Battery cell capacity prediction method and apparatus, device and medium

Abstract

The present application relates to the technical field of battery cell capacity grading, in particular to a battery cell capacity prediction method and apparatus, a device and a medium. The method comprises: acquiring original data of sample battery cells during a battery cell production process and before battery cell capacity grading; according to graph rules constructed on the basis of all of the original data, training an initial prediction model; and using a trained prediction model to predict the capacity of battery cells to be subjected to capacity grading. In the present application, the graph rules and data features of battery cell process data are extracted, the graph rules can fully mine high-dimensional features of the process data, and the corresponding graph rules and the corresponding data features are used as input features of a battery cell capacity prediction model for training, thereby improving the capability of the model. The trained prediction model is used for predicting the capacity of battery cells to be subjected to capacity grading without the need for charging and discharging, and it is only necessary to extract a first data feature of said battery cells, thus improving the efficiency of industrial battery cell production processes.

Description

A method, apparatus, equipment and dielectric for predicting battery cell capacity

[0001] This application is based on and claims priority to Chinese Invention Application No. 202410923214.X, filed on July 10, 2024, entitled "A method, apparatus, device and medium for predicting battery cell capacity". Technical Field

[0002] This application relates to the field of battery cell capacity assessment technology, and in particular to a method, apparatus, equipment and medium for predicting battery cell capacity. Background Technology

[0003] With the rapid development of renewable energy and electric vehicles, lithium-ion batteries, as an important energy storage device, have been widely used. The capacity of a lithium-ion battery is one of the key indicators for measuring its performance, and digital cell capacity grading technology is crucial for accurately predicting and managing the capacity of lithium-ion batteries during the production process. Cell capacity grading refers to the process of calibrating manufactured batteries by measuring their capacity. During battery manufacturing, due to their poor stability, even batteries produced on the same production line can vary. After capacity grading, the batteries can be classified according to the calibration results, selecting batteries with the same capacity to form cell groups, improving the consistency of the cell groups. Simultaneously, the calibration results can be used to screen out unqualified batteries, improving the yield rate. Currently, the industry practice is to perform a single charge-discharge cycle on the battery to obtain the capacity grading result, known as charge-discharge capacity grading. However, the inventors of this method found that it consumes a lot of electricity and time, resulting in low efficiency. Therefore, improving the efficiency of cell capacity grading has become an urgent problem to be solved.

[0004] Summary of the Invention

[0005] In view of this, embodiments of this application provide a method, apparatus, device and medium for predicting battery cell capacity, in order to solve the problem of low capacity allocation efficiency during the battery cell capacity allocation process.

[0006] In a first aspect, embodiments of this application provide a method for predicting battery cell capacity, the method comprising:

[0007] Obtain the original data of the sample cell during the cell production process and before the cell capacity assessment, the actual capacity of the sample cell, and the initial prediction model for cell capacity prediction;

[0008] Feature extraction is performed on the original data to obtain the first data feature of the original data, and a graph rule is constructed based on the first data feature;

[0009] The first data feature is input into the map rule, and the first enhanced feature of the original data is output.

[0010] Using the first data feature and the first enhanced feature as input data, and the true capacity as the label of the input data, the initial prediction model is trained using the input data and the label to obtain a trained prediction model;

[0011] The target data of the battery cell to be allocated capacity is obtained. The second data feature extracted from the target data and the second enhanced feature extracted from the target data through the graph rules are input into the trained prediction model. The output prediction result is the capacity of the battery cell to be allocated capacity.

[0012] Secondly, embodiments of this application provide a battery cell capacity prediction device, the battery cell capacity prediction device comprising:

[0013] The acquisition module is used to acquire the original data of the sample cell during the cell production process and before the cell capacity assessment, the actual capacity of the sample cell, and the initial prediction model for cell capacity prediction.

[0014] The feature extraction module is used to extract features from the original data to obtain the first data feature of the original data, and to construct a graph rule based on the first data feature;

[0015] The output module is used to input the first data feature into the map rule and output the first enhanced feature of the original data;

[0016] The training module is used to train the initial prediction model using the first data feature and the first enhanced feature as input data, and the true capacity as the label of the input data, to obtain a trained prediction model.

[0017] The prediction module is used to acquire the target data of the battery cell to be allocated capacity. The second data feature extracted from the target data and the second enhanced feature extracted from the target data through the spectral rules are input into the trained prediction model, and the output prediction result is the capacity of the battery cell to be allocated capacity.

[0018] Thirdly, embodiments of this application provide a computer device, the computer device including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the cell capacity prediction method as described in the first aspect.

