Retired lithium battery recycling process screening method based on table-based large model

Abstract

The application claims a retired lithium battery recycling process screening method based on a table large model, belonging to the technical field of lithium ion battery resource utilization. By obtaining experimental data under various battery systems and recycling process conditions, an initial data set is constructed. After preprocessing the data set, it is divided into a support set and a query set. Taking battery characteristics and process parameters as input, metal leaching rate, recycling cost and greenhouse gas emission as prediction labels, a table large model is constructed and screened. The model is used to predict candidate recycling processes, and a three-objective optimization function is constructed. Through Pareto optimization, a candidate process set is obtained. According to the decision weight, the recommended process scheme is determined and the corresponding retired lithium battery recycling process parameters are output. The application breaks through the limitations of single process or single metal system, realizes high-throughput screening and multi-objective collaborative optimization across systems and processes, and provides efficient and reliable technical support for the industrialization decision of retired lithium battery recycling process.

Description

Technical Field

[0001] This invention seeks protection for a screening method for the recycling process of decommissioned lithium batteries based on a large table model, belonging to the field of lithium-ion battery resource utilization technology. Background Technology

[0002] Driven by global emissions reduction targets and the trend of transportation electrification, the electric vehicle industry is expanding rapidly, leading to a simultaneous increase in both the number of installed and retired power batteries. The content of key metals such as lithium, nickel, cobalt, and manganese in these retired lithium batteries far exceeds that of primary ores. If these materials are not properly disposed of, they will not only increase the burden on the ecological environment and endanger public health, but also result in the waste and loss of key metal resources. Therefore, achieving low-carbon, low-cost, and multi-element efficient leaching not only has significant industrial application value but also helps reduce the environmental burden, while simultaneously achieving a synergistic improvement in resource and environmental benefits.

[0003] Achieving low cost and low greenhouse gas emissions while improving the efficiency of multi-metal synergistic leaching is crucial. Currently, optimizing recycling processes to achieve high metal leaching rates still largely relies on traditional empirical experiments. However, this method involves high trial-and-error costs, long experimental cycles, and complex factors affecting leaching. Against this backdrop, the ability to quickly and reliably screen and evaluate recycling processes for retired lithium-ion batteries that offer higher economic efficiency and lower environmental impact has become a prerequisite for supporting a circular economy in the battery industry and ensuring a secure supply of critical metals.

[0004] In recent years, machine learning has been introduced into the process modeling and optimization of recycling processes. It promises to achieve high-throughput screening and parameter optimization of a large number of candidate conditions while significantly reducing trial-and-error experiments, thereby guiding process window control and improving key performance indicators. For example, Chinese patent CN115760077A discloses a method for predicting the leaching rate of valuable metals from spent lithium batteries based on machine learning. However, its applicability is limited; this method only predicts the leaching rate of valuable metals in retired lithium batteries in hydrometallurgical scenarios where acid leaching is the only leaching method, ignoring the recycling cost and greenhouse gas emissions of the process. Summary of the Invention

[0005] The practical value of existing machine learning research is often limited by three key issues: First, training data often comes from a single process route or a single battery system, or even only models the leaching rate of a single metal, resulting in insufficient generalization ability of the model when applied across processes and systems. Second, model selection has long been concentrated on a few classic machine learning methods. While these methods have strong nonlinear fitting capabilities, they still lack representation learning that can be shared across tasks and transferable inductive biases, and it is difficult to form stable consistency constraints between subdomains with significant structural differences. Third, most studies often aim to pursue high leaching rates, neglecting economic and environmental impacts. Therefore, there is an urgent need to develop a general model applicable to different battery systems and different leaching processes, which can predict metal leaching effects while assessing recycling costs and greenhouse gas emissions. This would overcome the limitations of existing methods due to battery system and process type, reduce trial-and-error costs in the process parameter selection, and provide a reliable evaluation basis for the industrial scale-up of retired lithium-ion battery recycling processes.

[0006] The main problem addressed by this application is that existing machine learning methods, when applied to the recycling process of decommissioned lithium batteries, have insufficient generalization ability, a single evaluation objective, and ignore cost and carbon emissions, making it difficult to support industrial-scale decision-making.

