A data-driven state of health estimation method suitable for different types of lithium batteries

Abstract

A data-driven state of health estimation method suitable for different types of lithium batteries relates to a battery state estimation method. The charging voltage interval for constructing the aging feature is determined, containing the highest peak voltage value of the IC curve; considering the partial charging condition of the battery, the aging feature is constructed by using the circular arc chord between two points on the V-Q curve, including the length, slope, Sin value of the included angle with the Y axis and the length of the two right angles of the right triangle formed by the circular arc chord; based on the online learning and model updating strategy, the SOH estimation model is updated. The voltage interval for constructing the aging feature is determined based on the change of the slope of the IC curve and the charging voltage curve, the effective aging feature is constructed based on the circular arc chord under the consideration of the partial charging condition, and the online updating strategy of the data-driven model is designed, which is more suitable for practical application and can obtain accurate and stable battery SOH estimation results under different types and different environmental conditions.

Description

Technical Field

[0001] This invention relates to a battery state estimation method, and more particularly to a data-driven state of health estimation method applicable to different types of lithium batteries, belonging to the field of lithium-ion battery performance evaluation technology. Background Technology

[0002] Lithium-ion batteries have become the primary power source for electric vehicles due to their excellent performance. However, with continuous use and aging, battery performance and health deteriorate, potentially leading to safety accidents. Therefore, accurately estimating the State of Health (SOH) is crucial for safe driving of electric vehicles and optimal battery performance. Currently, data-driven SOH estimation methods have attracted widespread attention because they do not require consideration of the complex internal reaction mechanisms of the battery. However, data-driven SOH estimation methods require constructing effective aging characteristics and using data-driven algorithms to establish an effective SOH estimation model.

[0003] Currently, the construction of aging characteristics mainly includes methods based on incremental capacity (IC) analysis and methods based on charging voltage curves (V-Q curves). Both require a complete charging process and a complete charging voltage range to construct aging characteristics. For IC analysis-based methods, aging characteristics are constructed by creating an IC curve, including peak values ​​and the area enclosed by those peaks. For example, Chinese invention patent CN111398837A selects multiple health factors based on the IC curve, including the distance between different peaks / valleys and the peak area; another example is Chinese invention patent CN109031153B, which uses the voltage corresponding to the main peak of the IC curve as an aging characteristic for SOH estimation. However, aging characteristic construction methods based on IC curves require appropriate filtering algorithms to process the curve and are susceptible to noise and other factors. Methods based on charging voltage curves do not require constructing an IC curve; instead, they construct aging characteristics such as constant current charging time. For example, Chinese invention patent CN114578237A uses constant current charging time as an aging characteristic for SOH evaluation, and Chinese invention patent CN111985156B selects five characteristics, including constant current charging time and constant voltage charging time, to evaluate SOH. However, in practical applications, the charging voltage range of batteries is random, which means that partial charging may occur. Therefore, the method of constructing aging characteristics based on the full charging voltage range is no longer applicable.

[0004] For partial charging considerations, some works construct aging characteristics using fixed charging voltage ranges, such as charging capacity at fixed voltage points within a certain range, or voltage sampling points with fixed sampling times. On the one hand, using fixed charging voltage ranges to construct aging characteristics reduces dependence on full charging conditions; on the other hand, effective aging characteristics can be constructed within specific charging voltage ranges, but the charging voltage range used to construct these characteristics is not random. Although the above methods consider partial charging conditions, they still fundamentally require a charging process within a specific voltage range. Other works point out that the shape of the voltage curve approximates a circular arc or a straight line, thus utilizing the property that an arc can map to a complete circle and a line segment can map to a complete straight line to consider the partial charging situation of the battery, and selecting the center of the circle, the slope of the line, etc., as aging characteristics based on the constructed circle or straight line. However, the coordinate axes constructed by the above methods will change under practical application conditions, and some aging characteristics constructed based on the shape of existing charging voltage curves are still not applicable.