[0019] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the cell capacity prediction method as described in the first aspect.

[0020] The advantages of this application compared to the prior art are:

[0021] The process involves acquiring raw data of sample cells during cell production and before capacity allocation, the actual capacity of the sample cells, and an initial prediction model for cell capacity prediction. Feature extraction is performed on the raw data to obtain the first data feature. A graph rule is constructed based on the first data feature, and the first data feature is input into the graph rule to output the first enhanced feature of the raw data. Using the first data feature and the first enhanced feature as input data, and the actual capacity as the label, the initial prediction model is trained to obtain a trained prediction model. Target data for the cells to be allocated capacity is then acquired. The second data feature extracted from the target data and the second enhanced feature extracted through the graph rule are input into the trained prediction model, and the output prediction result is the capacity of the cells to be allocated capacity. In this application, a graph rule is constructed based on the data features extracted from the original data during the cell production process and before capacity allocation. The high-dimensional features of the original data are fully mined according to the graph rule to obtain enhanced features of the original data, thereby improving the feature accuracy of the original data. The corresponding data features and high-dimensional features are used as input features for training the cell capacity prediction model, thereby improving the training accuracy of the model. The trained prediction model is used to predict the capacity of the cell to be allocated without charging and discharging. It only needs to extract the data features and enhanced features of the target data of the cell to be allocated, thereby improving the efficiency of cell capacity allocation. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 is a schematic diagram of an application environment for a cell capacity prediction method provided in Embodiment 1 of this application;

[0024] Figure 2 is a flowchart illustrating a cell capacity prediction method provided in Embodiment 2 of this application;

[0025] Figure 3 is a flowchart illustrating a cell capacity prediction method provided in Embodiment 3 of this application;

[0026] Figure 4 is a flowchart illustrating a cell capacity prediction method provided in Embodiment 4 of this application;

[0027] Figure 5 is a flowchart illustrating a cell capacity prediction method provided in Embodiment 5 of this application;

[0028] Figure 6 is a flowchart illustrating a cell capacity prediction method provided in Embodiment Six of this application;

[0029] Figure 7 is a schematic diagram of a cell capacity prediction device provided in Embodiment 7 of this application;

[0030] Figure 8 is a schematic diagram of the structure of a computer device provided in Embodiment 8 of this application. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0033] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0034] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0035] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0036] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0037] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0038] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0039] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0040] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0041] To illustrate the technical solution of this application, specific embodiments are described below.

[0042] This application provides a cell capacity prediction method according to an embodiment, which can be applied in the application environment shown in Figure 1, wherein the local terminal communicates with the server. The local terminal includes, but is not limited to, terminal devices such as PDAs, desktop computers, laptops, ultra-mobile personal computers (UMPCs), netbooks, and personal digital assistants (PDAs). The server can be implemented using a separate server or a server cluster formed by multiple servers.

[0043] Referring to Figure 2, it is a flowchart illustrating a cell capacity prediction method provided in Embodiment 2 of this application. The cell capacity prediction method can be applied to the server in Figure 1. The server is connected to the corresponding local terminal. As shown in Figure 2, the cell capacity prediction method may include the following steps.

[0044] S201: Obtain the original data of the sample cell during the cell production process and before the cell capacity assessment, the actual capacity of the sample cell, and the initial prediction model for cell capacity prediction.

[0045] In step S201, the original data of the sample cell during the cell production process and before the cell capacity assessment, the actual capacity of the sample cell, and the initial prediction model for cell capacity prediction are obtained. The original data are the data of factors affecting cell capacity, the sample cell can be cells from different batches, and the initial prediction model for cell capacity prediction is a regression model.

[0046] In this embodiment, cell capacity prediction refers to a stage at the end of the cell production process. Before cell capacity prediction, a large amount of production data and process data before capacity assessment are generated during the cell production process, i.e., raw data. Based on the raw data of the cell before capacity prediction, the relevant status of the corresponding cell can be known, which can then be used to predict the cell capacity.

[0047] It should be noted that factors affecting cell capacity can include: raw material design, electrode design, capacity design, assembly process, formation process, and testing environment. Raw material design includes positive and negative electrode materials, separators, and electrolytes.

[0048] Obtain raw data of the sample cells during the cell production process and before cell capacity testing. The raw data may include raw material data, production data, cell preparation process data, cell formation data, and the equipment status for producing the sample cells.

[0049] Obtaining raw material data required for battery cell production includes batch numbers. This data will vary depending on the type of battery cell. For example, for lithium-ion cells, raw material data includes the composition, ratio, specific capacity, and electrolyte injection amount of the raw materials. Specific capacity refers to the ratio of the electrical capacity released by the active material within the cell to the mass of that active material. The batch number in the raw material data refers to the production batch of the corresponding raw material.