[0007] To achieve the above objectives, this application provides the following technical solution: A method for screening decommissioned lithium battery recycling processes based on a large tabular model, characterized by comprising: S1. Obtain a candidate recycling process set consisting of multiple candidate recycling processes, extract the process data of the candidate recycling process set, and construct an initial dataset; S2, preprocess the initial dataset to obtain the final dataset, and divide it into a support set and a query set; S3, determine the input features and predicted labels, and construct and filter the large table model by combining the support set and query set; S4. Based on the above large table model, obtain the prediction results of the candidate recycling process set; S5, construct an optimization function based on the prediction results, and obtain a Pareto candidate process set from the candidate recovery process set based on the optimization function; S6. Determine the recommended process scheme from the Pareto candidate process set, and output the corresponding retired lithium battery recycling process parameters through the graphical user interface.

[0008] Furthermore, S1 also includes: Collect experimental data on the recycling of retired lithium-ion batteries under various battery systems and recycling process conditions; Battery characteristics, process parameters, and leaching rates of at least one metal among lithium, nickel, cobalt, and manganese were extracted, and the corresponding recycling costs and greenhouse gas emissions were calculated using the EverBatt2023 model to construct an initial dataset. The battery system includes one or more of LCO, LFP, LMO, LMFP, NCA, NMC111, NMC333, NMC532, NMC622, NMC811 and their hybrid systems; The recycling process conditions include one or more of the following: hydrometallurgical process, pyrometallurgical process, pyrometallurgical-hydrometallurgical combined process, and direct recycling process.

[0009] Furthermore, in step S1, the battery characteristics include one or more of the following: battery type, positive electrode material system, lithium mass fraction, nickel mass fraction, cobalt mass fraction, manganese mass fraction, and physical form. The process parameters include one or more of the following: recovery process route, leaching or extraction method, leachate type, leachate concentration, reaction temperature, reaction time, pH, solid-liquid ratio, voltage, stirring speed, and reagent dosage. In step S1, the EverBatt2023 model is used to calculate the recycling cost and greenhouse gas emissions of each recycling process based on the battery system, metal composition, process route, reagent consumption and energy consumption parameters, taking 1 kg of retired lithium-ion battery black powder as the benchmark, and using these as target labels.

[0010] Furthermore, S2 also includes: The initial dataset is subjected to field unification, feature cleaning, missing value handling, continuous variable numericalization, and categorical variable encoding to obtain the final dataset, which is then divided into a support set and a query set. In step S2, the unification process includes field unification, unit conversion, feature cleaning, missing value processing, continuous variable numericalization, categorical variable encoding, and removal of target missing samples.

[0011] Furthermore, in step S2, the final dataset is divided into a support set and a query set in an 8:2 ratio, wherein the support set is used to provide contextual priors for the large table model, and the query set is used to evaluate the generalization prediction performance of the model.

[0012] Furthermore, S3 also includes: Using battery characteristics and process parameters as input features, and metal leaching rate, recycling cost and greenhouse gas emissions as prediction labels, a large tabular model prediction task is constructed based on the support set and query set. The prediction performance of the model is evaluated by comparing it with a traditional machine learning model. In step S3, the traditional machine learning model includes one or more of Random Forest, LightGBM, AdaBoost, and XGBoost; Metrics for evaluating the predictive performance of a model include mean absolute error (MAE) and coefficient of determination (R²). 2 One or more of the following: mean absolute percentage error (MAPE).

[0013] Furthermore, S4 also includes: Based on the tabular large model, the predicted results of metal leaching rate, recycling cost and greenhouse gas emissions of candidate recycling processes are obtained, and the key variables and their interactions are analyzed by SHAP, PDP or 2D-PDP. In step S4, SHAP, PDP or 2D-PDP are used to interpret the large table model to obtain the key variables affecting metal leaching rate, recycling cost and greenhouse gas emissions and their interactions. The SHAP analysis includes global key variable impact analysis and process route key variable impact analysis.

