Electronic device and method of analyzing factors affecting battery performance of electronic device

Abstract

A method for analyzing factors influencing battery performance in electronic devices is provided, the method comprising the steps of: identifying a causal relationship graph and process data, the causal relationship graph indicating the relationship between multiple factors related to the battery manufacturing process, the process data including data on the multiple factors identified each time the battery manufacturing process is performed as each data instance; and identifying information on the impact of each of the multiple factors on the characteristics of the final product based on the causal relationship graph, the process data, and a model trained to predict the characteristics of the final product of the battery manufacturing process by receiving the process data.

Description

Technical Field

[0001] This invention relates to a method for analyzing the factors affecting the performance of electronic devices and their batteries. Background Technology

[0002] In battery manufacturing, various sensors and process factors are correlated in complex nonlinear ways, making it difficult to identify parameters affecting the process. While research has been conducted using machine learning and explainable artificial intelligence (XAI) algorithms to identify key factors influencing battery cell performance, the difficulty with these techniques lies in the assumption of independence between individual processes throughout the entire process. Specifically, this assumption of independence between individual processes fails to reflect the characteristics of a battery manufacturing process where individual processes for several mechanisms are sequentially connected, and provides explanations far removed from actual system behavior.

[0003] When such flawed analytical methods are used as a basis, key factors among the many elements that shape the battery manufacturing process are incorrectly selected, making efficient quality management and product design difficult. Therefore, it may be necessary to improve the accuracy of identifying the contributions of key factors by searching for data-driven analytical methods that can reflect the causal relationship between these process factors and measured values. Summary of the Invention

[0004] Technical issues

[0005] This disclosure provides an exemplary embodiment of a method for analyzing factors affecting the performance of electronic devices and their batteries. Specifically, an exemplary embodiment of this disclosure provides a data-based analysis method that can reflect the causal relationship between process factors and measured values, thereby improving the accuracy of determining the contribution of key factors.

[0006] However, the objectives to be achieved by the exemplary embodiments of this disclosure are not limited to the above objectives, and other objectives can be clearly understood from the following exemplary embodiments.

[0007] Technical solution

[0008] According to one aspect, a method for analyzing factors affecting battery performance in an electronic device is provided. The method includes: identifying a causal relationship graph and process data, the causal relationship graph illustrating relationships between multiple factors associated with a battery manufacturing process, the process data including data for the multiple factors, the process data being identified as independent data instances each time the battery manufacturing process is executed; and identifying information on the influence of each of the multiple factors on the final product characteristics based on the causal relationship graph, the process data, and a model trained to predict final product characteristics in the battery manufacturing process by receiving the process data as input, and the causal relationship graph illustrating: physical property nodes connected in a unidirectional direction corresponding to the sequence of detailed processes included in the battery manufacturing process, and respectively corresponding to intermediate product characteristics and the final product characteristics generated separately in the detailed processes; and multiple process parameter nodes connected in a unidirectional direction to each of the physical property nodes and corresponding to process parameters controlled in each of the detailed processes respectively corresponding to the physical property nodes.

[0009] The battery performance influencing factor analysis method may further include: identifying multiple paths from the multiple process parameter nodes to the physical property nodes corresponding to the final product characteristics based on the causal relationship graph; identifying virtual data instances by adjusting at least a portion of the data of the data instances included in the process data based on the identified multiple paths; inputting the virtual data instances into the model; and identifying information about the influence by analyzing the model's output to the virtual data instances.

[0010] Identifying the virtual data may include: identifying at least one first path based on the plurality of paths, the at least one first path comprising an edge between a first node corresponding to a first factor in the causal graph and a second node corresponding to a child node of the first node; and the step of identifying at least one first virtual data instance based on the at least one first path, and identifying information about the influence, includes the following steps: identifying first value information about the first path based on the first virtual data instance and the model; and identifying the influence value of the first factor on a second factor corresponding to the second node based on the first value information.

[0011] Identifying the at least one first virtual data instance may include: identifying at least one path preceding the at least one first path in a plurality of permutations of the plurality of paths; for each of the plurality of permutations, identifying a first subset including the at least one path preceding the at least one first path and a second subset including the first path; identifying a first-first virtual data instance by randomly sampling and adjusting the data of nodes in the data instance that do not correspond to the first subset, and identifying a first-second virtual data instance by randomly sampling and adjusting the data of nodes in the data instance that do not correspond to the second subset, and identifying the first value information includes: identifying first-first value information and first-second value information based on the results of the first-first virtual data instance and the first-second virtual data instance identified for each of the plurality of permutations input to the model; and identifying the first value information based on information about the difference between the first-first value information and the first-second value information.

[0012] Identifying the first value information may include: identifying the first value information regarding the first path based on information about the average of the differences between the first-first value information and the first-second value information identified for each of the plurality of permutations.

[0013] Identifying the first value information may include: identifying first-first output information by inputting the first-first virtual data instance into the model; and identifying the first-first value information based on information about the difference between the average value of the output information identified by inputting the unadjusted data instance into the model and the first-first output information.

[0014] Information identifying the influence of the first factor may include: identifying a first influence value of the first factor on the second factor based on information about the sum of first value information about the first path, wherein the first path includes an edge between the first node and the second node.

[0015] Identifying information about the influence may include: identifying elements of a matrix comprising the influence values ​​of a first factor corresponding to a first coordinate on a first axis on a second factor corresponding to a second coordinate on a second axis; and identifying information about the influence of the plurality of factors on the characteristics of the final product based on the matrix, the information about the influence including information about direct influence and information about indirect influence.

[0016] Identifying information about the influence may include: identifying the influence value of at least one factor corresponding to at least one node, the at least one node being directly connected in the matrix to a node corresponding to the final product characteristic; and identifying information about the direct influence based on the influence value of the at least one factor.

[0017] Identifying the influence value of at least one factor may include: identifying coordinates on a second axis corresponding to the final product characteristic; identifying a vector in a matrix having coordinates on the second axis corresponding to the final product characteristic and parallel to the first axis; and identifying the influence value of at least one factor based on the vector.

[0018] Identifying information about the effect may include: identifying information about the sum of influence values ​​by means of factors corresponding to each of the process parameter nodes; and, based on the information about the sum, identifying information about the indirect effects by means of factors corresponding to each of the process parameter nodes.

[0019] Identifying information about the sum may include: identifying coordinates on the first axis corresponding to each of the process parameters; identifying each vector in the matrix, the vector having coordinates corresponding to each of the process parameters on the first axis and parallel to the second axis; and identifying information about the sum of the influence values ​​of each included in the vectors by means of factors corresponding to each of the process parameter nodes.