[0050] Production data generated during battery cell manufacturing can be obtained from the battery cell manufacturing system. This data may include testing data corresponding to each production step. The system can automatically collect production data in real time. For example, during battery cell manufacturing, each cell is individually numbered, and the system stores data corresponding to that number, including raw material data, manufacturing process data, and formation data. Therefore, the production data corresponding to a specific cell number can be quickly retrieved, enabling rapid data collection and storage.

[0051] Cell manufacturing process data includes coating weight, coating temperature, and core weight. Cell formation data includes formation cutoff voltage, formation temperature, and formation charge-discharge curves.

[0052] It should be noted that different production data have different effects on the cell capacity of lithium-ion cells. Generally speaking, for lithium-ion cells, the greater the core weight, the higher the corresponding cell capacity; while the electrolyte injection amount needs to reach a certain value to support the formation quality, so that a higher electrolyte injection amount results in a higher cell capacity.

[0053] For cell formation data, formation is a crucial step in the lithium-ion cell manufacturing process. Formation involves the initial charging of the cell after electrolyte injection and resting, forming a solid electrolyte interfacial film. The formation cutoff voltage and temperature significantly impact the resulting cell capacity, with different cutoff voltages producing varying thicknesses of the solid electrolyte interfacial film. Both the cutoff voltage and temperature need to be maintained within a suitable range. Too low a voltage will result in incomplete charging, while too high a voltage leads to a thicker, loose organic lithium salt layer on the outer surface of the solid electrolyte interfacial film, consuming more active lithium and causing a capacity decrease. Therefore, both excessively high and excessively low formation cutoff voltages will result in a reduction in cell capacity.

[0054] It should be noted that production data may also include necessary data and optimization data. Necessary data may include electrolyte injection amount, core weight, formation cutoff voltage and / or formation temperature, etc.; while optimization data may include raw material specific capacity, coating weight and / or formation charge-discharge curve, etc.

[0055] The equipment status for producing sample battery cells includes the status of all equipment used in the production process. For example, a coating machine is used to uniformly coat the positive and negative electrode surfaces with a slurry that has good stability, viscosity, and fluidity. The status of the coating machine is determined based on the uniformity of the coating. This determination can be based on different scores; a higher score is given for better uniformity, and a lower score for otherwise better uniformity. Other equipment is scored sequentially to determine its status.

[0056] The true capacity of a sample battery cell can be calculated through charge and discharge operations. For example, the discharge capacity method involves fully charging the cell under certain conditions, then fully discharging it with a certain current; the discharge current multiplied by the time equals the true capacity of the cell. The charging capacity method involves charging the cell to a certain SOC-1 (State of Charge, remaining charge) value under certain conditions, then charging it to a SOC-2 value using a different charging method. The charging capacity between SOC-1 and SOC-2 values ​​is calculated, and by comparing this charging capacity with the final capacity of the lithium-ion cell, the true capacity can be estimated. The open-circuit voltage method involves charging the cell to a certain SOC value using a constant current, determining the relationship between the open-circuit voltage and the discharge capacity, and then estimating the discharge capacity based on the open-circuit voltage.

[0057] The initial prediction model used for predicting battery cell capacity can be a decision tree model, support vector machine model, ensemble learning model, multilayer perceptron model, transformer model, or other machine learning or deep learning models.

[0058] S202: Extract features from the original data to obtain the first data features of the original data, and construct the graph rules based on the first data features.

[0059] In step S202, feature extraction is performed on the original data to obtain the first data feature of the original data. The first data feature is the feature of the original data in different dimensions. Graph rules are constructed based on the first data feature. The graph rules can be used to extract the relationship between various data in the original data and the high-dimensional features of the original data.

[0060] In this embodiment, feature extraction is performed on the original data to obtain the first data feature. This first data feature may include first-order difference features, statistical features, and interaction features. Graph rules are then constructed based on these first data features. When constructing graph rules, a GFD (Graph Functional Dependencies) approach can be used. In this approach, the relationships between the original data are stored in the graph rules in a graph structure. Each node in the graph rules represents a piece of original data. Graph rules refer to the rules governing the relationships between nodes and edges in the graph structure, or the patterns in the graph data. These rules can be used for business reasoning, decision-making, and verification, and can also serve as constraints. Graph rule features can be obtained through manual mining. Alternatively, graph rules can be obtained using rule mining algorithms such as SFE.