[0014] Furthermore, S5 also includes: The effective metal set is determined based on the cathode material composition. The minimum value of the predicted leaching rate of the effective metal is taken as the minimum guaranteed leaching rate to be maximized. Together with minimizing recycling costs and minimizing greenhouse gas emissions, a three-objective optimization function is constructed. The Pareto candidate process set is obtained through NSGA-II.

[0015] Furthermore, the method also includes: In step S5, the three-objective optimization function includes maximizing the guaranteed leaching rate, minimizing recycling costs, and minimizing greenhouse gas emissions. Lithium is included in the effective metal set by default, while nickel, cobalt, and manganese are included in the effective metal set when their mass fraction is greater than zero. The minimum value among the predicted leaching rates of each effective metal is used as the guaranteed leaching rate to be maximized. Furthermore, the method also includes: In steps S5 and S6, lithium is included in the effective metal set by default, and nickel, cobalt, and manganese are included in the effective metal set when their mass fraction is greater than zero. The minimum predicted leaching rate of each effective metal is taken as the minimum guaranteed leaching rate to be maximized. The minimum guaranteed leaching rate, recycling cost and greenhouse gas emissions in the Pareto candidate process set are normalized. The comprehensive decision score is calculated according to the preset decision weight to obtain the recommended process scheme of leaching rate priority, cost priority, low carbon priority or three-objective balance. The battery system, recycling process parameters, metal leaching rate, recycling cost and greenhouse gas emissions are output through the graphical operation interface.

[0016] This invention, based on reported experimental data from different retired lithium-ion battery systems under various recycling process conditions, extracts battery characteristics, process parameters, and metal leaching rate information. It then uses the EverBatt2023 model to calculate the recycling cost and greenhouse gas emissions corresponding to each process group. Building upon this, a large-scale tabular model prediction framework is constructed, and the predicted results of the combined leaching rate, recycling cost, and greenhouse gas emissions of valuable metals from different retired batteries are simultaneously output through a visual interface. Compared to traditional leaching experiments that require extensive trial-and-error optimization for different battery properties, and the limitations of existing machine learning methods that are often confined to a single process or a single leaching rate target, making it difficult to balance economic benefits and environmental impact, this invention only requires obtaining battery characteristic information and inputting the target battery to be processed. The model will recommend a set of recycling process parameters that achieve the highest leaching rate, lowest process operating cost, and lowest greenhouse gas emissions through the synergistic leaching of multiple metals. This invention reduces repetitive experiments and parameter exploration during process selection, lowers time, manpower, and material consumption, and is applicable to process selection needs in different industrial scenarios. It is an important tool for assisting in the optimization of retired lithium-ion battery recycling processes. This invention is not limited to a single leaching process or a single battery system, but can cover mainstream retired lithium battery systems and recycling processes on the market. It has cross-system and cross-process generalization capabilities, providing a new technical path for the scale-up of waste battery recycling processes and the efficient leaching of valuable metals. At the same time, this method can provide relevant recycling companies with more comprehensive information support on efficiency, cost, and environmental impact, improving the economics of process decisions and the overall benefits to enterprises while reducing environmental impact. Attached Figure Description

[0017] Figure 1 The flowchart illustrates a screening method for decommissioned lithium battery recycling processes based on a large table model, as claimed in an embodiment of the present invention. Figure 2 The prediction results of the lithium leaching efficiency prediction model of the decommissioned lithium battery recycling process screening method based on the large table model claimed in the embodiments of the present invention on the test set. Figure 3 The prediction results of the nickel leaching efficiency prediction model of the decommissioned lithium battery recycling process screening method based on the large table model claimed in the embodiments of the present invention on the test set; Figure 4 The prediction results of the cobalt leaching efficiency prediction model of the decommissioned lithium battery recycling process screening method based on the large table model claimed in the embodiments of the present invention are shown in the test set. Figure 5 The prediction results of the manganese leaching efficiency prediction model of the decommissioned lithium battery recycling process screening method based on the large table model claimed in the embodiments of the present invention are shown in the test set. Figure 6The prediction results of the recycling cost prediction model of the decommissioned lithium battery recycling process screening method based on the large table model claimed in the embodiments of the present invention on the test set; Figure 7 The greenhouse gas emission prediction model of the decommissioned lithium battery recycling process screening method based on a large table model, as claimed in the embodiments of this invention, is shown in the test set prediction results. Figure 8 The image shows a three-objective screening diagram for a decommissioned lithium battery recycling process screening method based on a large table model, as claimed in an embodiment of the present invention. Detailed Implementation

