A multi-feature lithium battery life prediction method based on cycle operation data

Abstract

The present application discloses a multi-feature lithium battery life prediction method based on cyclic operation data, which relates to the field of energy storage of new energy power systems. The method includes: obtaining cyclic operation data of multiple lithium batteries; constructing and extracting multiple aging feature sequences of each lithium battery according to the cyclic operation data of each lithium battery; and then constructing an aging feature matrix of each lithium battery; after feature fusion, obtaining a fused feature representation, and training a trained battery aging state prediction model; substituting the aging feature matrix of the lithium battery to be predicted into the model to obtain the life prediction result of the lithium battery to be predicted. The present application simplifies the data collection process and improves the data representativeness by obtaining the cyclic operation data of lithium batteries under actual working conditions, and fuses multiple aging features of lithium batteries through a self-attention mechanism, thereby enhancing the representation ability of the prediction model and improving the accuracy of lithium battery life prediction.

Description

Technical Field

[0001] The present application relates to the field of energy storage of new energy power systems, and in particular to a multi-feature lithium battery life prediction method based on cyclic operation data. Background Art

[0002] With the widespread application of lithium batteries in electric vehicles, energy storage systems, consumer electronics and other fields, their performance degradation and life prediction issues have gradually become key technical challenges. Lithium batteries will gradually age during repeated charging and discharging, which is manifested as capacity decay and increased internal resistance, which in turn affects the reliability and safety of the overall battery system. Therefore, accurately extracting the aging characteristics of lithium batteries is of great significance for understanding the battery's state of health (SOH) and predicting its remaining useful life (RUL).

[0003] The current aging feature extraction methods mainly rely on large-scale experimental data or complex physical and chemical models. Data collection is time-consuming and laborious, and the data is difficult to adapt to the changes in aging characteristics under the diverse application scenarios of lithium batteries. Especially in the analysis of cycle operation data, the accuracy and applicability of traditional methods are limited due to the uncertainty of battery status, the influence of environmental factors and differences in usage conditions. In addition, the current lithium battery life prediction process has problems such as low representativeness of aging characteristics and inaccurate prediction results. Summary of the invention

[0004] The purpose of this application is to provide a multi-feature lithium battery life prediction method based on cyclic operation data, which can simplify the data collection process, improve data representativeness, and at the same time enhance the characterization capability of the prediction model and improve the accuracy of lithium battery life prediction.

[0005] To achieve the above objectives, this application provides the following solutions.

[0006] In a first aspect, the present application provides a multi-feature lithium battery life prediction method based on cyclic operation data, comprising: obtaining cyclic operation data of multiple lithium batteries; the cyclic operation data is the operation data of the lithium battery under actual working conditions; the cyclic operation data includes at least: voltage, temperature, charge change value and cycle period at different sampling times; calculating the health status of each lithium battery in each cycle according to the cyclic operation data of each lithium battery, and constructing a health status sequence for each lithium battery; extracting multiple aging feature sequences of each lithium battery according to the cyclic operation data of each lithium battery; the aging features include at least: a feature sequence of the coefficient of variation of the charge change value, a feature sequence of the Euclidean distance of the charge change value , two or more of the characteristic sequence of capacity increment curve peak height, the characteristic sequence of capacity increment curve peak corresponding voltage, the characteristic sequence of temperature increment curve peak height and the characteristic sequence of temperature increment curve peak corresponding voltage; based on the health status sequence of each lithium battery, and, multiple aging feature sequences, construct an aging feature matrix for each lithium battery; use the self-attention mechanism to fuse the features of the aging feature matrix to obtain the fused feature representation; train the model based on all the fused feature representations to obtain the trained battery aging state prediction model; obtain the aging feature matrix of the lithium battery to be predicted, and substitute it into the trained battery aging state prediction model to obtain the life prediction result of the lithium battery to be predicted.

[0007] According to the specific embodiments provided in this application, this application discloses the following technical effects.