[0061] It should be noted that when using GFD-based graph construction rules, a graph G can be denoted as G = (V, E, Σ, L). Here, V = {v1, ..., vn} represents the set of vertices in the graph, and vi represents a single node; E = {e1, ..., em} represents the set of edges in the graph, and ei = (vi, vj) represents a single edge; Σ represents the set of vertex and edge labels in the graph; L represents the label mapping function, used to assign labels to vertices and edges, such as L(vi) ∈ Σ or L(ei) ∈ Σ. A graph pattern Q is denoted as Q[μ] = (VQ, EQ, ΣQ, LQ). Where μ is the set of variables, including all variables in the graph pattern; VQ = {v1,...,vn} represents the set of nodes in the graph pattern, and vi represents a single node; EQ = {e1,...,em} represents the set of edges in the graph pattern, and ei = (vi,vj) represents a single edge; ΣQ represents the set of node labels and edge labels in the graph pattern; LQ represents the label mapping function, used to assign corresponding labels to edges, denoted as LQ(ei)∈ΣQ. Graph approximate functional dependency is an extension of graph functional dependency in terms of semantic constraints. From the perspective of constraint conditions, a graph approximate functional dependency can be regarded as a GFD with some relaxation of constraints. At the same time, in order to better express the constraints between knowledge graph data, the graph pattern in the graph dependency is changed from the attribute graph model to the RDF (Resource Description Framework) graph model.

[0062] Referring to Figure 3, which is a flowchart illustrating a cell capacity prediction method provided in Embodiment 3 of this application, as shown in Figure 3, the feature extraction of the original data in step S202 to obtain the first data feature of the original data may include the following steps:

[0063] S301: Extract first-order difference features from the original data to obtain the first-order difference features of the original data;

[0064] S302: Extract statistical features from the original data to obtain the statistical features of the original data;

[0065] S303: Extract interactive features from the original data to obtain the interactive features of the original data;

[0066] S304: First-order difference features, statistical features, and interaction features are determined as the first data features of the original data.

[0067] In this embodiment, first-order difference feature extraction is performed on the original data to obtain the first-order difference features of the original data. Before performing first-order difference feature extraction, it is necessary to extract data based on time-series features from the original data. Based on the time-series features, the first-order difference of the time-series features is calculated, that is, the data in the original data that changes due to time, with time as the variable. The first-order difference feature is the feature data obtained by differentiating the time variable.

[0068] Statistical feature extraction is performed on the raw data to obtain its statistical features. These statistical features include quantitative features and attribute features. Quantitative features can be further divided into measurement features and count features. Measurement features refer to characteristics that can be measured with specific numerical values, such as the amount of electrolyte injected into a battery cell or the capacity of the battery cell. Count features describe the frequency or quantity of certain events occurring in a set of data, such as the number of times electrolyte is injected into a battery cell within a certain time period. Attribute features usually cannot be directly expressed by numerical values; they represent the basic attributes and categories of things. Statistical features need to be obtained through data analysis, rather than simple mathematical calculations.

[0069] Interaction features are extracted from the raw data to obtain its interaction characteristics. This extraction can be performed using a self-attention model, which is built upon a self-attention network and used to learn the interaction characteristics between various data features within the raw data. The self-attention network employs a self-attention mechanism, automatically learning higher-order interaction characteristics between features. This attention mechanism is inspired by human attention, focusing attention on key areas while ignoring other information.

[0070] The specific structure of the self-attention network layer can be set according to actual needs, such as being constructed based on one or more combinations of multi-head self-attention layers, residual layers, and regularization layers. This embodiment does not specifically limit this.

[0071] S203: Input the first data feature into the map rule and output the first enhanced feature of the original data.

[0072] In step S203, the first enhanced feature of the original data is obtained through the graph rules. The first enhanced feature is the relational feature representing the original data and the high-dimensional feature corresponding to the original data.

[0073] In this embodiment, the first data feature is input into the spectrum rules, and the first enhanced feature of the original data is output. Each sample cell can obtain a corresponding first enhanced feature.

[0074] S204: Using the first data feature and the first enhanced feature as input data, and the true volume as the label of the input data, the initial prediction model is trained using the input data and the label to obtain the trained prediction model.

[0075] In step S204, the first data feature and the first enhancement feature of each sample cell are input into the initial prediction model as input data, the predicted capacity of the corresponding cell is output, the loss is calculated by combining the predicted capacity with the label of the corresponding cell, the parameters of the initial prediction model are adjusted according to the loss, and the trained prediction model is obtained after training.