[0018] 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 a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0019] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0020] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0021] According to a first embodiment of the present invention, the present invention claims protection for a screening method for the recycling process of decommissioned lithium batteries based on a large table model, referring to... Figure 1 ,include: S1. Obtain a candidate recycling process set consisting of multiple candidate recycling processes, extract the process data of the candidate recycling process set, and construct an initial dataset; S2, preprocess the initial dataset to obtain the final dataset, and divide it into a support set and a query set; S3, determine the input features and predicted labels, and construct and filter the large table model by combining the support set and query set; S4. Based on the above large table model, obtain the prediction results of the candidate recycling process set; S5, construct an optimization function based on the prediction results, and obtain a Pareto candidate process set from the candidate recovery process set based on the optimization function; S6. Determine the recommended process scheme from the Pareto candidate process set, and output the corresponding retired lithium battery recycling process parameters through the graphical user interface.

[0022] Furthermore, S1 also includes: Collect experimental data on the recycling of retired lithium-ion batteries under various battery systems and recycling process conditions; Battery characteristics, process parameters, and leaching rates of at least one metal among lithium, nickel, cobalt, and manganese were extracted, and the corresponding recycling costs and greenhouse gas emissions were calculated using the EverBatt2023 model to construct an initial dataset. The battery system includes one or more of LCO, LFP, LMO, LMFP, NCA, NMC111, NMC333, NMC532, NMC622, NMC811 and their hybrid systems; The recycling process conditions include one or more of the following: hydrometallurgical process, pyrometallurgical process, pyrometallurgical-hydrometallurgical combined process, and direct recycling process.

[0023] In this embodiment, the model inputs battery NMC532, and the model selects the optimal process parameters that take into account the multi-metal leaching rate, recycling cost and greenhouse gas emissions.

[0024] Furthermore, in step S1, the battery characteristics include one or more of the following: battery type, positive electrode material system, lithium mass fraction, nickel mass fraction, cobalt mass fraction, manganese mass fraction, and physical form. The process parameters include one or more of the following: recovery process route, leaching or extraction method, leachate type, leachate concentration, reaction temperature, reaction time, pH, solid-liquid ratio, voltage, stirring speed, and reagent dosage. In step S1, the EverBatt2023 model is used to calculate the recycling cost and greenhouse gas emissions of each recycling process based on the battery system, metal composition, process route, reagent consumption and energy consumption parameters, taking 1 kg of retired lithium-ion battery black powder as the benchmark, and using these as target labels.

[0025] In this embodiment, experimental data on the recycling of retired lithium-ion batteries were collected from publicly available domestic and international literature, including battery characteristics, recycling process parameters, and metal leaching rates. Data was obtained from tables and text or extracted from graphs using Origin software (https: / / www.originlab.com / ). Battery characteristics included battery type, cathode material system, and the mass fractions of lithium, nickel, cobalt, and manganese. Recycling process parameters included the main recycling process route, leaching agent type, leaching agent concentration, reaction temperature, reaction time, pH, solid-liquid ratio, voltage, stirring speed, and reagent dosage. Metal leaching rates included the leaching rate of at least one metal among lithium, nickel, cobalt, and manganese. Using the EverBatt2023 model, with a baseline of processing 1 kg of retired lithium-ion battery black powder, the recycling cost and greenhouse gas emissions corresponding to each recycling process were calculated, and the metal leaching rate, recycling cost, and greenhouse gas emissions were used as labels. Ultimately, over 13,000 usable data records were obtained, and the database contained 23 feature variables and 6 prediction labels. The database is divided into a support set and a query set in an 8:2 ratio; Table 1 shows the main parameter ranges and label types for the decommissioned lithium battery recycling process data. Table 1. Parameter Range and Label Statistics of Experimental Data on Recycled Lithium Batteries Leaching rates are statistically analyzed based on effective values ​​from 0 to 100%; recovery costs and greenhouse gas emissions are statistically analyzed based on non-empty effective values, with units of ($ / kg black mass) and (kg CO2-eq / kg black mass), respectively; other continuous variables are statistical results of the original table, used only to illustrate the coverage of the dataset.

