Method for improving yield of battery foil through data modeling based on abnormal value elimination

Through outlier value removal and data normalization processing, the problem of large modeling errors in battery foil production in traditional methods is solved, the yield rate of battery foil is improved, and more accurate production process optimization is achieved.

CN118939952APending Publication Date: 2024-11-12LUOYANG INST OF SCI & TECH
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202410940002.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

In battery foil production, traditional data-driven methods have large modeling errors due to the unprocessed or simple processing of the original data, which cannot accurately simulate the production principle, making it difficult to improve the yield rate of battery foil.

Method used

Using data modeling method based on outlier value culling, through improved Glabs method and Z-score normalization, outliers in battery foil production data are eliminated, data smoothed, and modeling is ensured to ensure the effectiveness and generalization of modeling. performance.

Benefits of technology

Through outlier value removal and data normalization processing, the stability and effectiveness of battery foil production data are improved, the generalization ability of the model is enhanced, and the yield rate of battery foil is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118939952A_ABST
    Figure CN118939952A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of aluminum alloy battery foil production, and particularly relates to a method for improving the yield of battery foils by data modeling based on abnormal value rejection, which comprises the following steps of: firstly, screening original data, screening data not in a standard range, and then rejecting abnormal values in the data; the method comprises the following steps of: smoothing data by adopting an average value method to obtain a group of stable and effective data, and then standardizing the data by adopting a Z-score method, so that the data of each characteristic of the battery foil conforms to standard normal distribution, each index is in the same order of magnitude, and the influence of each different input parameter on the technical index of the battery foil is utilized to the maximum extent. The generalization capability of a mathematical model is improved, and the yield of battery foil products is improved. According to the method, the case library is constructed by using the battery foil data in actual operation, and powerful guidance is provided for an operator to analyze technical indexes of the battery foil and improve the yield of the battery foil.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of aluminum alloy battery foil production, and in particular relates to a method for improving the yield rate of battery foil based on data modeling based on outlier elimination. Background Art

[0002] As a deep-processed product of aluminum foil, battery foil has many key technologies and high technical index requirements. Its production requires the thinnest thickness, highest strength, best surface dyne value, smallest thickness difference, best version, and cleanest surface, while constantly pursuing six extreme values. In recent years, industries such as intelligent manufacturing and new energy vehicles have proposed to accelerate the process of industrial integration in my country. At the same time, technologies such as big data, cloud computing, and artificial intelligence have developed rapidly. These technologies have changed the industry production methods of traditional aluminum foil and also spawned new technologies in the aluminum foil industry. The mechanical properties prediction model is established using production data, and the intelligent production of battery foil is guided by establishing the relationship between production materials, environmental parameters and the main technical indicators of the product. In recent years, because the data-driven method is based on data, it uses several steps such as clarifying problems, collecting data, processing data, building models, evaluating and interpreting models, and drawing conclusions. It uses regression analysis, variance analysis, principal component analysis, factor analysis, cluster analysis, discriminant analysis, canonical correlation analysis, and correspondence analysis as means. It can combine modeling accuracy and complexity at the same time, is widely used, and is relatively practical. Therefore, it has become the mainstream research method for improving battery foil production technology and yield rate.

[0003] When modeling the battery foil production process, in actual applications, uncertain factors such as plate type, thickness difference, alloy composition, microstructure, and heat treatment process make it difficult to model the physical production of battery foil, and its nonlinear mechanism model is difficult to accurately establish. When the traditional data-driven method is used to model the battery foil production, the original data is not processed or the model is modeled after simple processing, which may result in large modeling errors. The model cannot correctly simulate the production principle of battery foil and cannot exist as a model for battery foil production process analysis. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method for improving the yield rate of battery foil based on data modeling by eliminating outliers.