[0076] In this embodiment, the first data feature and the first enhanced feature are used as input data, and the true capacity is used as the label of the input data. The initial prediction model is trained using the input data and labels to obtain a trained prediction model. During training, the mean squared error loss function can be used to calculate the model loss. During training, the parameters of the initial prediction model are adjusted according to the model loss, and this process is repeated multiple times to reduce the deviation between the predicted capacity output by the initial prediction model and the label. Training stops when the deviation is less than a preset deviation value or the number of iterations reaches a preset number, thus obtaining a trained model.

[0077] It should be noted that the initial prediction model can be various neural networks, such as CNN (Convolutional Networks), RNN (Recurrent Neural Networks), DNN (Deep Neural Networks), etc., or various decision tree networks, such as CART (Classification and Regression Tree), XGBoost (eXtremeGradient Boosting), etc. This embodiment does not limit the type of the initial prediction model.

[0078] Referring to Figure 4, which is a flowchart illustrating a cell capacity prediction method provided in Embodiment 4 of this application, as shown in Figure 4, before step S204, which uses the first data feature and the first enhanced feature as input data and the actual capacity as the label of the input data, to train the initial prediction model using the input data and the label, and to obtain the trained prediction model, the following steps may be included:

[0079] S401: Perform outlier detection on the true capacity of the sample cell to obtain the outlier values ​​of the true capacity of the sample cell;

[0080] S402: Based on outliers, perform initial screening of sample cells to obtain initially screened sample cells.

[0081] In this embodiment, the sample cells undergo initial screening, during which sample cells whose capacity does not meet the requirements are removed. Screening is further performed by outlier detection on the actual capacity of the sample cells. Outlier detection can utilize clustering detection, unsupervised model detection, or semi-supervised model detection. In this embodiment, clustering detection is used, with the actual capacity that cannot be clustered treated as discrete values. Sample cells corresponding to these discrete values ​​are then removed, resulting in the initially screened sample cells.

[0082] Optionally, after initially screening the sample cells based on outliers and obtaining the initially screened sample cells, the process further includes:

[0083] Based on the first data feature and the first enhancement feature in the initially screened sample cells, the initially screened sample cells are classified to obtain classified normal sample cells, which are then used to train the initial prediction model.

[0084] In this embodiment, the sample cells after initial screening are screened and classified according to the first data features and the first enhancement features in the sample cells after initial screening, and the classified normal sample cells are obtained. When screening and classifying the sample cells after initial screening, a classification model can be used for classification.

[0085] It should be noted that before using the classification model, it needs to be trained. When training the classification model, the first data feature and the first enhancement feature in the original data can be used to train the classification model to obtain a trained classification model. The trained classification model is then used to screen and classify the sample cells after the initial screening to obtain the classified normal sample cells. The normal sample cells are used to train the initial prediction model.

[0086] Referring to Figure 5, which is a flowchart illustrating a cell capacity prediction method provided in Embodiment 5 of this application, as shown in Figure 5, before step S204, which uses the first data feature and the first enhanced feature as input data and the actual capacity as the label of the input data, to train the initial prediction model using the input data and labels to obtain the trained prediction model, the following steps may be included:

[0087] S501: Obtain the historical capacity of cells produced in historical time periods prior to the sample cell's production time;

[0088] S502: Based on historical capacity and the actual capacity of sample cells, detect whether there is a trend in the actual capacity of sample cells;

[0089] S503: If a trend exists, then based on the trend, obtain a new prediction model that matches the trend, and use the new prediction model as the initial prediction model.

[0090] In this embodiment, before training the initial prediction model and obtaining the trained prediction model, the historical capacity of battery cells produced in historical time periods prior to the production time of the sample battery cells is obtained. Based on the historical capacity and the actual capacity of the sample battery cells, it is determined whether the actual capacity of the sample battery cells exhibits a trend based on the historical capacity, i.e., whether the actual capacity of the sample battery cells exhibits a certain characteristic distribution. If a trend exists, i.e., a specific distribution exists, a new prediction model matching the trend is obtained, and this new prediction model is used as the initial prediction model. If no trend exists, i.e., no specific distribution exists, the obtained initial prediction model is used for training.

[0091] It should be noted that detecting whether there is a trend in the actual capacity of the sample cells is to ensure that the prediction model can adapt to the capacitance distribution of the sample cells and improve the prediction accuracy of the prediction model.

[0092] Referring to Figure 6, which is a flowchart illustrating a cell capacity prediction method provided in Embodiment Six of this application, as shown in Figure 6, step S203 uses input data and labels to train an initial prediction model to obtain a trained prediction model, which may include the following steps:

[0093] S601: Obtain the base learner and meta-learner in the initial prediction model;

[0094] S602: Using input data and labels, train the base learner to obtain a trained base learner;

[0095] S603: Input the input data into the trained base learner and output the first prediction result;

[0096] S604: Based on the first prediction result and the label, train the meta-learner to obtain the trained meta-learner;

[0097] S605: Based on the trained base learner and the trained meta learner, a trained prediction model is obtained.