[0026] Furthermore, S2 also includes: The initial dataset is subjected to field unification, feature cleaning, missing value handling, continuous variable numericalization, and categorical variable encoding to obtain the final dataset, which is then divided into a support set and a query set. In step S2, the unification process includes field unification, unit conversion, feature cleaning, missing value processing, continuous variable numericalization, categorical variable encoding, and removal of target missing samples.

[0027] Furthermore, in step S2, the final dataset is divided into a support set and a query set in an 8:2 ratio, wherein the support set is used to provide contextual priors for the large table model, and the query set is used to evaluate the generalization prediction performance of the model.

[0028] Furthermore, S3 also includes: Using battery characteristics and process parameters as input features, and metal leaching rate, recycling cost and greenhouse gas emissions as prediction labels, a large tabular model prediction task is constructed based on the support set and query set. The prediction performance of the model is evaluated by comparing it with a traditional machine learning model. In step S3, the traditional machine learning model includes one or more of Random Forest, LightGBM, AdaBoost, and XGBoost; Metrics for evaluating the predictive performance of a model include mean absolute error (MAE) and coefficient of determination (R²). 2 One or more of the following: mean absolute percentage error (MAPE).

[0029] Furthermore, S4 also includes: Based on the tabular large model, the predicted results of metal leaching rate, recycling cost and greenhouse gas emissions of candidate recycling processes are obtained, and the key variables and their interactions are analyzed by SHAP, PDP or 2D-PDP. In step S4, SHAP, PDP or 2D-PDP are used to interpret the large table model to obtain the key variables affecting metal leaching rate, recycling cost and greenhouse gas emissions and their interactions. The SHAP analysis includes global key variable impact analysis and process route key variable impact analysis.

[0030] Furthermore, S5 also includes: The effective metal set is determined based on the cathode material composition. The minimum value of the predicted leaching rate of the effective metal is taken as the minimum guaranteed leaching rate to be maximized. Together with minimizing recycling costs and minimizing greenhouse gas emissions, a three-objective optimization function is constructed. The Pareto candidate process set is obtained through NSGA-II.

[0031] Furthermore, the method also includes: In step S5, the three-objective optimization function includes maximizing the guaranteed leaching rate, minimizing recycling costs, and minimizing greenhouse gas emissions. Lithium is included in the effective metal set by default, while nickel, cobalt, and manganese are included in the effective metal set when their mass fraction is greater than zero. The minimum value among the predicted leaching rates of each effective metal is used as the guaranteed leaching rate to be maximized. Furthermore, the method also includes: In steps S5 and S6, lithium is included in the effective metal set by default, and nickel, cobalt, and manganese are included in the effective metal set when their mass fraction is greater than zero. The minimum predicted leaching rate of each effective metal is used as the minimum guaranteed leaching rate to be maximized. The minimum guaranteed leaching rate, recycling cost, and greenhouse gas emissions in the Pareto candidate process set are normalized. A comprehensive decision score is calculated based on preset decision weights to obtain recommended process schemes that prioritize leaching rate, cost, low carbon, or a three-objective equilibrium. The system outputs battery configuration, recycling process parameters, metal leaching rate, recycling cost, and greenhouse gas emissions through a graphical user interface. In this embodiment, specifically in one particular instance, NMC532 type decommissioned lithium battery cathode material is used as the object to be recycled; refer to Figure 2-7 Based on the compositional characteristics of NMC532 cathode material, which simultaneously contains Li, Ni, Co, and Mn, Li, Ni, Co, and Mn are all included in the effective metal set. Subsequently, multiple sets of candidate recycling process schemes are generated in the candidate wet recycling process parameter space, and each candidate process scheme is input into the large table model to obtain the corresponding predicted values ​​of Li, Ni, Co, and Mn leaching rates, predicted values ​​of recycling costs per unit processing volume, and predicted values ​​of greenhouse gas emissions per unit processing volume.

