threshold; adjusting the health grading threshold based on the deviation confidence and the rebound confidence; the deviation confidence and the rebound confidence are negatively correlated with the health grading threshold; the adjusted health grading threshold is greater than or equal to the minimum value. 2.The method of claim 1, wherein: The state parameters at least include SOH, SOH decay amount, average temperature, voltage variation range, internal resistance of the target battery; and the working condition parameters at least include charging rate, SOC variation range, environmental temperature, cycle number, charging cycle interval. The health evolution track is represented by a health evolution sequence; the health evolution track of the target battery is constructed specifically as follows: for any charging cycle, each state parameter and each working condition parameter are normalized respectively; arranging each state parameter into a state vector corresponding to the charging cycle, and arranging each working condition parameter into a working condition vector corresponding to the charging cycle; the state vector and the working condition vector of any charging cycle form a health evolution factor of the corresponding charging cycle; the health evolution factors of each charging cycle are arranged in time sequence into the health evolution sequence of the target battery. 3.The method of claim 2, wherein: The method for calculating the health index is as follows: each state parameter is assigned a value for the health index, and the SOH is positively correlated with the health index, and the SOH decay, the average temperature, the voltage variation range, and the internal resistance are negatively correlated with the health index; The method for dividing the health level of the target battery is as follows: if the health index of the target battery is greater than the health classification threshold, the health level remains unchanged, otherwise, the health level of the target battery is downgraded; The degradation of the health level indicates that the health level of the target battery is downgraded in the current charging cycle. 4.The method of claim 3, wherein: The reference evolution trajectory includes reference state vectors of the last M charging cycles; the method for predicting the reference evolution trajectory is as follows: The health evolution factors of the last M charging cycles are extracted based on the health evolution trajectory and are marked as actual evolution factors; M is a positive integer; The state vectors of the last N charging cycles before the extraction of the M actual evolution factors are extracted and are marked as benchmark state vectors; N is a positive integer; Each working condition parameter is assigned a standardized value to obtain counterfactual condition vectors of each charging cycle; The continuous N benchmark state vectors and the counterfactual condition vectors of the corresponding N charging cycles are input into the trained first prediction model; the first prediction model outputs predicted values of the state vectors of the last M charging cycles; the predicted values are arranged in chronological order as the reference state vectors to form the reference evolution trajectory. 5.The method of claim 4, wherein: The method for calculating the weighted deviation degree of any charging cycle is as follows: The Euclidean distance between the state vectors in the actual evolution factors and the reference state vectors in the reference evolution factors is calculated as a first deviation index of the corresponding charging cycle; The Euclidean distance between the working condition vectors in the actual evolution factors and the counterfactual condition vectors in the reference evolution factors is calculated as a second deviation index of the corresponding charging cycle; The deviation weight of the corresponding charging cycle is assigned based on the second deviation index, and the second deviation index is positively correlated with the deviation weight; The first deviation index and the deviation weight are multiplied to obtain the weighted deviation degree of the corresponding charging cycle. 6.The method of claim 5, wherein: If the health level of the target battery does not degrade after the health level is redivided based on the adjusted health classification threshold, the health evolution trajectory of the target battery is continuously observed and updated, and the health classification threshold and the health level of the target battery are adjusted; The continuous observation and update of the health evolution trajectory of the target battery specifically include: The state parameters and working condition parameters of the target battery in the future n charging cycles are continuously observed, and the state vectors and working condition vectors of each charging cycle are constructed; The counterfactual condition vectors of the future n charging cycles are constructed, and the reference state vectors in the future n charging cycles are predicted again based on the first prediction model; The weighted deviation degrees of the future n charging cycles are calculated based on the state vectors, the working condition vectors, the reference state vectors, and the counterfactual condition vectors, respectively, and the average is taken to obtain the average deviation index of the target battery; The health index of the target battery in the future n charging cycles is calculated based on the state vectors. 7.The method of claim 6, wherein: The adjustment of the health classification threshold and the health level of the target battery specifically includes: The deviation index threshold is set; if the average deviation index of the target battery is less than the deviation index threshold, and the health index of the target battery gradually increases to be greater than the health classification threshold before adjustment in the next n charging cycles, the adjusted health classification threshold and the health level are kept unchanged; otherwise, the health classification threshold of the target battery is reset to the health classification threshold before adjustment, and the health level of the target battery is lowered.

Images