[0005] The purpose of the present invention is to achieve the following technical solution. According to the present invention, a method for improving the yield rate of battery foil based on data modeling based on outlier elimination is provided, which comprises the following steps:

[0006] Step 1: Preprocess the historical data of battery foil production;

[0007] Step 2: Use the improved Grubbs method to remove outliers in the battery foil production data;

[0008] Step 3: Flatten the battery foil production data and calculate the mean, eliminate redundant data, and retain a set of stable and effective data per unit time;

[0009] Step 4: Use the Z-score method to normalize the different parameter indicators of the battery foil production data so that each indicator is at the same order of magnitude, and analyze the impact of different parameters on the technical indicators of the battery foil under the same dimension.

[0010] Furthermore, step one specifically includes:

[0011] Eliminate unqualified battery foil production data, including: battery aluminum foil tensile strength deviates from the 190-280Mpa range, surface wetting tension deviates from 30-32dyn, and surface quality does not meet the standards.

[0012] Furthermore, step 2 specifically includes the following process:

[0013] (1) Assume that the battery foil production data in a certain unit time period is expressed as

[0014]

[0015] Where n is the number of variables, and m is the number of battery foils produced per unit time;

[0016] (2) Assume that the tensile strength in the battery foil production data is X j1 , calculate the median M, mean μ and mean square error σ in m groups of data; if the number of battery foils produced per unit time m is less than the specified number, the improved Grubbs method is used to eliminate outliers: calculate the absolute value of the residual error of the tensile strength of each ton of battery foil |V j |=|X j1 -M|, select a set of data with the largest absolute value and find the G value:

[0017]

[0018] Assuming the significance level is α, then the control Grubbs critical value G (n,α) The table shows the Grubbs critical value G when the number of data is n. (n,α) ; Compare the actual value G obtained by the above formula with G (n,α) Value, if G>G (n,α) , then the corresponding j-th group of battery foil data is abnormal, the system issues an early warning, and the data is discarded;

[0019] (3) If the number of battery foils produced per unit time m is greater than the specified number, the tensile strength corresponding to each unit of battery foil will present a normal distribution. In this case, the large-scale criterion is adopted: the residual error V of the tensile strength of a unit of battery foil isj =X j1 -μ, yes

[0020] |V j |=|X j1 -μ|>3σ

[0021] Yield Strength Then the tensile strength of the unit battery foil is abnormal data and is eliminated;

[0022] (4) Referring to the outlier elimination process of tensile strength, outlier data values ​​of other characteristic parameters in the battery foil production data are eliminated.

[0023] Furthermore, step three specifically includes the following process:

[0024] (1) Take the data after steps 1 and 2 that have eliminated outliers within the effective range, flatten the data, calculate the average value, and eliminate redundant data;

[0025] (2) After removing redundant data from the data, a set of stable and effective optimal data is retained in each unit time;

[0026] (3) The best way to retain data is to find the average value within a unit time period.

[0027]

[0028] Among them, m′ is the number of remaining data, and the final result after data smoothing is

[0029]

[0030] After data smoothing, the data value of each feature of the battery foil is stable and effective, ensuring the effectiveness and generalization performance of modeling.

[0031] (4) According to the methods described in steps (2) and (3), the data smoothing of different characteristics in the battery foil production process is obtained in sequence.

[0032] Furthermore, step 4 includes the following process:

[0033] (1) Take the stable and effective data from the above steps, use the Z-score method to normalize the different parameter feature data, and calculate the mean and mean square error of each parameter feature so that the parameter feature data is normalized.

[0034]

[0035] where μ j is the mean of the jth dimension of the data, σ jis the standard deviation of the jth dimension of the data. The processed data conforms to the standard normal distribution, that is, the mean is 0 and the variance is 1;

[0036] (2) According to the method of step (1), normalize the data of other characteristics of the battery foil in turn;

[0037] (3) The data after outliers are eliminated, smoothed and normalized are used for mathematical modeling to ensure the effectiveness and generalization performance of the modeling.

[0038] The present invention is beneficial in that:

[0039] The collected data are subjected to outlier elimination by using the improved Grubbs method, and the data are smoothed to eliminate redundant information and increase the generalization ability of the model. The data after outlier elimination and smoothing are subjected to Z-score standardization so that each index is at the same order of magnitude, which is convenient for comprehensive comparative evaluation in the later stage. The present invention constructs a case library using actual battery foil production data, which provides powerful guidance for operators to analyze battery foil technical indicators and improve battery foil yield.