[0020] The battery performance influencing factor analysis method may further include: identifying contribution ratio information based on information about the influence, the contribution ratio information showing the ratio of the values ​​of information about the direct influence and information about the indirect influence for each of the plurality of factors.

[0021] The battery performance influencing factor analysis method may further include: supporting the control of the intermediate product characteristics based on information about the direct influence.

[0022] The battery performance influencing factor analysis method may further include: using information about the indirect effects to support the control of the process parameters of each of the detailed processes.

[0023] Identifying the causal relationship diagram and the process data may include: identifying the entire process data identified each time the battery manufacturing process is executed; calculating the correlation coefficients between all factors of the battery manufacturing process based on the entire process data; identifying the plurality of factors by filtering at least a portion of the factors whose correlation coefficients are greater than or equal to a threshold; and identifying at least a portion of the data in the entire process data corresponding to the plurality of factors as the process data.

[0024] Identifying at least a portion of the data corresponding to the multiple factors in the entire process data as the process data may include: identifying the data instances in the entire process data that correspond to the measured temperature and effective temperature range of the final product characteristics as the process data.

[0025] The battery performance influencing factor analysis method may further include: inputting the process data into the model on a data instance basis; and training the model based on the comparison results between the actual data corresponding to the final product characteristics identified on a data instance basis and the output of the model.

[0026] According to another aspect, an electronic device for analyzing factors affecting battery performance is also provided, the electronic device comprising: a processor and a memory configured to store one or more instructions, the processor being configured to, by executing the one or more instructions: identify a causal relationship graph and process data, the causal relationship graph illustrating relationships between multiple factors associated with a battery manufacturing process, the process data including data for the multiple factors, the process data being identified as independent data instances each time the battery manufacturing process is executed; and, based on the causal relationship graph, the process data, and a model trained to predict final product characteristics in the battery manufacturing process by receiving the process data as input, identify information on the influence of each of the multiple factors on the final product characteristics, and the causal relationship graph illustrating: physical property nodes connected in a unidirectional direction corresponding to the sequence of detailed processes included in the battery manufacturing process, and respectively corresponding to intermediate product characteristics and the final product characteristics generated separately in the detailed processes; and a plurality of process parameter nodes connected in a unidirectional direction to each of the physical property nodes and corresponding to process parameters controlled in each of the detailed processes respectively corresponding to the physical property nodes.

[0027] According to another aspect, a non-transitory computer-readable recording medium is also provided, wherein a program is recorded for performing the above-described method for analyzing factors affecting battery performance on a computer.

[0028] Additional aspects of the example implementation will be set forth in part in the description which follows, and will be apparent in part from the description.

[0029] Beneficial effects

[0030] Based on the proposed example implementation, one or more of the following effects can be expected.

[0031] According to the example implementation, the accuracy of determining the contribution of key factors can be improved by using a data-based analysis method that can reflect the causal relationship between process factors and measured values.

[0032] According to the example implementation, the impact of process factors can be explained by classifying them as direct or indirect effects based on the objective.

[0033] The effects of this disclosure are not limited to those described above, and other effects will become apparent to those skilled in the art from the following description of the appended claims. Attached Figure Description

[0034] Figure 1 This is a diagram illustrating an example of the operation of an electronic device that analyzes factors affecting battery performance according to an exemplary embodiment.

[0035] Figure 2 This is a flowchart describing a method for analyzing factors affecting battery performance according to an example implementation.

[0036] Figure 3 This is an example diagram illustrating a causal relationship diagram according to an exemplary implementation.

[0037] Figure 4 This is an example diagram illustrating the path from the process parameter node to the property node corresponding to the final product characteristic in the causal relationship diagram, according to an example implementation.

[0038] Figure 5 This is a diagram illustrating an example of a matrix including influence values ​​according to an example implementation.

[0039] Figure 6 This is a diagram illustrating an example of identifying information about direct impacts using a matrix, according to an exemplary implementation.

[0040] Figure 7 This is a diagram illustrating an example of identifying information about indirect effects using a matrix, according to an exemplary implementation.

[0041] Figure 8a and Figure 8b This is a diagram illustrating information about the direct effects and information about the indirect effects according to an example implementation.

[0042] Figure 9 This is an example diagram illustrating contribution ratio information according to an example implementation.

[0043] Figure 10 This is a block diagram illustrating an electronic device according to an example embodiment. Detailed Implementation

[0044] While taking into account the functionality obtained under this disclosure, as many terms as possible are selected from currently widely used general terminology used in the example embodiments; however, these terms may be replaced by other terms based on the intent, habits, or emergence of new technologies of those skilled in the art. Furthermore, in certain cases, terms arbitrarily chosen by the applicant of this disclosure may be used. In such cases, the meanings of these terms may be described in the corresponding descriptive sections of this disclosure. Therefore, it should be noted that the terms used herein should be interpreted based on their actual meaning and the entirety of this specification, rather than simply on their names.

[0045] Throughout the specification, when an element is referred to as "including" another element, that element should not be construed as excluding other elements, unless otherwise specified in the description, and that element may include at least one other element.

[0046] Throughout the specification, the phrase "at least one of a, b, and c" may include all of 'a' only, 'b' only, 'c' only, 'a and b', 'a and c', 'b and c', or 'a, b, and c'.

[0047] In this disclosure, a "terminal" can be implemented as a computer or portable terminal capable of accessing a server or another device via a network. The computer may include, for example, a laptop computer, a desktop computer, and a notebook equipped with a web browser. The portable device may be a wireless communication device that ensures portability and mobility, and includes any type of handheld wireless communication device, such as a tablet PC, a smartphone, or a communication-based device such as International Mobile Telecommunications (IMT), Code Division Multiple Access (CDMA), W-CDMA, and Long Term Evolution (LTE).

[0048] In the following description, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings, enabling those skilled in the art to readily implement the present disclosure. However, the present disclosure may be embodied in many different forms and is not limited to the exemplary embodiments described herein.

[0049] In the following description, exemplary embodiments of the present disclosure will be described with reference to the accompanying drawings.

[0050] Figure 1This is a diagram illustrating an example of the operation of an electronic device that analyzes factors affecting battery performance according to an exemplary embodiment.

[0051] refer to Figure 1 The electronic device 100 can operate by receiving causal relationship diagrams and process data as inputs, as will be described below. Additionally, the electronic device 100 may include model 110.