[0008] This application reduces the time and resource consumption of data collection in a laboratory environment by obtaining the cyclic operation data of lithium batteries under actual working conditions, simplifies the data collection process, and the cyclic operation data under actual working conditions can accept the errors caused by the environment and data collection accuracy, thereby improving the data representativeness and the robustness of feature extraction; this application uses a comprehensive prediction of multiple aging features, assigns different weights to each feature, and uses a self-attention mechanism to fuse multiple aging features of lithium batteries, thereby enhancing the characterization ability of the prediction model and improving the accuracy of lithium battery life prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0010] Figure 1 A schematic flow chart of a multi-feature lithium battery life prediction method based on cyclic operation data provided in an embodiment of the present application.

[0011] Figure 2 A curve diagram of discharge capacity, voltage and temperature for a cyclic charge and discharge segment provided in an embodiment of the present application.

[0012] Figure 3 The embodiments of the present application provide a scatter plot of the relationship between various aging characteristics and health status. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0014] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0015] Currently, the traditional battery aging feature data extraction method mainly has the following problems.

[0016] 1) High sensitivity to environment and usage conditions: Existing feature extraction methods mostly rely on stable laboratory environments and high-precision data acquisition equipment. In actual battery usage environments, lithium battery characteristics are often affected by environmental changes and errors in data acquisition equipment, and traditional methods have low effectiveness in extracting features.

[0017] 2) Low feature extraction accuracy, difficult to reflect the true aging state: The battery aging process includes different stages, such as slow decay period, rapid decay period, etc. Traditional feature extraction methods are less applicable in different stages. The aging feature data extracted by many methods are difficult to accurately reflect the true state of battery aging, especially when the battery enters the rapid decay period, feature extraction and monitoring are difficult to adapt to the accelerated decay state changes, which in turn affects the accuracy of health status assessment and life prediction.

[0018] 3) Lack of a universal method applicable to multiple battery types: Different types of lithium batteries, such as lithium iron phosphate batteries and ternary lithium batteries, have significantly different aging mechanisms, and traditional methods are usually only applicable to specific types of batteries. The lack of a universal feature extraction method to adapt to lithium batteries of different material systems has limited its promotion and application in various application scenarios.

[0019] 4) Dependence on specific operating conditions: Traditional feature extraction methods are limited to relatively strict specific operating conditions, such as full charge and discharge data segments, or fixed charge and discharge current rates, etc., resulting in sparse feature sampling points and unable to cover the entire life cycle of battery aging.

[0020] like Figure 1 As shown, this embodiment provides a multi-feature lithium battery life prediction method based on cycle operation data, including the following steps.

[0021] S1. Acquire cyclic operation data of multiple lithium batteries; the cyclic operation data is the operation data of the lithium batteries under actual working conditions; the cyclic operation data at least includes: voltage, temperature, power change value and cycle period at different sampling times.

[0022] Optionally, the lithium battery is at least one or more of a lithium iron phosphate battery and a ternary lithium battery.

[0023] Optionally, the cycle operation data also includes a battery remaining capacity (SOC).

[0024] Optionally, the power change value is a charge amount or a discharge amount.

[0025] S2. Calculate the health status of each lithium battery in each cycle according to the cycle operation data of each lithium battery, and construct a health status sequence for each lithium battery.

[0026] Furthermore, the calculation formula of the health status sequence of the lithium battery is as follows:

[0027] soh=(soh1,soh2,…,soh p ).

[0028]

[0029] Among them, soh p It indicates the health status of the lithium battery at the pth cycle, Q0 indicates the nominal capacity of the lithium battery, They represent the start and end time of the charge and discharge of the pth cycle, Respectively represent the pth cycle at time and The change value of the power, Respectively represent the pth cycle at time and The remaining power of the lithium battery, p∈{1,2,…,P}, P represents the number of total cycle cycles of the lithium battery.