[0098] In this embodiment, the initial prediction model is an ensemble learning model, which may include a base learner and a meta-learner. An ensemble learning model integrates multiple models through a certain strategy to obtain better generalization performance and higher decision accuracy by utilizing group decision-making. Commonly used ensemble strategies include weighted averaging and direct averaging.

[0099] In this embodiment, to improve the generalization ability of the prediction model, multiple base learners can be constructed and trained. Each base learner is trained using the first data feature and the first enhancement feature, resulting in multiple trained base learners. The training process can be based on cross-validation. Cross-validation typically divides the sample cells into N subsets and trains the initial prediction model N times. In each round of training, N-1 training subsets are used to train the initial prediction model, and one test subset is used to validate it. For example, the dataset can be divided into 10 subsets, with 9 subsets used as the training set and 1 subset as the test set, and the test set differing for each round. The base learners can be XGBoost, LightGBM, CatBoost, GBDT, RF, SVR, Ridge, Lasso, and MLP models. The specific algorithm used to construct each base learner can be flexibly selected based on the actual situation.

[0100] From multiple trained base learners, select at least two trained target base learners. There are several ways to select trained base learners, such as selecting based on the loss value of each base learner in the last round of training, to obtain at least two trained target base learners.

[0101] The meta-learner is trained using the output data of the target base learner during the training process to obtain a trained prediction model. The trained prediction model includes at least two trained target base learners and a trained meta-learner. Based on the trained base learners and the trained meta-learner, the final trained prediction model is obtained.

[0102] S205: Obtain the target data of the battery cell to be rated, input the second data feature extracted from the target data and the second enhanced feature of the target data extracted through the graph rules into the trained prediction model, and obtain the output prediction result as the capacity of the battery cell to be rated.

[0103] In step S205, target data of the cells to be classified are obtained, wherein the cells to be classified are cells with unknown capacity. The second data features extracted from the target data and the second enhanced features of the target data extracted through the graph rules are input into the trained prediction model to obtain the output prediction result as the capacity of the cells to be classified, so as to classify the cells to be classified into different capacity categories according to the capacity of the cells.

[0104] In this embodiment, the method for obtaining the target data of the battery cell to be capacity-rated is the same as the method for obtaining the original data of the sample battery cell during the battery cell production process and before capacity-rated battery cell, and will not be described again. The second feature data obtained by feature extraction of the target data and the second enhanced feature of the target data extracted by the graph rules are used as input data and input into the trained prediction model to obtain the output prediction result as the capacity of the battery cell to be capacity-rated.

[0105] The process involves acquiring raw data of sample cells during cell production and before capacity allocation, the actual capacity of the sample cells, and an initial prediction model for cell capacity prediction. Feature extraction is performed on the raw data to obtain the first data feature. A graph rule is constructed based on the first data feature, and the first data feature is input into the graph rule to output the first enhanced feature of the raw data. Using the first data feature and the first enhanced feature as input data, and the actual capacity as the label, the initial prediction model is trained to obtain a trained prediction model. Target data for the cells to be allocated capacity is then acquired. The second data feature extracted from the target data and the second enhanced feature extracted through the graph rule are input into the trained prediction model, and the output prediction result is the capacity of the cells to be allocated capacity. In this application, a graph rule is constructed based on the data features extracted from the original data during the cell production process and before capacity allocation. The high-dimensional features of the original data are fully mined according to the graph rule to obtain enhanced features of the original data, thereby improving the feature accuracy of the original data. The corresponding data features and high-dimensional features are used as input features for training the cell capacity prediction model, thereby improving the training accuracy of the model. The trained prediction model is used to predict the capacity of the cell to be allocated without charging and discharging. It only needs to extract the data features and enhanced features of the target data of the cell to be allocated, thereby improving the efficiency of cell capacity allocation.

[0106] Please refer to Figure 7, which is a schematic diagram of a battery cell capacity prediction device provided in Embodiment 7 of this application. For details, please refer to Figures 2-6 and the relevant descriptions in the embodiments corresponding to Figures 2-6. For ease of explanation, only the parts relevant to this embodiment are shown. As shown in Figure 7, the battery cell capacity prediction device 70 includes: an acquisition module 71, a feature extraction module 72, an output module 73, a training module 74, and a prediction module 75.