[0005] Furthermore, the aforementioned Chinese invention patents are all based on fixed battery types and fixed application environments, and analyze and construct aging features based on aging data obtained from a complete cycle life. This is not conducive to constructing battery aging features under actual application conditions. On the one hand, it is difficult to obtain aging data from a complete cycle life for new batteries of different types and application environments; on the other hand, aging features constructed based on existing aging datasets are not applicable to new batteries. Moreover, even if effective aging features are extracted, the large differences in data range and distribution between input aging features for different types of batteries under different environmental conditions, including the potentially significant differences in the mapping relationship between input aging features and the target SOH, will lead to large estimation errors for both traditional data-driven algorithms such as backpropagation (BP) and decision trees, and deep learning algorithms such as convolutional neural networks (CNNs) and long short-term memory (LSTM) neural networks, if these data are not learned and trained. Therefore, traditional modeling methods are insufficient to obtain a stable and accurate SOH estimation result applicable to different types and application environments. In real-world applications, it is impossible to obtain complete aging cycle data for every type of battery under equal conditions, which poses a significant challenge to data-driven SOH estimation methods. Battery aging is a time series prediction problem. Without access to data for the entire lifespan, even after updating the data-driven model based on data obtained from time-series relationships, it is still impossible to make a completely accurate prediction of the future capacity degradation of the battery, because the future aging trend of the battery cannot be known in advance.

[0006] In summary, current aging feature construction methods still cannot effectively solve the problem of constructing aging features considering partial charging. Furthermore, current methods, based on complete cycle aging data, are limited to fixed battery types or specific environments. The input aging features differ significantly between different battery types and under different environmental conditions, leading to substantial SOH estimation errors when such data is unavailable for training and modeling. Therefore, after constructing reasonable aging features, designing a data-driven SOH estimation strategy is a pressing issue that needs to be addressed. Summary of the Invention

[0007] To address the shortcomings of the prior art, this invention provides a data-driven state of health estimation method applicable to different types of lithium batteries. It determines the voltage range for constructing aging characteristics based on the changes in the slope of the IC curve and the charging voltage curve. Considering partial charging conditions, it constructs effective aging characteristics based on circular arc chords and designs an online update strategy for the data-driven model. This method is more suitable for practical applications and can obtain accurate and stable SOH estimation results for batteries under different types and environmental conditions.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: a data-driven health status estimation method applicable to different types of lithium batteries, comprising the following steps:

[0009] Step 1: Determine the charging voltage range used to construct aging characteristics

[0010] The V-Q curve within the charging voltage range has an arc shape. Within the charging voltage range that includes the peak voltage value of the IC curve, the slope change of the V-Q curve is monitored. If the slope of the V-Q curve has a monotonic changing trend over a period of time, then this charging voltage range is determined to be used to construct aging characteristics.

[0011] Step 2: Considering the aging characteristics of the circular arc chord structure under partial battery charging conditions

[0012] Based on the determined charging voltage range used to construct aging characteristics, the aging characteristics are constructed using the circular arc chord between two points on the V-Q curve. Let the coordinates of the two points on the V-Q curve be A(x) and Q(q). A ,y A ) and D(x D ,y D The aging characteristics are constructed using the arc chord AD of the circle formed by the line connecting two points:

[0013] Feature 1: The length of the circular arc chord AD

[0014]

[0015] Feature 2: The slope of the circular arc chord AD

[0016]

[0017] Feature 3: The sin value of the angle between the circular arc chord AD and the Y-axis

[0018]

[0019] Feature 4: The sum of the lengths of the two legs of the right triangle formed by the circular arc chord AD

[0020] l AO+DO =|x D -x A |+|y D -y A |

[0021] The aging characteristics of the above structures are all related to the relative distance and are not affected by the partial charging situation on the coordinate axis under actual application conditions.

[0022] Step 3: SOH Estimation Based on Online Learning and Model Update Strategy

[0023] 1) Feature scaling processing based on aging features constructed in the initial loop

[0024] Let the aging characteristics constructed in the initial cycle be as follows: Where m is the dimension of the selected aging feature. This represents the extreme value of the aging characteristics throughout the entire aging cycle, and the aging characteristics constructed during the aging process are... Based on this, scaling is performed, that is:

[0025]

[0026] In the formula, F i N For the processed data-driven input aging features, N is the number of aging cycles;

[0027] 2) SOH Online Learning and Model Update

[0028] ① Establish the initial model

[0029] M0 = f(X0)

[0030] In the formula, M0 is the initial data-driven model, f is the data-driven algorithm, and X0 is the initial accumulated data;

[0031] ② Online learning and model updates: The initial model is updated by setting the accumulated data amount and the update time, as detailed below:

[0032] When the accumulated data reaches a set value or the update time is met, online learning and model updates of the data-driven model will be performed.