[0052] Here, Model 110 can be implemented using machine learning algorithms that can reflect the nonlinear correlations between process factors, such as various algorithms like XGBoost, Random Forest, or deep neural networks (DNNs), but not limited to those described above. Model 110 can be a model trained to output the final product property (FPP) generated by the entire battery manufacturing process by receiving data corresponding to factors controlled in each detailed process of the battery manufacturing process and receiving data corresponding to intermediate product features (IPFs) generated in each detailed process. Its training method can be determined in association with the various algorithms described above or another algorithm and model building scheme.

[0053] at the same time, Figure 1 The diagram shows elements that are associated only with this example embodiment. Therefore, those skilled in the art who associate this example embodiment with other elements will understand that, in addition to… Figure 1 In addition to the components shown, other commonly used components may also be included.

[0054] Figure 2 This is a flowchart describing a method for analyzing factors affecting battery performance according to an example implementation.

[0055] In operation S210, the electronic device 100 can identify a causal relationship diagram and process data. The causal relationship diagram illustrates the relationships between multiple factors associated with the battery manufacturing process. The process data includes data for multiple factors, and the process data is identified as an independent data instance each time the battery manufacturing process is executed. In operation S220, the electronic device 100 can identify information about the impact of each of the multiple factors on the final product characteristics based on the causal relationship diagram, the process data, and a model 110 trained to predict the final product characteristics in the battery manufacturing process by receiving the process data as input. The battery performance influencing factor analysis method according to an exemplary embodiment of this disclosure will be described in detail below.

[0056] First, the electronic device 100 can acquire a causal relationship graph. According to an example implementation, the causal relationship graph can be acquired through input to the electronic device 100, by a user with domain knowledge related to the battery manufacturing process, by input sent to the electronic device 100 from another server, or by the electronic device 100's analysis of process factors related to the battery manufacturing process.

[0057] Such a causal graph may include: property nodes, which are connected in a unidirectional direction corresponding to the sequence of detailed processes included in the battery manufacturing process, and respectively correspond to intermediate product features and final product characteristics generated separately in the detailed processes; and multiple process parameter nodes, which are connected in a unidirectional direction to each property node and correspond to process parameters controlled in each detailed process corresponding to the property node. As can be inferred from the above description, the causal graph may be of the type of a directed acyclic graph (DAG).

[0058] Examples of each detailed process included in the battery manufacturing process may include, but are not limited to, mixing, coating, rolling, cutting and grooving, assembly, activation, aging, and degassing processes, and need not include all of them. Furthermore, examples of process parameters may include slurry temperature, slurry flow rate, pump, coating gap length, drying temperature and air volume, coating speed, coating thickness and coating load for the coating process; rolling roll gap length, rolling pressure, rolling speed and rolling thickness for the rolling process; laminate temperature, lamination pressure, lamination roll temperature, lamination speed and dimensions for the assembly process; and J / F temperature and pressure, aging time, inter-process waiting time, electrolyte quantity, performance measurement temperature, and final cell thickness for the activation, post-processing, and overall processes. Similarly, examples of process parameters are not limited to those described above and need not include all of them. Additionally, examples of intermediate product characteristics may include rolling rate, load after rolling, negative electrode thickness after activation, load of a single cell, and air gap amount in the positive and negative electrodes after coating, rolling, and activation. Examples of final product characteristics may include discharge capacity, discharge capacity at the measured temperature, charging resistance, etc. However, examples of intermediate and final product characteristics are not limited to those described above, and need not include all of the above examples.

[0059] According to an example implementation, a subset of all factors associated with the battery manufacturing process can be selected, enabling the battery performance influencing factor analysis method of this disclosure to be executed. As an example, electronic device 100 can identify the entire process data identified each time the battery manufacturing process is executed. Electronic device 100 can then calculate the correlation coefficients between all factors of the battery manufacturing process based on the entire process data. Electronic device 100 can identify multiple factors by filtering at least a subset of factors whose correlation coefficients are greater than or equal to a threshold. In other words, representative factors among those with correlation coefficients greater than or equal to the threshold can be exclusively retained, and other factors can be filtered. As an example, when factors A, B, C, and D with correlation coefficients greater than or equal to the threshold are combined, one factor (e.g., factor A) can be exclusively retained, and other factors can be filtered. Here, according to the example implementation, among a set of factors with correlation coefficients greater than or equal to the threshold, the factors that remain unfiltered can have the highest ranking importance. According to exemplary embodiments, correlation coefficients can be calculated using various schemes for calculating correlation coefficients, such as the Pearson correlation coefficient, Spearman rank correlation coefficient, or Kendall's π coefficient, but are not limited to those described above. A threshold can be set as a boundary value, which can be determined to have a relevant correlation based on the characteristics of each correlation coefficient calculation scheme. The aforementioned causal relationship graph can be generated and input as a reference along with a selected portion of the entire set of factors associated with the battery manufacturing process.

[0060] Reference Figure 3 So that we can see an example of the causal relationship diagram above.

[0061] Figure 3 This is an example diagram illustrating a causal relationship diagram according to an exemplary implementation.

[0062] refer to Figure 3 An example of a causal relationship diagram can be shown. Figure 3 In the middle, as one of the physical property nodes 310 311 can be a node corresponding to an intermediate product feature based on the An1 process. 321 and 322 can be a node corresponding to process parameters that can be controlled in association with the An1 process. For example, the An1 process is a rolling process. 321 can be a node corresponding to the rolling pressure, and 322 can be a node corresponding to the rolling speed. Therefore, a causal relationship graph can be identified, where unidirectional edges connect from process parameter nodes 320, corresponding to process parameters that can be controlled in the corresponding detailed process, to property nodes 310, corresponding to intermediate product characteristics in each detailed process. Furthermore, as... Figure 3 The noise can be identifiable as a unidirectional edge connecting to the node associated with each intermediate product feature, reflecting measurement errors or other various errors. Meanwhile, FPP node 315 can be a node corresponding to a final product feature, and nodes 311, 312, 313, and 314 marked with IPF can each be a node corresponding to an intermediate product feature. Figure 3 The causal relationship diagram above is merely an example, and for ease of description, much of it can be simplified. For instance, there could be several process parameters (FPPs), and each process parameter could be linked to various intermediate products, or the relationships between process parameters (IPFs) could be connected in a more complex structure. Figure 3 From 321 connected to Edge 330 of 311 is marked for what will be described below.

[0063] The electronic device 100 can identify information about process data in parallel with the process of identifying the aforementioned causal relationship diagram. Process data may include data on multiple factors, which are identified as independent data instances each time a battery manufacturing process is executed. Here, an independent data instance may be data about factors associated with the process of manufacturing an individual battery cell. For example, a predetermined data instance associated with a predetermined battery cell may include information about process parameters for each detailed process set in the manufacturing process of the predetermined battery cell, as well as intermediate product characteristics. Process data may include such data instances each time a battery manufacturing process is executed.