[0030] S3. Extract multiple aging feature sequences of each lithium battery based on the cyclic operation data of each lithium battery; the aging features include at least two or more of: a feature sequence of the coefficient of variation of the charge change value, a feature sequence of the Euclidean distance of the charge change value, a feature sequence of the peak height of the capacity increment curve, a feature sequence of the voltage corresponding to the peak of the capacity increment curve, a feature sequence of the peak height of the temperature increment curve, and a feature sequence of the voltage corresponding to the peak of the temperature increment curve.

[0031] Step S3 specifically includes the following steps.

[0032] S31. Figure 2 As shown, the operation data of the voltage within the preset voltage range in the cycle operation data of each lithium battery is intercepted as the cycle charge and discharge segment of each lithium battery.

[0033] Optionally, the preset voltage interval is [V lower ,V upper ], the feature extraction window length is w.

[0034] S32. Based on the different aging characteristics in the cyclic charge and discharge segments of each lithium battery, multiple aging characteristic sequences of each lithium battery are constructed.

[0035] S4. Based on the health status sequence of each lithium battery and multiple aging feature sequences, an aging feature matrix of each lithium battery is constructed.

[0036] Optionally, the formula for the aging characteristic matrix is ​​as follows.

[0037] D=[(f 1 ) T ,(f 2 ) T ,(f 3 ) T ,(f 4 ) T ,(f 5 ) T ,(f 6 ) T ,(soh) T ].

[0038] in,(·) T represents transpose, D is a matrix with P rows and 7 columns, and the p-th row of the matrix D represents the aging characteristics and health status of the p-th cycle. Figure 3 A scatter plot showing the relationship between various battery aging characteristics and SOH.

[0039] Step S4 specifically includes the following steps.

[0040] S41. Intercepting the operation data of the cycle operation data of each lithium battery, where the voltage is within a preset voltage range, as the cycle charge and discharge segment of each lithium battery.

[0041] S42. Based on the different aging characteristics in the cyclic charge and discharge segments of each lithium battery, multiple aging characteristic sequences of each lithium battery are constructed.

[0042] S5. Use the self-attention mechanism to fuse the aging feature matrix and obtain the fused feature representation.

[0043] Step S5 specifically includes the following steps.

[0044] S51. Input the aging feature matrix of all lithium batteries into the neural network model to obtain the hidden layer representation of the aging characteristics of all lithium batteries.

[0045] S52. Calculate the dot product of the hidden layer representation of each aging feature of each lithium battery and the corresponding health state to obtain a similarity score of each aging feature of each lithium battery.

[0046] S53. The similarity scores of all aging features of each lithium battery are normalized by the Softmax function to obtain the attention weights of each aging feature, and the weights are added together to obtain the fused feature representation.

[0047] S6. Train the model based on all fused feature representations to obtain a trained model for predicting the battery aging status.

[0048] Optionally, the battery aging state prediction model can be constructed based on a neural network model.

[0049] In actual application, the aging feature matrix of lithium batteries can be used for battery aging state estimation and battery life prediction model training. For example, the first 6 columns of the aging feature matrix of lithium batteries are used as input and the last column as output. The Attention mechanism is introduced to automatically extract the most relevant information about the battery aging state from a variety of aging features, assign weights to different aging features, and perform feature fusion to improve prediction accuracy. Among them, the battery aging feature matrix can be used for battery aging state estimation and battery life prediction model training. Different from the prior art, which usually only predicts the state for a single aging feature, the comprehensive prediction of multiple aging features is used to allow the model to learn the aging law of the battery in different state spaces. By assigning different weights to each feature, the expression ability of the model is increased and the prediction accuracy is improved.

[0050] S7. Obtain the aging feature matrix of the lithium battery to be predicted, and substitute it into the trained battery aging state prediction model to obtain the life prediction result of the lithium battery to be predicted.

[0051] Furthermore, the construction process of different aging feature sequences is as follows.

[0052] 1) The construction process of the characteristic sequence of the coefficient of variation of the electric quantity change value is as follows.