[0033] M k =f(ΔN) a )

[0034] No update is required if the accumulated data amount has not reached the set value and the update time has not been met.

[0035] M k =M k-1

[0036] In the formula, M k For the data-driven model updated at the current moment, ΔN a M represents the accumulated data at the current moment. k-1 The data-driven model updated from the previous moment;

[0037] By continuously updating the model using the online learning and model update strategy described above, the required SOH estimation results can be obtained.

[0038] Compared with existing technologies, the beneficial effects of this invention are as follows: First, this invention determines the voltage range for constructing aging features based on the changes in the slope of the IC curve and the charging voltage curve. Based on the actual charging voltage curve, it eliminates the need for complex data processing and algorithms, making it simpler and more effective, and facilitating online deployment in practical applications. The invention constructs aging features in the form of circular arcs based on the shape of the charging voltage curve, applicable to partial charging conditions, requiring no large amount of battery aging cycle data, and better aligning with the selection of aging features and SOH prediction under practical application conditions. Furthermore, it designs an online update strategy for the data-driven model, providing effective input aging features for data-driven SOH estimation methods, obtaining accurate and stable SOH estimation results. This is applicable to the construction of aging features and SOH estimation under different types and environmental conditions, making it more conducive to practical embedded applications and deployment. It also provides a foundation for accurate state of charge estimation and driving range prediction, contributing to the safe driving of electric vehicles. Attached Figure Description

[0039] Figure 1 This is a flowchart of the data-driven health status estimation method for lithium batteries according to the present invention;

[0040] Figure 2 The IC and V-Q curves are based on the CS2-36 battery structure from the University of Maryland.

[0041] Figure 3 It is a mathematical diagram of the right triangle formed by the circular arc chord connecting two points on the V-Q curve. Detailed Implementation

[0042] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0043] Reference Figures 1-3 As shown, a data-driven health state estimation method applicable to different types of lithium batteries is proposed, combining... Figure 1 As shown, it includes the following steps:

[0044] Step 1: Determine the charging voltage range used to construct aging characteristics

[0045] The charging voltage range used to construct aging characteristics differs and is not fixed for different types of batteries under different environmental conditions. Since the IC curve can convert the voltage plateau in the V-Q curve into the IC peak value, it can more sensitively reflect key information about the battery's internal state. Therefore, the charging voltage range used to construct aging characteristics must include the voltage position of the IC peak value.

[0046] Combination Figure 2 As shown, taking the CS2-36 battery based on the University of Maryland as an example, the constructed IC curve (ΔQ / ΔU ~ U) shows that a certain segment of ΔU in the IC curve and the area enclosed by the ΔQ / ΔU curve, denoted as ΔQ, has been proven to be an effective aging characteristic. Furthermore, other characteristics related to the IC curve during this ΔU period, such as the IC peak value, can also be used to effectively assess SOH. The IC peak value represents a specific voltage plateau of the battery, and is directly related to the phase transition of the electrode material, which exhibits high sensitivity to battery aging. As the battery ages, Li... + The continuous loss of electrode active material causes the IC peak value of the IC curve to change continuously.

[0047] Therefore, the following conclusions can be drawn:

[0048] 1. Based on incremental capacity analysis, the IC peak value is highly sensitive to battery aging. The IC peak value represents the voltage plateau inside the battery. Therefore, the V-Q curve within the IC peak value range is also highly sensitive to battery aging and can also construct effective aging characteristics within the charging voltage range.

[0049] 2. Based on incremental capacity analysis, the essence of determining the charging voltage range constructed by the effective aging characteristics is to find the voltage range in which the partial charging capacity ΔQ is strongly correlated with battery aging, while the fully charged Q is strongly correlated with battery aging.