[0064] Furthermore, when generating the aforementioned causal relationship diagram using a selected portion from the entire set of factors associated with the battery manufacturing process as a reference, the process data may include data corresponding to the selected portion of the factors. For example, when filtering a portion of the entire set of factors related to the battery manufacturing process based on correlation coefficients, for each data instance, the process data may include data regarding the unfiltered factors. Therefore, the electronic device 100 can identify at least a portion of the entire process data corresponding to multiple selected factors as process data.

[0065] According to the example implementation, factors can be filtered based on the correlation coefficients described above, and then some data instances can be filtered based on the measurement temperature of the final product characteristic. That is, the electronic device 100 can identify data instances from the data instances included in the entire process data that correspond to the measurement temperature of the final product characteristic within an effective temperature range as process data. Since temperature has a significant impact on the measurement of the final product characteristic, such an effective temperature range can be set to analyze performance influencing factors of the final product characteristic as a reference, which is measured at similar temperatures within the effective temperature range. The effective temperature range can be set to a range within which the final product characteristic will not be measured differently due to temperature, and can be set to a predetermined temperature range (e.g., from 28.5 degrees Celsius to 29.5 degrees Celsius). Here, the individual data instances included in the process data may not include data regarding the performance of the final product characteristic based on the corresponding manufacturing process, and the model 110 may be in a state where it has been trained using data instances as input and using data regarding the actual measured final product characteristic corresponding to the data instance as the answer.

[0066] In the following, an example implementation will be described in which the electronic device 100 identifies information about the impact of each of a plurality of factors on battery performance by using model 110 together with the aforementioned causal relationship diagram and process data.

[0067] According to an exemplary implementation, electronic device 100 can identify multiple paths connecting multiple process parameter nodes to physical property nodes corresponding to final product characteristics based on a causal relationship graph. For example, electronic device 100 can identify all allowed paths in the causal relationship graph. According to an exemplary implementation, electronic device 100 can therefore identify multiple paths connecting multiple process parameter nodes to physical property nodes corresponding to final product characteristics using depth-first search (DFS). (Refer to...) Figure 4 Identify the path about Figure 3 The above examples are examples that are thus identified.

[0068] refer to Figure 4 It can be identified from the above reference. Figure 3 The described process parameter node 320 Node 321 is an example of the first path 410 connected to FPP node 315 corresponding to the final product characteristics. Although not explicitly shown in the description... Figure 4 As shown, but similar to the first path 410, multiple paths may include those from each process parameter node 320 (i.e., by...). Figure 3 The reference numerals 322 to 329 in the accompanying drawings indicate , , , , , , , Continue the path to FPP node 315. For ease of description, in the following paragraphs, numbers 322 to 329 will be used to indicate the path to FPP node 315. , , , , , , , The path continuing to FPP node 315 can be described separately as the second through ninth paths.

[0069] Subsequently, electronic device 100 can identify virtual data instances by adjusting at least a portion of the data of the data instance included in the process data based on multiple paths thus identified. In a virtual data instance, data corresponding to some factors in the data of the data instance can be used without any change, but the value corresponding to another factor can be batch-randomly sampled data (which is the process data in this disclosure) and identified. Electronic device 100 can identify virtual data instances thus identified to model 110, and therefore identify information about the influence by analyzing the output of model 110 to the virtual data instance. Such a process will be described in further detail below.

[0070] According to an example implementation, the electronic device 100 can identify the influence value of each edge in a causal relationship graph. Here, since the edges are formed in a unidirectional direction, the influence value of an edge can be the value of the influence of the factor corresponding to the parent node on the factor corresponding to the child node. For example, the electronic device 100 can calculate the influence value corresponding to the factor corresponding to the child node. The configuration of process parameters at node 321 is related to... The value of the impact of the intermediate product characteristics corresponding to node 311.

[0071] To calculate the influence value of an edge in this way, the electronic device 100 can identify at least one first path, which includes an edge between a first node corresponding to a first factor in a causal graph and a second node corresponding to a child node of the first node. For example, Figure 3 The aforementioned edge 330 may be included in Figure 4 In the first path 410. In this example, there exists a single path that includes edges, but in another case, for example, there could be paths that include... Several paths between the edges.

[0072] The electronic device 100 can identify at least one first virtual data instance based on a first path. Here, a process for identifying the first virtual data instance according to an example embodiment will be described.

[0073] According to an example implementation, electronic device 100 can identify multiple arrangements of multiple paths randomly arranged. For example, in Figure 4 In the above example, electronic device 100 can identify the random arrangement of paths 1 to 9. An array of permutations. All of these permutations can be identified, but sampled and identified portions of them are also permitted in example implementations. Electronic device 100 can identify at least one path positioned prior to the aforementioned first path for each permutation. For example, when the permutation is formed as "second path, third path, first path, fourth path, etc.", electronic device 100 can identify the second and third paths as at least one path positioned prior to the first path for each permutation. Electronic device 100 can identify for each of the plurality of permutations a first subset comprising at least one path positioned prior to the aforementioned first path and a second subset of the first subset including the first path. For example, in the example above, the first subset can be formed as "second path," "third path," and the second subset can be formed as "second path," "third path," "first path."

[0074] Then, the electronic device 100 can identify the first-first virtual data instance by randomly sampling and adjusting the data of nodes in the data instance that do not correspond to the first subset, and identify the first-second virtual data instance by randomly sampling and adjusting the data of nodes in the data instance that do not correspond to the second subset. For example, when combined with Figure 4 When describing the above example using path examples, for a first subset formed as "second path, third path", electronic device 100 can maintain the values ​​of nodes included in the paths of the first subset for each data instance, and randomly sample the values ​​represented by reference numerals 321, 324, 325, 326, 327, 328, 329 and 312 from another data instance included in the process data. , , , , , , , To identify the first-to-first virtual data instances, these nodes are those not included in the path of the first subset. For example, such a random sampling task can be performed for all data instances to identify the first-to-first virtual data instances, or a random sampling task can be performed specifically for a portion of the data instances to identify the first-to-first virtual data instances. Similarly, for the second subset, values ​​corresponding to nodes included in the path of the second subset can be maintained for each data instance, and can be randomly sampled from another data instance included in the process data, denoted by reference numerals 324, 325, 326, 327, 328, 329, and 312. , , , , , , The values ​​are used to identify the first and second virtual data instances, which are nodes not included in the path of the second subset. Similarly, for the first and second virtual data instances, such a random sampling task can be performed on all data instances to identify the first and second virtual data instances, or the random sampling task can be performed specifically on a portion of the data instances to identify the first and second virtual data instances.