[0053] The following formula is used to extract the power change value of each lithium battery within the preset time window at different times:

[0054]

[0055] in, Indicates time is the power change value within the time window at the start time, For the moment The amount of electricity, For the moment The amount of electricity, is the i-th sampling time of the p-th cycle, is the i+wth sampling moment of the pth cycle, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle, and w is the window length.

[0056] According to the charge change value of each lithium battery within the preset time window at different times, a sequence of charge change values ​​of each cycle of the lithium battery is constructed.

[0057]

[0058] Where m is the total number of sampling moments in each cycle.

[0059] According to the charge change value of each lithium battery within the preset time window, the following formula is used to calculate the mean of the charge change value sequence of different cycle charge and discharge segments of the lithium battery.

[0060]

[0061] in, It is the mean value of the charge change value sequence of the p-th cycle charge and discharge segment of the lithium battery.

[0062] According to the mean of the charge change value sequence of different cycle charge and discharge segments of the lithium battery, the characteristic sequence of the coefficient of variation of the charge change value of each lithium battery is calculated using the following formula.

[0063]

[0064] Among them, f 1 is the characteristic sequence of the coefficient of variation of the power change value, The coefficient of variation characteristics of the power change value in the pth cycle.

[0065] 2) The process of constructing the Euclidean distance feature sequence of the power change value is as follows.

[0066] The following formula is used to extract the power change value of each lithium battery within the preset time window at different times:

[0067]

[0068] in, Indicates time is the power change value within the time window at the start time, For the moment The amount of electricity, For the moment The amount of electricity, is the i-th sampling time of the p-th cycle, is the i+wth sampling moment of the pth cycle, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle, and w is the window length.

[0069] According to the charge change value of each lithium battery within the preset time window at different times, a sequence of charge change values ​​of each cycle of the lithium battery is constructed.

[0070]

[0071] Where m is the total number of sampling moments in each cycle.

[0072] According to the sequence of power change values ​​of each cycle of the lithium battery, the Euclidean distance feature sequence of the power change value of the lithium battery is calculated using the following formula.

[0073]

[0074] Among them, f 2 Represents the Euclidean distance feature sequence of the power change value of the lithium battery, It represents the Euclidean distance of the charge change value of the lithium battery in the pth cycle.

[0075] 3) The construction of the peak height characteristic sequence of the capacity increment curve is as follows.

[0076] According to the voltage sequence of different cycle charge and discharge segments, the charge change value sequence is linearly interpolated at intervals of 0.001V, and the interpolation sequence of the charge change value of different cycle charge and discharge segments of the lithium battery is calculated using the following formula as follows.

[0077]

[0078] in, Interpolation sequence of charge change values ​​of lithium battery in different cycle charge and discharge segments, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, and P is the number of cycles of the total lithium battery cycle.

[0079] According to the interpolation sequence of the charge change values ​​of different cycle charge and discharge segments of the lithium battery, the following formula is used to construct the peak height characteristic sequence of the capacity increment curve of the lithium battery.

[0080]

[0081] Among them, f 3 is the peak height characteristic sequence of the capacity increment curve of the lithium battery, is the peak height of the capacity increment curve of the pth cycle, min(·) is the minimum function, It is the sequence of charge change values ​​of the p-th cycle charge and discharge segment of the lithium battery.

[0082] 4) The process of constructing the voltage characteristic sequence corresponding to the peak of the capacity increment curve is as follows.

[0083] According to the voltage sequence of different cycle charge and discharge segments, the charge change value sequence is linearly interpolated at intervals of 0.001V, and the interpolation sequence of the charge change value of different cycle charge and discharge segments of the lithium battery is calculated using the following formula.

[0084]

[0085] in, Interpolation sequence of charge change values ​​of lithium battery in different cycle charge and discharge segments, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, and P is the number of cycles of the total lithium battery cycle.