[0040] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following specifically cites a preferred embodiment and describes it in detail with the accompanying drawings as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of an embodiment of a method for improving the yield rate of battery foil based on data modeling based on outlier elimination of the present invention;

[0042] Figure 2 This is the distribution diagram of the characteristic data after removing outliers and normalizing;

[0043] Figure 3 This is a comparison chart before and after the modeling was carried out to improve the yield rate of battery foil after data processing. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] The present invention provides a method for improving the yield rate of battery foil based on data modeling based on outlier elimination, comprising the following steps:

[0046] Step 1: Preprocess the historical data of battery foil production;

[0047] Step 2: Use the improved Grubbs method to remove outliers in the battery foil production data;

[0048] Step 3: Flatten the battery foil production data and calculate the mean value, eliminate excessive redundant information, and retain a set of stable and effective data per unit time (in the same production line or production workshop);

[0049] Step 4: Use the Z-score method to normalize the different parameter indicators of battery foil production data so that each indicator is at the same order of magnitude. Analyze the impact of different parameter indicators on the technical indicators of battery foil under the same dimension to improve the yield rate of battery foil.

[0050] Step 1 specifically includes:

[0051] Collect historical data on battery foil production, obtain original sample data, collect data on parameters of various dimensions in the battery foil production process, and eliminate unqualified battery foil production data that does not meet the standard range, including: battery aluminum foil tensile strength deviates from the range of 190-280Mpa; surface wetting tension deviates from the range of 30-32dyn (dyne value); surface quality does not meet the standard (including severe streaks, color difference, strands, scratches) and other data for pre-processing to obtain data that meets the basic product qualification range.

[0052] Step 2 specifically includes the following processes:

[0053] (1) Assume that the battery foil production data within a certain unit time period is expressed as

[0054]

[0055] Where n is the number of variables; m is the amount of aluminum foil produced per unit time.

[0056] (2) Taking the removal of abnormal data values ​​of tensile strength parameter as an example, let’s assume that the tensile strength of aluminum foil is X j1 , calculate the median M, mean μ and mean square error σ in m groups of data. If the number of battery foils produced per unit time is less than the specified number, the improved Grubbs method is used to eliminate abnormal data values. Calculate the absolute value of the residual error of the tensile strength of each ton of battery foil |V j |=|X j1 -M|, select a set of data with the largest absolute value and find the G value:

[0057]

[0058] Assuming the significance level is α, then the control Grubbs critical value G(n,α) The table shows the Grubbs critical value G when the number of data is m. (n,α) Compare the actual value G obtained by the above formula with the value in Table G (n,α) Value, if G>G (n,α) , then the corresponding j-th group of battery foil data is abnormal, the system issues an alert, and the data is discarded.

[0059]

[0060] Grubbs critical value G (n,α) surface

[0061] (3) If the number of battery foils produced per unit time m is greater than a fixed number, the tensile strength corresponding to each unit (ton) of battery foil will show a normal distribution. In this case, the Pauta criterion can be used. j =X j1 -μ, yes

[0062] |V j |=|X j1 -μ|>3σ

[0063] Yield Strength The tensile strength of the unit battery foil is considered to be abnormal data and is discarded.

[0064] Referring to the outlier elimination process of tensile strength, the outlier data values ​​of other characteristic parameters in the battery foil production data are eliminated.

[0065] Step 3 specifically includes the following processes:

[0066] (1) Take the data that has been eliminated of outliers within the effective range through steps 1 and 2, build a case library with the data values ​​that meet the requirements, flatten the data, calculate the average value, and eliminate excessive redundant data.

[0067] (2) After removing redundant data, a set of stable and effective optimal data is retained in each unit time (in the same production line or production workshop).

[0068] (3) The best way to retain data is to find the average value within a unit time period.

[0069]

[0070] Where m′ is the number of remaining data. The final result after data smoothing is

[0071]

[0072] After data smoothing, the data values ​​of each feature of the battery foil are stable and effective, ensuring the effectiveness of modeling and generalization performance characteristics.