[0050] Therefore, for new batteries of different types and under different environments in practical applications, a relatively complete charging cycle can be monitored first. Then, by determining the peak voltage value of the IC curve after constructing the IC curve, the charging voltage range for constructing aging characteristics can be selected. Considering that the start and end charging voltage ranges of the battery charging cycle are not fixed in practical applications, meaning that the battery experiences random partial charging, the determined charging voltage range for constructing aging characteristics must still be able to construct effective aging characteristics under certain partial charging conditions.

[0051] Combination Figure 2 As shown, taking the CS2-36 battery based on the University of Maryland as an example, the constructed V-Q curve shows that the shape of a certain segment of the V-Q curve resembles an arc or line segment on a circle or a straight line. Circles and straight lines have some desirable properties, such as the radius and the x and y coordinates of the circle formed by any arc segment on a circle being the same, and the slope and the intercepts of the line with the coordinate axes being constant for any line segment on a straight line.

[0052] Therefore, in the actual charging voltage range, the shape of a certain V-Q curve resembles a circle or an arc or line segment on a straight line. If a certain charging voltage range is a subset of the determined charging voltage range used to construct aging characteristics, then the aging characteristics can be constructed based on the complete circle or straight line mapped by that charging voltage range. Since the length of an arc on a circle can be equivalent to a straight line when it is very small compared to the circumference of the entire circle, regardless of whether the shape of a V-Q curve resembles an arc or a straight line, it can be uniformly considered as an arc on a certain circle. For any charging condition, theoretically, a V-Q curve resembling a line segment or arc can be found within a certain charging voltage range.

[0053] Based on the above analysis, taking into account partial charging, the V-Q curves within the charging voltage range used to construct aging characteristics need to meet certain shapes, and the charging voltage range needs to include the highest peak voltage value of the IC curve.

[0054] The slope of the V-Q curve is:

[0055]

[0056] In the formula, K V～Q Let Y be the slope of the V-Q curve under the conditions of ΔU and ΔQ. IC The ordinate value of the IC curve under the same ΔU condition. The reciprocal of.

[0057] Based on the above formula, in the actual constructed IC curve, the reciprocal of the ordinate value of the IC curve will not tend to 0 or infinity. Furthermore, within a certain voltage interval ΔU, the above formula guarantees that it is not zero, and the charging capacity ΔQ of the voltage interval ΔU is also not zero. Therefore, the slope of the V-Q curve is not infinite. The above analysis shows that regardless of the charging voltage range of the V-Q curve, as long as that segment of the V-Q curve has an arc shape, the slope of that segment of the V-Q curve will only continuously increase or decrease over the entire charging voltage range.

[0058] Therefore, based on the slope variation characteristics of the V-Q curve during charging, the charging voltage range containing the peak voltage value of the IC curve can be further determined. That is, within the charging voltage range containing the peak voltage value of the IC curve, the slope variation of the V-Q curve is monitored. If the slope of the V-Q curve shows a monotonic variation trend over a period of time, then this charging voltage range can be determined for constructing aging characteristics.

[0059] Step 2: Considering the aging characteristics of the circular arc chord structure under partial battery charging conditions

[0060] Based on the determined charging voltage range used to construct aging characteristics, the equation of a circle can be determined by any three points on the V-Q curve. Assuming the coordinates of the three points are (x1, y1), (x2, y2), and (x3, y3), and the radius of the circle is r, with the center at (x0, y0), then according to the equation of the circle:

[0061]

[0062] Solving the equations simultaneously yields the coordinates of the center and the radius of the circle, as shown below:

[0063]

[0064] in:

[0065]

[0066] Under practical application conditions, the vertical axis (U) of the coordinate axis formed by the V-Q curves remains unchanged. However, the coordinate axis (charging time) shifts due to changes in the charging initiation voltage. This results in different coordinates for points on the V-Q curve on different coordinate axis references, affecting the construction of aging characteristics. Therefore, considering partial charging scenarios under practical application conditions, the impact of coordinate axis shifts must be taken into account when constructing aging characteristics based on the shape of the V-Q curve. However, in reality, the relative distance between two points on the V-Q curve remains constant regardless of coordinate axis shifts. Therefore, aging characteristics can be constructed based on the circular arc chord between two points on the V-Q curve, combined with... Figure 3 As shown, let the coordinates of two points on the V-Q curve be A(x) and Q(y).A ,y A ) and D(x D ,y D The aging characteristics are constructed using the arc chord AD of the circle formed by the line connecting two points:

[0067] Feature 1: The length of the circular arc chord AD

[0068]

[0069] Feature 2: The slope of the circular arc chord AD

[0070]

[0071] Feature 3: The sin value of the angle between the circular arc chord AD and the Y-axis

[0072]

[0073] Feature 4: The sum of the lengths of the two legs of the right triangle formed by the circular arc chord AD

[0074] l AO+DO =|x D -x A |+|y D -y A |

[0075] The aging characteristics constructed above are all related to relative distance, and therefore are not affected by the partial charging condition on the coordinate axes under actual application conditions. Furthermore, aging characteristics can also be constructed using values ​​such as cosα, α, and β.

[0076] Step 3: SOH Estimation Based on Online Learning and Model Update Strategy

[0077] 1) Feature scaling processing based on aging features constructed in the initial loop

[0078] Based on the aging feature construction methods in steps one and two, the constructed aging features may vary significantly in range under different types and environmental conditions. When complete aging cycle data is unavailable, the data-driven model may exhibit substantial SOH estimation errors. Therefore, a feature scaling strategy based on the aging features constructed in the initial cycle is designed to reduce the impact of differences in the range of input aging features under different types and environmental conditions on the data-driven SOH estimation method.

[0079] Assuming the aging characteristics constructed in the initial cycle are as follows: Where m is the dimension of the selected aging characteristic. Since the constructed aging characteristic has a strong positive or negative correlation with SOH, it can be predicted that the constructed aging characteristic will also show an increasing or decreasing trend with battery aging, i.e. This represents the extreme value of the aging characteristics throughout the entire aging cycle. Therefore, to reduce the range differences in aging characteristics under different types and environmental conditions, the aging characteristics constructed during the aging process will... Based on this, scaling is performed, that is:

[0080]

[0081] In the formula, F i N The input aging features are processed data-driven, and N is the number of aging cycles.

[0082] Based on the feature data processing in the above formula, the input aging characteristics of batteries of different types and under different environmental conditions are data sequences that start from 1 and change over time. Compared with unprocessed methods or methods based on traditional data normalization or standardization, this greatly increases the stability of modeling and estimation.

[0083] 2) SOH Online Learning and Model Update

[0084] In practical applications, it is difficult to obtain complete cycle aging data for batteries of different types and under different environmental conditions at once. This means that traditional data-driven methods cannot obtain offline SOH estimation models applicable to different battery types and environmental conditions. However, under certain charging conditions, aging characteristics can be constructed and reference SOH calculated during actual battery use. Therefore, data-driven models can learn and update online based on the obtained real-time data to adapt to the aging pattern of the battery under the current condition. However, when the difference between the updated data and the data learned by the original model is large, the data-driven model may suffer from unstable estimation results. Therefore, the online learning and updating of SOH estimation models for batteries of different types and under different environmental conditions no longer relies on previously trained models, but focuses on the data pattern of the current battery type. That is, the designed online learning and model updating strategy for establishing online learning and model updating of SOH estimation for batteries of different types and under different environmental conditions is as follows:

[0085] ① Establish the initial model

[0086] M0 = f(X0)

[0087] In the formula, M0 is the initial data-driven model, f is the data-driven algorithm, and X0 is the initial accumulated data.

[0088] ② Online learning and model updates: The initial model is updated by setting the accumulated data amount and the update time, as detailed below:

[0089] When the accumulated data reaches a set value or the update time is met, online learning and model updates of the data-driven model will be performed.

[0090] M k =f(ΔN) a )

[0091] No update is required if the accumulated data amount has not reached the set value and the update time has not been met.

[0092] M k =M k-1

[0093] In the formula, M k For the data-driven model updated at the current moment, ΔN a M represents the accumulated data at the current moment. k-1 This is the data-driven model updated from the previous moment.