[0032] Reference Figure 8 In this embodiment, a set of comprehensive optimal candidate process schemes was obtained through NSGA-II multi-objective optimization screening. The main recovery route of this scheme is hydrometallurgy, the leaching method is organic acid-biomass-assisted acid leaching, the stirring speed parameters are 156.2543 r / min, the main leaching agent concentration is 0.9736 mol / L, the auxiliary leaching agent concentration is 0.5856 mol / L, the main reagent dosage parameter is 14.9020, the auxiliary reagent dosage parameter is 7.9515, the pH is 1.1372, the reaction temperature is 90.9741℃, the reaction time is 101.3527 min, and the solid-liquid ratio is 0.1256 kg / L.

[0033] According to the large-scale model predictions in the table, the predicted leaching rates of Li, Ni, Co, and Mn under this candidate process are 86.8%, 84.6%, 87.5%, and 85.9%, respectively. The minimum predicted leaching rate for each metal in the effective metal set is determined, which is 84.59% for this candidate process. Meanwhile, the predicted recovery cost per unit throughput for this candidate process is $9.4537 / kg black mass, and the predicted greenhouse gas emissions per unit throughput are 3.0618 kgCO2-eq / kg black mass, resulting in a comprehensive score of 76.01 / 100.

[0034] The aforementioned candidate process schemes belong to the comprehensive and balanced schemes within the Pareto candidate process set. They can maintain a high synergistic leaching rate of Li, Ni, Co, and Mn multi-metals while also achieving low recycling costs and low greenhouse gas emissions. This demonstrates that the present invention can, for a given battery system, provide a comprehensive solution based on the large-scale model prediction results and the three-dimensional... The optimization function is used to select the recommended combination of recycling process parameters that balances multi-metal leaching efficiency, economy, and low carbon emissions.

[0035] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of 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, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0036] Furthermore, the functional units in the various embodiments of this application 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 units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0037] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A method for screening recycling processes for decommissioned lithium batteries based on a large tabular model, characterized in that, include: S1. Obtain a candidate recycling process set consisting of multiple candidate recycling processes, extract the process data of the candidate recycling process set, and construct an initial dataset; S2, preprocess the initial dataset to obtain the final dataset, and divide it into a support set and a query set; S3, determine the input features and predicted labels, and construct and filter the large table model by combining the support set and query set; S4. Based on the above large table model, obtain the prediction results of the candidate recycling process set; S5. Based on the prediction results of multi-metal leaching rate, recycling cost, and greenhouse gas emissions output by the large table model for the candidate recycling process set, a three-objective optimization function is constructed with the objectives of maximizing the guaranteed leaching rate, minimizing the recycling cost, and minimizing greenhouse gas emissions. The guaranteed leaching rate is the minimum predicted value of the leaching rate of each metal in the effective metal set, which is determined based on the mass fractions of Li, Ni, Co, and Mn in the cathode material to be recycled. Multi-objective optimization is then performed on the candidate recycling process set to obtain the Pareto candidate process set. S6. Determine the recommended process scheme from the Pareto candidate process set, and output the corresponding retired lithium battery recycling process parameters through the graphical user interface.

2. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 1, characterized in that, S1 further includes: Collect experimental data on the recycling of retired lithium-ion batteries under various battery systems and recycling process conditions; Battery characteristics, process parameters, and leaching rates of at least one metal among lithium, nickel, cobalt, and manganese were extracted, and the corresponding recycling costs and greenhouse gas emissions were calculated using the EverBatt2023 model to construct an initial dataset. The battery system includes one or more of LCO, LFP, LMO, LMFP, NCA, NMC111, NMC333, NMC532, NMC622, NMC811 and their hybrid systems; The recycling process conditions include one or more of the following: hydrometallurgical process, pyrometallurgical process, pyrometallurgical-hydrometallurgical combined process, and direct recycling process.

3. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 2, characterized in that, In step S1, the battery characteristics include one or more of the following: battery type, positive electrode material system, lithium mass fraction, nickel mass fraction, cobalt mass fraction, manganese mass fraction, and physical form. The process parameters include one or more of the following: recovery process route, leaching or extraction method, leachate type, leachate concentration, reaction temperature, reaction time, pH, solid-liquid ratio, voltage, stirring speed, and reagent dosage. In step S1, the EverBatt2023 model is used to calculate the recycling cost and greenhouse gas emissions of each recycling process based on the battery system, metal composition, process route, reagent consumption and energy consumption parameters, taking 1 kg of retired lithium-ion battery black powder as the benchmark, and using these as target labels.

4. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 2, characterized in that, The S2 further includes: The initial dataset is subjected to field unification, feature cleaning, missing value handling, continuous variable numericalization, and categorical variable encoding to obtain the final dataset, which is then divided into a support set and a query set. In step S2, the unification process includes field unification, unit conversion, feature cleaning, missing value processing, continuous variable numericalization, categorical variable encoding, and removal of target missing samples.

5. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 4, characterized in that, In step S2, the final dataset is divided into a support set and a query set in an 8:2 ratio. The support set is used to provide contextual priors for the large table model, and the query set is used to evaluate the generalization prediction performance of the model.

6. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 2, characterized in that, The S3 further includes: Using battery characteristics and process parameters as input features, and metal leaching rate, recycling cost and greenhouse gas emissions as prediction labels, a large tabular model prediction task is constructed based on the support set and query set. The prediction performance of the model is evaluated by comparing it with a traditional machine learning model. In step S3, the traditional machine learning model includes one or more of Random Forest, LightGBM, AdaBoost, and XGBoost; Metrics for evaluating the predictive performance of a model include mean absolute error (MAE) and coefficient of determination (R²). 2 One or more of the following: mean absolute percentage error (MAPE).

7. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 2, characterized in that, The S4 further includes: Based on the tabular large model, the predicted results of metal leaching rate, recycling cost and greenhouse gas emissions of candidate recycling processes are obtained, and the key variables and their interactions are analyzed by SHAP, PDP or 2D-PDP. In step S4, SHAP, PDP or 2D-PDP are used to interpret the large table model to obtain the key variables affecting metal leaching rate, recycling cost and greenhouse gas emissions and their interactions. The SHAP analysis includes global key variable impact analysis and process route key variable impact analysis.

8. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 2, characterized in that, The S5 also includes: The effective metal set is determined based on the cathode material composition. The minimum value of the predicted leaching rate of the effective metal is taken as the minimum guaranteed leaching rate to be maximized. Together with minimizing recycling costs and minimizing greenhouse gas emissions, a three-objective optimization function is constructed. The Pareto candidate process set is obtained through NSGA-II.

9. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 8, characterized in that, Also includes: In step S5, the three objective optimization functions include maximizing the guaranteed leaching rate, minimizing recycling costs, and minimizing greenhouse gas emissions. Lithium is included in the effective metal set by default, while nickel, cobalt, and manganese are included in the effective metal set when their mass fraction is greater than zero. The minimum value among the predicted leaching rates of each effective metal is used as the guaranteed leaching rate to be maximized.

10. The method for screening decommissioned lithium battery recycling processes based on a large table model according to claim 2, characterized in that, Also includes: In steps S5 and S6, lithium is included in the effective metal set by default, and nickel, cobalt, and manganese are included in the effective metal set when their mass fraction is greater than zero. The minimum predicted leaching rate of each effective metal is taken as the minimum guaranteed leaching rate to be maximized. The minimum guaranteed leaching rate, recycling cost and greenhouse gas emissions in the Pareto candidate process set are normalized. The comprehensive decision score is calculated according to the preset decision weight to obtain the recommended process scheme of leaching rate priority, cost priority, low carbon priority or three-objective balance. The battery system, recycling process parameters, metal leaching rate, recycling cost and greenhouse gas emissions are output through the graphical operation interface.

Images