[0073] (4) According to the methods described in steps (2) and (3), the data smoothing of different characteristics in the battery foil production process is obtained in sequence.

[0074] Step 4 includes the following process:

[0075] (1) Take the stable and effective data from the above steps and use the Z-score method to normalize the data of different parameters, such as Figure 2 As shown, the mean and mean square error of each feature are calculated to normalize the data.

[0076]

[0077] where μ j is the mean of the jth dimension of the data, σ j is the standard deviation of the jth dimension of the data. The processed data conforms to the standard normal distribution, that is, the mean is 0 and the variance is 1.

[0078] (2) According to the method of step (1), the data of other characteristics of the battery foil in each unit time are normalized in the same manner to obtain a processed case library.

[0079] (3) The data after outliers are eliminated, smoothed and normalized are used for mathematical modeling analysis, and the established model has the characteristics of effectiveness and generalization performance.

[0080] According to the above steps, the best battery foil production data obtained in each unit time is used for mathematical modeling, and the established data-driven model is used to carry out prediction experiments. By comparing and analyzing parameter indicators, the influence of various input parameters on the technical indicators of the battery foil is maximized, and the production process of the battery foil is optimized online using the analyzed technical indicators. The traditional data-driven method is used to model the production of battery foil and analyze the production process, and the method of the present invention is used to model the production of battery foil and analyze the production process. Then, the battery foil is produced through the production process optimized by the corresponding analysis. According to the comparative experiment on the yield rate of the produced battery foil, the results of the yield rate of the battery foil before and after are as follows. Figure 3 As shown by Figure 3 It can be seen that the method of the present invention can improve the yield rate of battery foil. Figure 3 It can be clearly seen that through data modeling after outliers are eliminated, the impact of different input parameters on battery foil indicators is accurately analyzed, the production process is optimized, and the yield rate of battery foil is improved.

[0081] Example:

[0082] First, the original data is cleaned and the data that is not within the scope of the production process is screened out; then, the improved Grubbs method is used to remove outliers in the data, and the characteristic data of each dimension in the battery foil is removed in the same way; then the average method is used to smooth the data to obtain a set of stable and effective data; then, the Z-score method is used to standardize the data so that the data of each characteristic of the battery foil conforms to the standard normal distribution and all indicators are in the same order of magnitude; in this way, the influence of different input parameters on the technical indicators of the battery foil is maximized, the generalization ability of the mathematical model is improved, and the yield rate of the battery foil product is improved. The specific steps are as follows:

[0083] (1) Collect historical data on battery foil production and obtain original sample data

[0084] Collect data on various dimensional parameters in the battery foil production process and screen out data that does not meet the standard range.

[0085] (2) Abnormal data elimination

[0086] For data that meet the basic product qualification range, the improved Grubbs method is used to eliminate abnormal data in battery foil production data.

[0087] (3) Build a case library

[0088] Smooth the data to reduce redundancy so that a set of effective and stable data is retained per unit time. Then, the data is normalized using the Z-score method. The results are as follows: Figure 2 The data in each unit time is processed in the same way to obtain a processed case library for the next step of mathematical modeling analysis.

[0089] (4) Use the best data to perform data modeling, analyze parameter indicators, and conduct comparative experiments on battery foil yield

[0090] The battery foil production data in each unit time is used for mathematical modeling. The established data-driven model is used to conduct prediction experiments. Through comparative analysis, the influence of various input parameters on the technical indicators of battery foil is maximized. The production process of battery foil is optimized online using the analyzed technical indicators. The results of improving the yield rate of battery foil before and after are as follows: Figure 3 As shown. Figure 3 It can be clearly seen that by removing outliers and modeling the data, the impact of different input parameters on the battery foil indicators is accurately analyzed, the production process is optimized, and the yield rate of the battery foil is improved.

[0091] The results of this embodiment show that the method for improving the yield rate of battery foil based on data modeling based on outlier elimination of the present invention has important economic value and application value.