[0107] The acquisition module 71 is used to acquire the original data of the sample cell during the cell production process and before the cell capacity assessment, the actual capacity of the sample cell, and the initial prediction model for cell capacity prediction.

[0108] Feature extraction module 72 is used to extract features from the original data to obtain the first data features of the original data, and to construct graph rules based on the first data features;

[0109] Output module 73 is used to input the first data features into the map rules and output the first enhanced features of the original data;

[0110] Training module 74 is used to train the initial prediction model using the first data feature and the first enhanced feature as input data and the true volume as the label of the input data, and to obtain the trained prediction model.

[0111] The prediction module 75 is used to obtain the target data of the battery cell to be allocated capacity. The second data feature extracted from the target data and the second enhanced feature of the target data extracted by the graph rules are input into the trained prediction model, and the output prediction result is the capacity of the battery cell to be allocated capacity.

[0112] Optionally, the feature extraction module 72 includes:

[0113] The first extraction unit is used to extract first-order difference features from the original data to obtain the first-order difference features of the original data.

[0114] The second extraction unit is used to extract statistical features from the original data to obtain the statistical features of the original data.

[0115] The third extraction unit is used to extract interactive features from the original data to obtain the interactive features of the original data.

[0116] The determination unit is used to determine the first difference feature, statistical feature and interaction feature as the first data feature of the original data.

[0117] Optionally, the cell capacity prediction device 70 also includes:

[0118] The detection module is used to detect outliers in the true capacity of the sample cells and obtain the outliers in the true capacity of the sample cells.

[0119] The screening module is used to perform initial screening of sample cells based on outliers, and obtain the initially screened sample cells.

[0120] Optionally, the cell capacity prediction device 70 also includes:

[0121] The classification module is used to classify the initially screened sample cells based on the first data feature and the first enhancement feature, and obtain the classified normal sample cells. The normal sample cells are used to train the initial prediction model.

[0122] Optionally, the cell capacity prediction device 70 also includes:

[0123] The historical capacity acquisition module is used to acquire the historical capacity of cells produced in historical time periods prior to the production time of the sample cells.

[0124] The judgment module is used to detect whether there is a trend in the actual capacity of the sample cells based on the historical capacity and the actual capacity of the sample cells.

[0125] The new prediction model acquisition module is used to acquire a new prediction model that matches the trend if a trend exists, and use the new prediction model as the initial prediction model.

[0126] Optionally, training module 74 includes:

[0127] The acquisition unit is used to acquire the base learners and meta-learners in the initial prediction model.

[0128] The first training unit is used to train the base learner using input data and labels to obtain a trained base learner.

[0129] The output unit is used to input the input data into the trained base learner and output the first prediction result.

[0130] The second training unit is used to train the meta-learner based on the first prediction result and the label, so as to obtain the trained meta-learner.

[0131] The obtained unit is used to obtain the trained prediction model based on the trained base learner and the trained meta learner.

[0132] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0133] Figure 8 is a schematic diagram of a computer device provided in Embodiment 8 of this application. As shown in Figure 8, the computer device of this embodiment includes: at least one processor (only one is shown in Figure 8), a memory, and a computer program stored in the memory and executable on at least one processor. When the processor executes the computer program, it implements the steps in any of the above-described embodiments of the cell capacity prediction method.

[0134] The computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that Figure 8 is merely an example of a computer device and does not constitute a limitation thereof. The computer device may include more or fewer components than illustrated, or a combination of certain components, or different components; for example, it may also include a network interface.

[0135] The processor referred to can be a CPU, but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0136] Memory includes readable storage media, internal memory, etc., wherein internal memory can be the RAM of a computer device, providing an environment for the operation of the operating system and computer-readable instructions stored in the readable storage media. The readable storage media can be the hard disk of a computer device, or in some embodiments, the external storage device of the computer device. The readable storage media can be non-volatile or volatile, such as plug-in hard disks, smart media cards (SMC), secure digital cards (SD), flash cards, etc., equipped on a computer device. Furthermore, memory can include both internal storage units and external storage devices of a computer device. Memory is used to store the operating system, applications, bootloader, data, and other programs, such as program code of computer programs. Memory can also be used to temporarily store data that has been output or will be output.

[0137] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above method embodiments. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium can include at least: any entity or device capable of carrying computer program code, a recording medium, a computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0138] The implementation of all or part of the processes in the methods of the above embodiments can also be accomplished by a computer program product. When the computer program product is run on a computer device, it enables the computer device to execute the steps in the above method embodiments.