[0075] According to an example implementation, electronic device 100 can identify first-first value information and first-second value information based on the results of inputting first-first virtual data instances and first-second virtual data instances into model 110. Specifically, electronic device 100 can identify first-first output information by inputting first-first data instances into model 110. Then, electronic device 100 can identify information regarding the difference between the average of multiple output information identified by inputting unadjusted raw data instances included in process data into model 110 and the first-first output information, and can use this difference information as the first-first value information. Information regarding the average of multiple output information identified by inputting unadjusted raw data instances into model 110 can be pre-calculated. Similarly, electronic device 100 can identify first-second output information by inputting first-second data instances into model 110. Then, electronic device 100 can identify information regarding the difference between the average of multiple output information identified by inputting unadjusted raw data instances included in process data and the first-second output information, and can use this difference information as the first-second value information.

[0076] According to the example implementation, the electronic device 100 can calculate information regarding the difference between first-first value information and first-second value information. Since the first subset, second subset, first-first virtual data instance, and first-second virtual data instance are identified for each permutation, first-first value information and first-second value information can be identified for each permutation. The electronic device 100 can calculate information regarding the difference between the first-first value information and first-second value information identified for each permutation. Then, the electronic device 100 can calculate information about the average value of the information regarding the difference between the first-first value information and first-second value information, and can identify the average value of the information regarding the difference as the first value information regarding the first path. That is, referring back to the example above, the electronic device 100 can calculate information regarding the difference between the first-first value information of the first subset formed as "second path, third path" and the first-second value information of the second subset formed as "second path, third path, first path". Therefore, information regarding the difference between the first-first value information and first-second value information regarding the corresponding permutation can be calculated. When the information regarding the difference between the first-first value information and the first-second value information of the corresponding permutation is calculated in a similar manner, such an average of the information regarding the difference can be identified as the first value information regarding the first path. The above scheme may be suitable for identifying the degree of contribution of the values ​​included in the first path when model 110 receives data instances and outputs final product characteristics, and will be described below.

[0077] According to an example implementation, the electronic device 100 can identify a first influence value of a first factor on a second factor based on information about the sum of first value information regarding a first path including the edge between the first node and the second node. Because Figure 3 In the example above, edge 330 is included Figure 4 In the first path 410, that is, in a unique path, the first value information about the first path 410 can therefore be identified as the first influence value of the first factor on the second factor without any change. However, conversely, due to the connection The edges with FPP can be included in all paths from the first to the ninth, so the sum of the value information about paths from the first to the ninth can be calculated as the influence value of the edges, i.e., the intermediate product. The impact value on FPP. As another example, the connection... and The edges can be included in the fifth and sixth paths, and the sum of multiple value information about the fifth and sixth paths can be calculated as the edge influence value, i.e., the intermediate product. intermediate product The impact value.

[0078] The following sections will describe the calculation process of the aforementioned series of influence values ​​using Equations 1 to 3. First, Equation 1, which is associated with the function used to calculate the value information, will be described.

[0079] [Equation 1]

[0080] In equation 1, It can represent a set that includes multiple permutations as described above. It can represent It includes a permutation. Furthermore, 'a' can represent the first path. This can represent the path x in the permutation The position within. Using this definition, we can understand, The second subset can be represented as a subset including paths whose elements in the corresponding permutation are positions preceding or equal to the first path. The first subset mentioned above can be represented as a subset including paths whose elements are positions preceding the first path in the corresponding permutation. Furthermore, A function that can be represented as a function of computed value information receives a subset including at least one path as input, keeps the values ​​corresponding to the nodes included in the corresponding path in the data instance unchanged, and outputs information about the difference between the average of multiple output information obtained by randomly sampling the values ​​of factors corresponding to nodes not included in the path from other data instances of the process data and inputting the values ​​into model 110 and the average of multiple output information obtained by inputting the entire data instance of the process data, as described above. Its purpose is consistent with Equation 2. The functions have the same purpose, but have relatively different input values ​​as described below.

[0081] [Equation 2]

[0082] In Equation 2, S can represent a set (or union) that includes at least one node. This can represent the values ​​corresponding to p integer factors. It can be expressed that the integral is calculated for allowed combinations of parameters not included in the set S. That is, the function corresponding to model 110, and can be understood as equal to the average of the results of inputting data instances obtained by randomly sampling parameters not included in set S into model 110. This can represent the average of multiple output messages from model 110 across the entire process data set X. As can be seen in the above description of Equation 2, although because... It receives a subset of nodes as input, and The input includes a subset of paths, which may differ, but those skilled in the art can understand this from Equation 2. It receives a subset of paths as input.

[0083] [Equation 3]

[0084] In equation 3, It can represent a cause-and-effect diagram. In the above arrangement, when path p includes edge e, It can have only the value 1, otherwise it has the value 0. This can represent value information about path p. According to equation 3, The output value can be the sum of multiple values ​​on all paths p including edge e. This can be the same as the definition of the influence value of edge e above.

[0085] The impact value calculation scheme based on the above equation can be an application of the scheme for calculating Shapley values. Shapley values ​​are derived from game theory and are used to evaluate the contribution of each participant. In other words, the impact value of each edge can be understood as a value that evaluates the contribution of each edge in a sense similar to the Shapley value.

[0086] According to the above example implementation, when calculating the influence value of the factor corresponding to the parent node on the factor corresponding to the child node for each edge (where the parent node and child node are connected by an edge), the electronic device 100 can identify a matrix including the influence values. For example, the electronic device 100 can identify a matrix including the influence values ​​of the first factor corresponding to the first coordinate on the first axis on the second factor corresponding to the second coordinate on the second axis as elements. Since the influence value of the first factor corresponding to the first coordinate on the first axis on the second factor corresponding to the second coordinate on the second axis is included as an element, the matrix can have the form of a triangular matrix. Diagonal elements and elements showing the influence values ​​between nodes that are not directly connected to each other can be automatically entered as 0. (Refer to...) Figure 5 In order to describe such an example.

[0087] Figure 5 This is a diagram illustrating an example of a matrix including influence values ​​according to an example implementation.