[0094] Battery capacity degradation is a time-series problem, meaning that the constructed aging features, including the calculated reference state of harm (SOH), are continuously acquired over time as the battery ages. Within a certain window, the statistical characteristics between the aging features and SOH constantly change, and the future aging trend of the battery is unknown, making it impossible to obtain a good SOH estimate based on the currently available data. However, the updated data-driven model can make good predictions about the aging state within the current range. By continuously updating the model according to the designed online learning and model update strategy, the desired SOH estimate can be obtained.

[0095] Example

[0096] This embodiment validates the method based on three different aging datasets. The basic information of the selected batteries is shown in Table 1.

[0097] Table 1 shows the battery dataset information used for verification.

[0098]

[0099] The charging voltage ranges for constructing aging characteristics for each battery based on this method are shown in Table 2.

[0100] Table 2. Charging voltage ranges for each battery to construct aging characteristics.

[0101]

[0102] The Pearson correlation coefficient values ​​between aging characteristics and capacity constructed based on this method are shown in Table 3.

[0103] Table 3 shows the correlation between the aging characteristics and capacity of each battery type.

[0104]

[0105] The results show that the constructed aging characteristics are reasonable and effective, and have a strong correlation with the SOH characterized by capacity, which can provide a basis for obtaining good SOH estimation results.

[0106] Based on the online learning and model update strategy designed in this method, the SOH estimation results for each battery in the next four steps are shown in Table 4:

[0107] Table 4. SOH estimation results for each battery

[0108]

[0109] The evaluation indicators for the SOH estimation results in the table are: mean absolute error (%) and maximum absolute error (%).

[0110] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0111] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A data-driven health status estimation method applicable to different types of lithium batteries, characterized in that: Includes the following steps: Step 1: Determine the charging voltage range used to construct aging characteristics Within the charging voltage range The curve has an arc shape, and monitoring is performed within the charging voltage range that includes the peak voltage value of the IC curve. The slope of the curve changes over a period of time. If the slope of the curve has a monotonically changing trend, then this charging voltage range is determined to be used to construct aging characteristics. Step 2: Considering the aging characteristics of the circular arc chord structure under partial battery charging conditions Based on determining the charging voltage range used to construct aging characteristics, utilizing The aging characteristics of the circular arc chord structure between two points on the curve are assumed to be... The coordinates of the two points on the curve are respectively and The aging characteristics are constructed by using the arc chord AD of the circle formed by the line connecting two points as the basis: Feature 1: The length of the circular arc chord AD Feature 2: The slope of the circular arc chord AD Feature 3: The sin value of the angle between the circular arc chord AD and the Y-axis Feature 4: The sum of the lengths of the two legs of the right triangle formed by the circular arc chord AD The aging characteristics of the above structures are all related to the relative distance and are not affected by the partial charging situation on the coordinate axis under actual application conditions. Step 3: SOH Estimation Based on Online Learning and Model Update Strategy 1) Feature scaling processing based on aging features constructed in the initial loop Let the aging characteristics constructed in the initial cycle be as follows: ,in It is the dimension of the selected aging feature, and the aging feature constructed during the aging process. Based on this, scaling is performed, that is: In the formula, To process the input aging features driven by the post-processing data, The number of aging cycles; 2) SOH Online Learning and Model Update ① Establish the initial model In the formula, As the initial data-driven model, Data-driven algorithms This represents the initial accumulated aging characteristic data; ② Online learning and model updates: The initial model is updated by setting the accumulated data amount and the update time, as detailed below: When the accumulated data reaches a set value or the update time is met, online learning and model updates of the data-driven model will be performed. No update is required if the accumulated data amount has not reached the set value and the update time has not been met. In the formula, The updated data-driven model at the current moment. The data accumulated at the current moment. The data-driven model is the one updated from the previous moment; By continuously updating the model using the online learning and model update strategy described above, the required SOH estimation results can be obtained.

2. The data-driven health status estimation method applicable to different types of lithium batteries according to claim 1, characterized in that: In step one The slope of the curve is: In the formula, In order to be in and Under the conditions The slope of the curve, In the same Under the condition of IC curve ordinate value The reciprocal of.

Images