[0086] According to the interpolation sequence of the charge change values ​​of different cycle charge and discharge segments of the lithium battery, the voltage characteristic sequence corresponding to the peak of the capacity increment curve of the lithium battery is calculated using the following formula.

[0087]

[0088]

[0089] Among them, f 4 The voltage characteristic sequence corresponding to the peak of the capacity increment curve of the lithium battery, is the voltage corresponding to the peak of the capacity increment curve of the pth cycle of the lithium battery, and argmin(·) is the index result of the minimum function.

[0090] 5) The process of constructing the characteristic sequence of the peak height of the temperature increment curve is as follows.

[0091] According to the voltage sequence of different cycle charge and discharge segments, the temperature sequence is linearly interpolated at intervals of 0.001V, and the temperature interpolation sequence of different cycle charge and discharge segments of the lithium battery is calculated using the following formula as follows.

[0092]

[0093] in, Temperature interpolation sequence of different cycle charge and discharge segments of lithium battery, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, and P is the number of cycles of the total lithium battery cycle.

[0094] According to the temperature interpolation sequence of different cycle charge and discharge segments of the lithium battery, the following formula is used to construct the peak height characteristic sequence of the temperature increment curve of the lithium battery.

[0095]

[0096] Among them, f 5 is the peak height characteristic sequence of the temperature increment curve of the lithium battery, is the peak height of the temperature increment curve of the pth cycle, min(·) is the minimum value function, It is the temperature variation sequence of the p-th cycle charge and discharge segment of the lithium battery.

[0097] 6) The process of constructing the voltage characteristic sequence corresponding to the peak value of the temperature increment curve is as follows.

[0098] According to the voltage sequence of different cycle charge and discharge segments, the temperature sequence is linearly interpolated at intervals of 0.001V, and the temperature interpolation sequence of different cycle charge and discharge segments of the lithium battery is calculated using the following formula:

[0099]

[0100]

[0101] in, Temperature interpolation sequence of different cycle charge and discharge segments of lithium battery, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, and P is the number of cycles of the total lithium battery cycle.

[0102]

[0103] Among them, f 6 is the voltage characteristic sequence corresponding to the peak value of the temperature increment curve, is the voltage corresponding to the peak value of the temperature increment curve in the pth cycle, and argmin(·) is the index result of the minimum function.

[0104] This application reduces the time and resource consumption of data collection in a laboratory environment by obtaining the cycle operation data of lithium batteries under actual working conditions, simplifies the data collection process, and the cycle operation data under actual working conditions can accept the errors caused by the environment and data collection accuracy, thereby improving the data representativeness and robustness of feature extraction; this application uses a comprehensive prediction of multiple aging features, assigns different weights to each feature, and uses a self-attention mechanism to fuse multiple aging features of lithium batteries, thereby enhancing the characterization ability of the prediction model and improving the accuracy of lithium battery life prediction. This application is applicable to lithium batteries of different material systems, and improves the ability to promote and apply in a variety of application scenarios. This application allows the use of fragmented data in the charge and discharge cycle for feature extraction, does not rely on specific battery usage conditions, and improves the coverage of feature extraction points for the entire life cycle of the battery.

[0105] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A multi-feature lithium battery life prediction method based on cycle operation data, characterized in that: The multi-feature lithium battery life prediction method based on cycle operation data includes: Acquire cyclic operation data of multiple lithium batteries; the cyclic operation data is the operation data of the lithium batteries under actual working conditions; the cyclic operation data at least includes: voltage, temperature, power change value and cycle period at different sampling times; Calculate the health status of each lithium battery in each cycle according to the cycle operation data of each lithium battery, and construct a health status sequence of each lithium battery; Extracting multiple aging feature sequences of each lithium battery according to the cycle operation data of each lithium battery; the aging features include at least two or more of: a feature sequence of a coefficient of variation of a power change value, a feature sequence of a Euclidean distance of a power change value, a feature sequence of a peak height of a capacity increment curve, a feature sequence of a voltage corresponding to a peak of a capacity increment curve, a feature sequence of a peak height of a temperature increment curve, and a feature sequence of a voltage corresponding to a peak of a temperature increment curve; Based on the health state sequence of each lithium battery and multiple aging feature sequences, an aging feature matrix of each lithium battery is constructed; The self-attention mechanism is used to fuse the aging feature matrix to obtain the fused feature representation; The model is trained based on all fused feature representations to obtain a trained model for predicting battery aging status; The aging feature matrix of the lithium battery to be predicted is obtained, and substituted into the trained battery aging state prediction model to obtain the life prediction result of the lithium battery to be predicted.

2. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1 is characterized in that: According to the cycle operation data of each lithium battery, multiple aging feature sequences of each lithium battery are extracted, including: Intercepting the operation data of the voltage within the preset voltage range from the cycle operation data of each lithium battery as the cycle charge and discharge segment of each lithium battery; According to the different aging characteristics in the cyclic charge and discharge segments of each lithium battery, multiple aging characteristic sequences of each lithium battery are constructed.

3. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1 is characterized in that: The self-attention mechanism is used to fuse the aging feature matrix to obtain the fused feature representation, which includes: Input the aging feature matrix of all lithium batteries into the neural network model to obtain the hidden layer representation of the aging features of all lithium batteries; Calculate the dot product of the hidden layer representation of each aging feature of each lithium battery and the corresponding health state to obtain the similarity score of each aging feature of each lithium battery; The similarity scores of all aging features of each lithium battery are normalized by the Softmax function to obtain the attention weights of each aging feature, which are summed up to obtain the fused feature representation.

4. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1 is characterized in that: The calculation formula of the health status sequence of the lithium battery is as follows: soh=(soh1,soh2,…,soh p ); Among them, soh p It indicates the health status of the lithium battery at the oth cycle, Q0 indicates the nominal capacity of the lithium battery, They represent the start and end time of the charge and discharge of the pth cycle, Respectively represent the pth cycle at time and The change value of the power, Respectively represent the pth cycle at time and The remaining power of the lithium battery, p∈{1,2,…,P}, P represents the number of total cycle cycles of the lithium battery.

5. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1, characterized in that: The construction process of the characteristic sequence of the coefficient of variation of the power change value is as follows: The following formula is used to extract the power change value of each lithium battery within the preset time window at different times: in, Indicates time is the power change value within the time window at the start time, For the moment of electricity, For the moment of electricity, is the i-th sampling time of the p-th cycle, is the i+wth sampling time of the pth cycle, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle, and w is the window length; According to the power change value of each lithium battery in the preset time window at different times, the power change value sequence of each cycle of the lithium battery is constructed: Where m is the total number of sampling moments in each cycle; According to the power change value within the preset time window of each lithium battery, the mean value of the power change value sequence of different cycle charge and discharge segments of the lithium battery is calculated using the following formula: in, is the mean value of the charge change value sequence of the p-th cycle charge and discharge segment of the lithium battery; According to the mean of the power change value sequence of different cycle charge and discharge segments of the lithium battery, the characteristic sequence of the coefficient of variation of the power change value of each lithium battery is calculated using the following formula: Among them, f 1 is the characteristic sequence of the coefficient of variation of the power change value, The coefficient of variation characteristics of the power change value in the pth cycle.

6. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1, characterized in that: The construction process of the Euclidean distance feature sequence of the power change value is as follows: The following formula is used to extract the power change value of each lithium battery within the preset time window at different times: in, Indicates time is the power change value within the time window at the start time, For the moment of electricity, For the moment of electricity, is the i-th sampling time of the p-th cycle, is the i+wth sampling time of the pth cycle, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle, and w is the window length; According to the power change value of each lithium battery in the preset time window at different times, the power change value sequence of each cycle of the lithium battery is constructed: Where m is the total number of sampling moments in each cycle; According to the sequence of power change values ​​of each cycle of the lithium battery, the Euclidean distance feature sequence of the power change value of the lithium battery is calculated using the following formula: Among them, f 2 Represents the Euclidean distance feature sequence of the lithium battery’s power change value, It represents the Euclidean distance of the charge change value of the lithium battery in the pth cycle.

7. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1, characterized in that: The construction of the peak height characteristic sequence of the capacity increment curve is as follows: According to the voltage sequence of different cycle charge and discharge segments, the charge change value sequence is linearly interpolated at intervals of 0.001V. The interpolation sequence of charge change values ​​of different cycle charge and discharge segments of the lithium battery is calculated using the following formula as follows: in, Interpolation sequence of charge change values ​​of lithium battery in different cycle charge and discharge segments, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle; According to the interpolation sequence of the charge change value of different cycle charge and discharge segments of the lithium battery, the following formula is used to construct the peak height characteristic sequence of the capacity increment curve of the lithium battery: Among them, f 3 is the peak height characteristic sequence of the capacity increment curve of the lithium battery, is the peak height of the capacity increment curve of the pth cycle, min(·) is the minimum function, It is the sequence of charge change values ​​of the p-th cycle charge and discharge segment of the lithium battery.

8. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1, characterized in that: The construction process of the voltage characteristic sequence corresponding to the peak of the capacity increment curve is as follows: According to the voltage sequence of different cycle charge and discharge segments, the charge change value sequence is linearly interpolated at intervals of 0.001V, and the interpolation sequence of charge change values ​​of different cycle charge and discharge segments of the lithium battery is calculated using the following formula: in, Interpolation sequence of charge change values ​​of lithium battery in different cycle charge and discharge segments, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle; According to the interpolation sequence of the charge change value of different cycle charge and discharge segments of the lithium battery, the voltage characteristic sequence corresponding to the peak of the capacity increment curve of the lithium battery is calculated using the following formula: Among them, f 4 The voltage characteristic sequence corresponding to the peak of the capacity increment curve of the lithium battery, is the voltage corresponding to the peak of the capacity increment curve of the pth cycle of the lithium battery, and argmin(·) is the index result of the minimum function.

9. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1, characterized in that: The construction process of the temperature increment curve peak height characteristic sequence is as follows: According to the voltage sequence of different cycle charge and discharge segments, the temperature sequence is linearly interpolated at intervals of 0.001V, and the temperature interpolation sequence of different cycle charge and discharge segments of the lithium battery is calculated using the following formula: in, Temperature interpolation sequence of different cycle charge and discharge segments of lithium battery, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle; According to the temperature interpolation sequence of different cycle charge and discharge segments of lithium batteries, the following formula is used to construct the peak height characteristic sequence of the temperature increment curve of lithium batteries: Among them, f 5 is the peak height characteristic sequence of the temperature increment curve of the lithium battery, is the peak height of the temperature increment curve of the pth cycle, min(·) is the minimum value function, It is the temperature variation sequence of the p-th cycle charge and discharge segment of the lithium battery.

10. The multi-feature lithium battery life prediction method based on cycle operation data according to claim 1, characterized in that: The construction process of the voltage characteristic sequence corresponding to the peak value of the temperature increment curve is as follows: According to the voltage sequence of different cycle charge and discharge segments, the temperature sequence is linearly interpolated at intervals of 0.001V, and the temperature interpolation sequence of different cycle charge and discharge segments of the lithium battery is calculated using the following formula: in, Temperature interpolation sequence of different cycle charge and discharge segments of lithium battery, V lower is the minimum value of the preset voltage range, V upper is the maximum value of the preset voltage range, p∈{1,2,…,P}, P is the number of cycles of the total lithium battery cycle; Among them, f 6 is the voltage characteristic sequence corresponding to the peak value of the temperature increment curve, is the voltage corresponding to the peak value of the temperature increment curve in the pth cycle, and argmin(·) is the index result of the minimum function.

Images