[0088] refer to Figure 5 The matrix can include a vertical axis corresponding to the parent node and a horizontal axis corresponding to the child node. Each element can be the influence value of a factor corresponding to a coordinate on the vertical axis on a factor corresponding to a coordinate on the horizontal axis. For example, example element 501 could be a factor... Factors The impact value. In Figure 5 In the example, the first axis and the second axis can correspond to the vertical axis and the horizontal axis, respectively.

[0089] According to the example implementation, the electronic device 100 can identify information about the impact of multiple factors on the characteristics of the final product based on the matrix described above, including information about direct impact and information about indirect impact.

[0090] First, regarding information about direct impact, the electronic device 100 can identify the impact value of at least one factor corresponding to at least one node in the matrix that is directly connected to the node corresponding to the final product characteristic. For example, according to the above... Figure 3 For example, electronic device 100 can recognize , and The impact value on FPP. Electronic device 100 can identify information about direct impact based on at least one impact value thus identified. The impact value of at least one factor corresponding to the at least one directly connected node can be easily identified via a matrix. In other words, electronic device 100 can identify the coordinates on the second axis corresponding to the final product characteristic corresponding to the child node, identify a vector in the matrix having coordinates on the second axis corresponding to the final product characteristic and parallel to the first axis, and identify the impact value of at least one factor corresponding to the at least one directly connected node based on this vector. More specifically, elements in the identified vector included in the matrix can be identified as the impact value of at least one factor corresponding to the at least one directly connected node. (Refer to...) Figure 6 So that we can see an example of a vector.

[0091] Figure 6 This is a diagram illustrating an example of identifying information about direct impacts using a matrix, according to an exemplary implementation.

[0092] refer to Figure 6 The electronic device 100 can identify that the coordinates on the second axis corresponding to the FPP are the coordinates at the right end, and identify a vector 600 having said coordinates and parallel to the first axis. The electronic device 100 can identify the values ​​included in the vector 600 as the influence values ​​of at least one factor corresponding to at least one directly connected node.

[0093] Then, regarding information about indirect effects, the electronic device 100 can identify information about the sum of influence values ​​through factors corresponding to each process parameter node. Furthermore, information about indirect effects can be identified based on information about the sum, through factors corresponding to each process parameter node. For example, according to the above... Figure 3For example, electronic device 100 can target corresponding factors , , , , , , , , Information about the sum of influence values ​​is identified. Electronic device 100 can identify information about indirect influences based on at least one influence value so identified for each factor. The factor-wise influence value corresponding to each of the aforementioned process parameter nodes can be easily identified via a matrix. For example, electronic device 100 can identify the coordinates corresponding to each process parameter on a first axis in the matrix. Electronic device 100 can identify each vector in the matrix, which has coordinates corresponding to each process parameter on the first axis and is parallel to the second axis. Electronic device 100 can identify information about the sum of influence values ​​included in each vector by the factors corresponding to each process parameter node, and identify information about indirect influences based on the information about the sum. (Refer to...) Figure 7 So that we can see an example of this vector.

[0094] Figure 7 This is a diagram illustrating an example of identifying information about indirect effects using a matrix, according to an exemplary implementation.

[0095] refer to Figure 7 The electronic device 100 can identify the coordinates corresponding to the process parameter nodes, i.e., PP on the first axis, and identify the vector 700 with coordinates parallel to the second axis. The electronic device 100 can identify information about the indirect influence of each process parameter by obtaining the sum of the values ​​included in the vector 700 via the vector. Figure 7 The vector 700 parallel to the second axis is shown to correspond to the corresponding portion of the row with coordinates, rather than the entire row. This is because since PP (i.e., process parameters) are set to 0 and therefore do not affect each other, the result will be the same regardless of whether the portion associated with the influence values ​​between PP is included.

[0096] According to the example implementation, the electronic device 100 can identify and plot information about direct effects and information about indirect effects based on the above example implementation. (Refer to...) Figure 8a and Figure 8b This is to show examples of information drawn in this way regarding direct effects and information regarding indirect effects.

[0097] Figure 8a and Figure 8b This is a diagram illustrating information about the direct effects and information about the indirect effects according to an example implementation.

[0098] Figure 8a This is a graph showing information about the direct impact. Its vertical axis represents each factor. The direct impact value of each factor on the final product characteristics can be identified by bars parallel to its horizontal axis. Factors prefixed with X (e.g., ...) are listed on the vertical axis. , ) can be process parameters, factors prefixed with Y (e.g., , (etc.) can be characteristics of intermediate products.

[0099] Figure 8b This is a graph illustrating information about indirect effects. Its vertical axis represents each factor. The indirect effect value for each process parameter can be identified by bars parallel to its horizontal axis. Because this graph shows information about indirect effects, it is related to… Figure 8a The difference allows for the identification of intermediate product features that do not have a prefix of Y.

[0100] When comparing Figure 8a and Figure 8b At that time, it can be identified There is no information regarding direct impact, but It exists and ranks third in information regarding indirect influence. This may mean... The direct impact is relatively small because it does not primarily affect process parameters that generate the final product from nearly final intermediate products, but rather has a significant impact on preceding intermediate products. Similarly, factors not included in the information about direct impacts but having a substantial portion in the information about indirect impacts may include... and Therefore, information about direct effects and information about indirect effects can be used differently, because each presents information about relatively different factors.

[0101] According to an example implementation, electronic device 100 can support control over intermediate product characteristics based on information about direct effects. Specifically, the information about direct effects can be beneficially used by product designers who wish to analyze and adjust intermediate products and effectively design battery cells. Therefore, electronic device 100 can support product designers by providing information about direct effects to product designer terminals, enabling product designers to adjust intermediate products and perform battery cell design.

[0102] According to an example implementation, electronic device 100 can support the control of process parameters for each detailed process based on information about indirect effects. Specifically, the information about indirect effects can be beneficially used by a process operator who wishes to effectively set process parameters associated with each detailed process and control quality. Therefore, electronic device 100 can support a process operator by providing information about indirect effects to the process operator's terminal, enabling the process operator to effectively control process parameters and ensure high cell quality.

[0103] According to an example implementation, the electronic device 100 can identify additional analytical information. Specifically, the electronic device 100 can identify contribution ratio information based on information about the impact, which shows the ratio of values ​​for information about direct impact and information about indirect impact for each of a plurality of factors. In other words, the electronic device 100 can calculate the ratio of direct to indirect impact in the overall battery manufacturing process by obtaining the sum of the values ​​shown for each factor by information about direct impact and information about indirect impact for each factor, and then dividing that value by the sum, and can identify the contribution ratio information accordingly. (See reference...) Figure 9 This is so that we can see an example of contribution ratio information.