[0139] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0140] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0141] In the embodiments provided in this application, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0142] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0143] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for predicting battery cell capacity, wherein, The cell capacity prediction method includes: Obtain the original data of the sample cell during the cell production process and before the cell capacity assessment, the actual capacity of the sample cell, and the initial prediction model for cell capacity prediction; Feature extraction is performed on the original data to obtain the first data feature of the original data, and a graph rule is constructed based on the first data feature; The first data feature is input into the map rule, and the first enhanced feature of the original data is output. Using the first data feature and the first enhanced feature as input data, and the true capacity as the label of the input data, the initial prediction model is trained using the input data and the label to obtain a trained prediction model; The target data of the battery cell to be allocated capacity is obtained. The second data feature extracted from the target data and the second enhanced feature extracted from the target data through the graph rules are input into the trained prediction model. The output prediction result is the capacity of the battery cell to be allocated capacity.

2. The cell capacity prediction method as described in claim 1, wherein, The step of extracting features from the original data to obtain the first data feature of the original data includes: First-order difference features are extracted from the original data to obtain the first-order difference features of the original data; Statistical features are extracted from the original data to obtain the statistical features of the original data; Interaction features are extracted from the original data to obtain the interaction features of the original data; The first-order difference feature, the statistical feature, and the interaction feature are determined as the first data feature of the original data.

3. The cell capacity prediction method as described in claim 1, wherein, Before training the initial prediction model using the first data feature and the first enhanced feature as input data, and the true capacity as the label of the input data, to obtain the trained prediction model, the method further includes: Outlier detection is performed on the actual capacity of the sample battery cell to obtain the outlier value of the actual capacity of the sample battery cell; Based on the outlier values, the sample cells are initially screened to obtain the initially screened sample cells.

4. The cell capacity prediction method as described in claim 3, wherein, After initially screening the sample cells based on the outlier values ​​to obtain the initially screened sample cells, the method further includes: Based on the first data feature and the first enhancement feature in the initially screened sample cells, the initially screened sample cells are screened and classified to obtain classified normal sample cells, which are used to train the initial prediction model.

5. The cell capacity prediction method as described in claim 1, wherein, Before training the initial prediction model using the first data feature and the first enhanced feature as input data, and the true capacity as the label of the input data, to obtain the trained prediction model, the method further includes: Obtain the historical capacity of battery cells produced during a historical period prior to the production time of the sample battery cell; Based on the historical capacity and the actual capacity of the sample cells, detect whether there is a trend in the actual capacity of the sample cells; If a trend exists, a new prediction model matching the trend is obtained, and the new prediction model is used as the initial prediction model.

6. The cell capacity prediction method as described in claim 1, wherein, The step of training the initial prediction model using the input data and the labels to obtain a trained prediction model includes: Obtain the base learner and meta-learner in the initial prediction model; The base learner is trained using the input data and the labels to obtain a trained base learner. The input data is fed into the trained base learner, and a first prediction result is output. Based on the first prediction result and the label, the meta-learner is trained to obtain a trained meta-learner. Based on the trained base learner and the trained meta learner, a trained prediction model is obtained.

7. A battery cell capacity prediction device, wherein, The cell capacity prediction device includes: The acquisition module is used to acquire the original data of the sample cell during the cell production process and before the cell capacity assessment, the actual capacity of the sample cell, and the initial prediction model for cell capacity prediction. The feature extraction module is used to extract features from the original data to obtain the first data feature of the original data, and to construct a graph rule based on the first data feature; The output module is used to input the first data feature into the map rule and output the first enhanced feature of the original data; The training module is used to train the initial prediction model using the first data feature and the first enhanced feature as input data, and the true capacity as the label of the input data, to obtain a trained prediction model. The prediction module is used to acquire the target data of the battery cell to be allocated capacity. The second data feature extracted from the target data and the second enhanced feature extracted from the target data through the spectral rules are input into the trained prediction model, and the output prediction result is the capacity of the battery cell to be allocated capacity.

8. The cell capacity prediction device as described in claim 7, wherein, The feature extraction module includes: The first extraction unit is used to extract first-order difference features from the original data to obtain the first-order difference features of the original data. The second extraction unit is used to extract statistical features from the original data to obtain the statistical features of the original data. The third extraction unit is used to extract interactive features from the original data to obtain the interactive features of the original data. The determining unit is used to determine the first-order difference feature, the statistical feature, and the interaction feature as the data features of the original data.

9. A computer device, wherein, The computer device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the cell capacity prediction method as described in claims 1 to 6.

10. A computer-readable storage medium storing a computer program, wherein, When the computer program is executed by the processor, it implements the cell capacity prediction method as described in claims 1 to 6.

Images