[0092] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for improving the yield rate of battery foil based on data modeling with outlier elimination, characterized in that: The following steps are involved: Step 1: Preprocess the historical data of battery foil production; Step 2: Use the improved Grubbs method to remove outliers in the battery foil production data; Step 3: Flatten the battery foil production data and calculate the mean, eliminate redundant data, and retain a set of stable and effective data per unit time; Step 4: Use the Z-score method to normalize the different parameter indicators of battery foil production data so that each indicator is at the same order of magnitude, and analyze the impact of different parameters on the technical indicators of battery foil under the same dimension.

2. The method for improving the yield rate of battery foil by data modeling based on outlier elimination according to claim 1 is characterized in that: Step 1 specifically includes: Eliminate unqualified battery foil production data, including: battery aluminum foil tensile strength deviates from the 190-280Mpa range, surface wetting tension deviates from 30-32dyn, and surface quality does not meet the standards.

3. The method for improving the yield rate of battery foil by data modeling based on outlier elimination according to claim 1 is characterized in that: Step 2 specifically includes the following processes: (1) Assume that the battery foil production data in a certain unit time period is expressed as Where n is the number of variables, and m is the number of battery foils produced per unit time; (2) Assume that the tensile strength in the battery foil production data is X j1 , calculate the median M, mean μ and mean square error σ in m groups of data; if the number of battery foils produced per unit time m is less than the specified number, the improved Grubbs method is used to eliminate outliers: calculate the absolute value of the residual error of the tensile strength of each ton of battery foil |V j |=|X j1 -M|, select a set of data with the largest absolute value and find the G value: Assuming the significance level is α, then the control Grubbs critical value G (n,α) The table shows the Grubbs critical value G when the number of data is m. (n,α) ; Compare the actual value G obtained by the above formula with G (n,α) Value, if G>G (n,α) , then the corresponding j If the battery pack foil data is abnormal, the system will issue an alert and remove the data; (3) If the number of battery foils produced per unit time m is greater than the specified number, the tensile strength corresponding to each unit of battery foil will present a normal distribution. In this case, the large-scale criterion is adopted: the residual error V of the tensile strength of a unit of battery foil is j =X j1 -μ, yes |V j |=|X j1 -μ|>3σ Yield Strength Then the tensile strength of the unit battery foil is abnormal data and is eliminated; (4) Referring to the outlier elimination process of tensile strength, outlier data values ​​of other characteristic parameters in the battery foil production data are eliminated.

4. The method for improving the yield rate of battery foil by data modeling based on outlier elimination according to claim 1 is characterized in that: Step 3 specifically includes the following processes: (1) Take the data after steps 1 and 2 that have eliminated outliers within the effective range, flatten the data, calculate the average value, and eliminate redundant data; (2) After removing redundant data from the data, a set of stable and effective optimal data is retained in each unit time; (3) The best way to retain data is to find the average value within a unit time period. Among them, m′ is the number of remaining data, and the final result after data smoothing is After data smoothing, the data value of each feature of the battery foil is stable and effective, ensuring the effectiveness and generalization performance of modeling. (4) According to the methods described in steps (2) and (3), the data smoothing of different characteristics in the battery foil production process is obtained in sequence.

5. The method for improving the yield rate of battery foil by data modeling based on outlier elimination according to claim 1 is characterized in that: Step 4 includes the following process: (1) Take the stable and effective data from the above steps, use the Z-score method to normalize the different parameter feature data, and calculate the mean and mean square error of each parameter feature so that the parameter feature data is normalized. where μ j is the mean of the jth dimension of the data, σ j is the standard deviation of the jth dimension of the data. The processed data conforms to the standard normal distribution, that is, the mean is 0 and the variance is 1; (2) According to the method of step (1), normalize the data of other characteristics of the battery foil in turn; (3) The data after outliers are eliminated, smoothed and normalized are used for mathematical modeling to ensure the effectiveness and generalization performance of the modeling.

Citation Information

Patent Citations

  • Abnormal value identification method based on improved Grubbs test method

    CN116644368A

  • Dynamic K screening method for lithium ion battery

    CN118275895A

  • Hot-rolled steel mechanical property machine learning method guided by physical metallurgy

    CN118298978A