[0104] Figure 9 This is an example diagram illustrating contribution ratio information according to an example implementation.

[0105] refer to Figure 9 It can identify information about process parameters. The proportion of information directly affecting process parameters is much higher. and The proportion of multiple pieces of information that indirectly influence the outcome is much higher. Based on this contribution ratio information, managers can identify situations where there is no appropriate reflection. The characteristics of intermediate products are affected, and conversely, and This can significantly impact the characteristics of the intermediate products it connects to. Furthermore, as an additional example, managers can [address the characteristics of the intermediate products connected to it]. and When a node is connected to a node corresponding to the same intermediate product feature, it can be adjusted... and To control the characteristics of the corresponding intermediate products.

[0106] The model 110 used to perform the above-described method for analyzing factors affecting battery performance can be a model trained by the electronic device 100, or it can be a pre-trained model input into the electronic device 100. An example of training model 110 by the electronic device 100 will be described below.

[0107] According to the example implementation, the electronic device 100 can input the aforementioned process data into the model 110 on a data instance basis. The electronic device 100 can identify the comparison results between the actual data corresponding to the final product characteristics identified on a data instance basis and the output of the model 110. Then, the electronic device 100 can train the artificial intelligence model 110 based on the results. For example, when building the model 110 using a DNN, the electronic device 100 can train the model 110 by calculating a loss function based on the comparison results and backpropagating the loss.

[0108] Figure 10 This is a block diagram illustrating an electronic device according to an example embodiment.

[0109] According to an example implementation, electronic device 100 may include memory 101 and processor 102. Regarding Figure 1 The illustrated electronic device 100 shows elements associated only with this exemplary embodiment. Therefore, those skilled in the art, in connection with this exemplary embodiment, will understand that, in addition to… Figure 10 In addition to the components shown, other commonly used components may also be included. In the example embodiment, the processor 102 may be included in the controller.

[0110] Processor 102 can control the overall operation of electronic device 100 and process data and signals. Processor 102 may be formed with at least one hardware unit. Furthermore, processor 102 may be operated by one or more software modules generated by executing program code stored in memory 101. Because processor 102 may include memory, processor 102 can control the overall operation of electronic device 100 and process data and signals by executing program code stored in memory.

[0111] The processor 102 can be configured to: identify a causal graph and process data by executing one or more instructions, the causal graph illustrating the relationships between multiple factors associated with the battery manufacturing process, the process data including data for the multiple factors, the process data being identified as independent data instances each time the battery manufacturing process is executed; and identify information about the impact of each of the multiple factors on the final product characteristics based on the causal graph, the process data, and a model trained to predict the final product characteristics in the battery manufacturing process by receiving the process data as input.

[0112] In some example implementations, electronic device 100 may additionally include a transceiver for performing wired / wireless communication. Electronic device 100 can communicate with external electronic devices using the transceiver. The external electronic device may be a terminal or a server. Furthermore, the communication technologies used by the transceiver may include Global System for Mobile Communications (GSM), Code Division Multiple Access (CDMA), Long Term Evolution (LTE), 5G, Wireless Local Area Network (WLAN), Wi-Fi, Bluetooth, Radio Frequency Identification (RFID), Infrared Data Association (IrDA), ZigBee, Near Field Communication (NFC), etc.

[0113] Electronic devices according to the above-described exemplary embodiments may include a processor, a memory for storing and executing program data, permanent memory such as a disk drive, a communication port for communicating with external devices, and user interface devices such as touch panels, keys, and buttons. Methods implemented by software modules or algorithms can be stored as computer-readable code or program instructions executable in a computer-readable recording medium. Here, the computer-readable recording medium may include magnetic storage media (e.g., read-only memory (ROM), random access memory (RAM), floppy disk, hard disk, etc.), optical reading media (e.g., CD-ROM or digital versatile disc (DVD)), etc.). The computer-readable recording medium can be distributed across computer systems connected via a network, allowing computer-readable code to be stored and executed in a distributed manner. The medium can be read by a computer, stored in memory, and executed by a processor.

[0114] This exemplary implementation can be represented by functional blocks and various processing steps. These functional blocks can be implemented by various numbers of hardware and / or software configurations that perform specific functions. For example, this exemplary implementation can employ integrated circuit configurations, such as memory, processors, logic circuits, and lookup tables, which can perform various functions by controlling one or more microprocessors or other control devices. Similar to elements that can be executed by software programming or software components, this exemplary implementation can be implemented by programming or scripting languages ​​such as C, C++, Java, and assembly languages, including various algorithms implemented by combinations of data structures, procedures, routines, or other programming configurations. Functional aspects can be implemented by algorithms executed by one or more processors. In addition, this exemplary implementation can employ related techniques for, for example, electronic environment setup, signal processing, and / or data processing. The terms “mechanism,” “component,” “device,” and “configuration” can be used broadly and are not limited to mechanical and physical components. These terms can include the meaning of a series of software routines associated with a processor.

[0115] The above embodiments are merely examples, and other embodiments may be implemented within the scope of the appended claims.

Claims

1. A method for analyzing factors affecting battery performance in electronic devices, the method comprising the following steps: Identify causal relationship diagrams and process data, wherein the causal relationship diagrams illustrate the relationships between multiple factors associated with the battery manufacturing process, and the process data includes data for the multiple factors, wherein the process data is identified as an independent data instance each time the battery manufacturing process is executed; as well as Based on the causal relationship diagram, the process data, and a model trained to predict the final product characteristics in the battery manufacturing process by receiving the process data as input, information is provided to identify the impact of each of the plurality of factors on the final product characteristics. The causal relationship diagram illustrates: Property nodes are connected in a unidirectional direction corresponding to the sequence of detailed processes included in the battery manufacturing process, and respectively correspond to intermediate product features and final product characteristics generated separately in the detailed processes. as well as Multiple process parameter nodes are connected in a unidirectional direction to each of the property nodes and correspond to process parameters controlled in each of the detailed processes respectively corresponding to the property nodes.

2. The method for analyzing factors affecting battery performance according to claim 1, further comprising the following steps: Based on the causal relationship graph, multiple paths are identified connecting the multiple process parameter nodes to the physical property nodes corresponding to the final product characteristics; Virtual data instances are identified by adjusting at least a portion of the data included in the process data based on the identified plurality of paths; Input the virtual data instance into the model; and Information about the impact is identified by analyzing the model's output to the virtual data instance.

3. The method for analyzing factors affecting battery performance according to claim 2, wherein, The steps for identifying the virtual data instance include the following: Based on the multiple paths, at least one first path is identified, the at least one first path comprising an edge between a first node corresponding to a first factor in the causal relationship graph and a second node corresponding to a child node of the first node; and At least one first virtual data instance is identified based on the at least one first path, and The steps for identifying information about the impact include the following: Based on the first virtual data instance and the model, identify first value information about the first path; and Based on the first value information, identify the influence value of the first factor on the second factor corresponding to the second node.

4. The method for analyzing factors affecting battery performance according to claim 3, wherein, The step of identifying the at least one first virtual data instance includes the following steps: Identify at least one path preceding the at least one first path in a plurality of random arrangements of the plurality of paths; For each of the plurality of permutations, a first subset of the at least one path preceding the at least one first path and a second subset of the first subset containing the first path are identified; and First-first virtual data instances are identified by randomly sampling and adjusting the data of nodes in the data instances that do not correspond to the first subset, and second-first virtual data instances are identified by randomly sampling and adjusting the data of nodes in the data instances that do not correspond to the second subset. The steps for identifying the first value information include the following: First-first value information and first-second value information are identified based on the results of the first-first virtual data instance and the first-second virtual data instance identified for each of the plurality of permutations input to the model; and The first value information is identified based on information about the difference between the first-first value information and the first-second value information.

5. The method for analyzing factors affecting battery performance according to claim 4, wherein, The step of identifying the first value information includes the following steps: identifying the first value information about the first path based on information about the average value of the difference between the first-first value information and the first-second value information identified for each of the plurality of permutations.

6. The method for analyzing factors affecting battery performance according to claim 4, wherein, The steps for identifying the first value information include the following: First-first output information is identified by inputting the first-first virtual data instance into the model; and The first-first value information is identified based on the difference between the average value of the output information identified by the model when the unadjusted data instances are input into it and the first-first output information.

7. The method for analyzing factors affecting battery performance according to claim 3, wherein, The step of identifying information about the influence of the first factor includes the following steps: identifying a first influence value of the first factor on the second factor based on information about the sum of first value information about the first path, the first path including the edge between the first node and the second node.

8. The method for analyzing factors affecting battery performance according to claim 1, wherein, The steps for identifying information about the impact include the following: The matrix comprising the influence values ​​of the first factor corresponding to the first coordinate on the first axis on the second factor corresponding to the second coordinate on the second axis is identified as elements; and Based on the matrix, information regarding the impact of the multiple factors on the final product characteristics is identified, including information about direct impacts and information about indirect impacts.

9. The method for analyzing factors affecting battery performance according to claim 8, wherein, The steps for identifying information about the impact include the following: Identify the influence value of at least one factor corresponding to at least one node, wherein the at least one node is directly connected in the matrix to a node corresponding to the final product characteristic; and Information about the direct impact is identified based on the impact value of the at least one factor.

10. The method for analyzing factors affecting battery performance according to claim 9, wherein, The step of identifying the influence value of the at least one factor includes the following steps: Identify the coordinates on the second axis that correspond to the final product characteristics; Identify vectors in the matrix, the vectors having coordinates on the second axis corresponding to the final product characteristics and parallel to the first axis; and The influence value of the at least one factor is identified based on the vector.

11. The method for analyzing factors affecting battery performance according to claim 8, wherein, The steps for identifying information about the impact include the following: Information about the sum of influence values ​​is identified by factors corresponding to each of the process parameter nodes; and Based on the information about the sum, information about the indirect effects is identified by factors corresponding to each of the process parameter nodes.

12. The method for analyzing factors affecting battery performance according to claim 11, wherein, The steps for identifying information about the sum include the following: Identify the coordinates on the first axis corresponding to each of the process parameters; Identify each vector in the matrix, the vector having coordinates corresponding to each of the process parameters on the first axis and parallel to the second axis; and Information about the sum of the influence values ​​of each of the factors included in the vector is identified by using factors corresponding to each of the process parameter nodes.

13. The method for analyzing factors affecting battery performance according to claim 8, further comprising the following steps: Contribution ratio information is identified based on information about the effects, which shows the ratio of the values ​​of information about the direct effects and information about the indirect effects for each of the plurality of factors.

14. The method for analyzing factors affecting battery performance according to claim 8, further comprising the following steps: Information about the direct impact is used to support control over the characteristics of the intermediate product.

15. The method for analyzing factors affecting battery performance according to claim 8, further comprising the following steps: Information about the indirect effects is used to support the control of the process parameters for each of the detailed processes.

16. The method for analyzing factors affecting battery performance according to claim 1, wherein, The steps for identifying the causal relationship diagram and the process data include the following: Identify the entire process data identified each time the battery manufacturing process is executed; Calculate the correlation coefficients between all factors in the battery manufacturing process based on the entire process data; The multiple factors are identified by filtering at least a subset of factors whose correlation coefficients are greater than or equal to a threshold; and Identify at least a portion of the data in the entire process data that corresponds to the multiple factors as the process data.

17. The method for analyzing factors affecting battery performance according to claim 16, wherein, The step of identifying at least a portion of the data corresponding to the multiple factors in the entire process data as the process data includes the following steps: identifying the data instances in the entire process data that correspond to the measured temperature and effective temperature range of the final product characteristics as the process data.

18. The method for analyzing factors affecting battery performance according to claim 1, further comprising the following steps: The process data is input into the model on a per-data-instance basis; and The model is trained based on a comparison between the actual data corresponding to the final product characteristics identified on a per-data-instance basis and the output of the model.

19. A non-transitory computer-readable recording medium having a program recorded therein for performing the method according to any one of claims 1 to 18 on a computer.

20. An electronic device for analyzing factors affecting battery performance, the electronic device comprising: processor; as well as The memory, configured to store one or more instructions, The processor is configured to execute one or more instructions: Identify causal relationship diagrams and process data, wherein the causal relationship diagrams illustrate the relationships between multiple factors associated with the battery manufacturing process, and the process data includes data for the multiple factors, wherein the process data is identified as independent data instances each time the battery manufacturing process is executed; and Based on the causal relationship diagram, the process data, and a model trained to predict the final product characteristics in the battery manufacturing process by receiving the process data as input, information is identified regarding the impact of each of the plurality of factors on the final product characteristics. The causal relationship diagram illustrates: Property nodes, connected in a unidirectional direction corresponding to the sequence of detailed processes included in the battery manufacturing process, and respectively corresponding to intermediate product features and final product characteristics generated individually in the detailed processes; and Multiple process parameter nodes are connected in a unidirectional direction to each of the property nodes and correspond to process parameters controlled in each of the detailed processes respectively corresponding to the property nodes.

Images