Method and apparatus for producing lithium

Abstract

A lithium manufacturing method according to an embodiment includes a step of determining, by an artificial intelligence model, a main variable of a yield of a production tank in an electrodialysis apparatus; a step of setting a range of an input condition for the calculated main variable; and a step of calculating a yield and a production result corresponding to an input condition within the range by inputting the input condition to the artificial intelligence model, and determining a final operation condition according to the yield and the production result when the calculated yield and production result satisfy a certain condition.

Description

Technical Field

[0001] This invention relates to a lithium manufacturing method and apparatus, and more specifically, to a lithium manufacturing method and apparatus for determining operating conditions for optimal yield and production results. Background Technology

[0002] Bipolar electrodialysis (BPED) is an electro-separation process that uses ion exchange materials to separate or concentrate specific ions from a solution. BPED is attracting attention as an important water separation and regeneration technology. However, the efficiency and performance of BPED operation vary significantly depending on various operating conditions. Therefore, finding the optimal operating conditions can greatly improve the process's economy and efficiency. Summary of the Invention

[0003] (a) Technical problems to be solved An embodiment of the present invention provides a lithium manufacturing method and apparatus, in which an enhancement-type artificial intelligence model is applied during the preprocessing process to determine the optimal operating conditions for the calculated main variables.

[0004] An embodiment of the present invention provides a lithium manufacturing method and apparatus, which generates optimal operating conditions that satisfy the optimal yield and production results by preprocessing based on artificial intelligence for the calculated main variables.

[0005] (II) Technical Solution In an embodiment, the lithium manufacturing method includes: a step of calculating and determining the main variable of the output of the generation tank in an electrodialysis device using an artificial intelligence model; a step of setting a range of input conditions for the calculated main variable; and a step of calculating the output and production results corresponding to the input conditions by inputting the input conditions within the range into the artificial intelligence model, and determining final operating conditions based on the output and production results when the calculated output and production results meet specific conditions.

[0006] The steps for determining the final operating conditions may include: a first step of generating randomly selected input conditions within the set range; a second step of inputting the generated input conditions into an improvement-type artificial intelligence model to calculate the output and production results corresponding to the input conditions; a third step of determining the output and the input conditions as candidate operating conditions when the production results meet preset constraints; and a step of determining the candidate operating condition that calculates the output and production results closest to the preset target conditions as the final operating condition by repeating the first to the third steps among the determined multiple candidate operating conditions.

[0007] The steps for calculating the principal variables may include: a cleaning step, in which multiple initial variables used in the electrodialysis device are cleaned to obtain multiple first variables; a first preprocessing step, in which the first variables are input into an enhancement-type artificial intelligence model, the importance of each first variable is calculated, and multiple second variables are obtained through preprocessing based on the importance of the variables; and a second preprocessing step, in which the second variables are input into the enhancement-type artificial intelligence model to obtain multiple principal variables that meet or exceed a first criterion in terms of predictive fit.

[0008] The importance of the variables can be expressed as permutation importance, and the first preprocessing step may include removing first variables whose permutation importance is lower than a preset second standard from the first variables.

[0009] The second preprocessing step may include the step of calculating the prediction fit using the mean absolute percentage error (MAPE).

[0010] The step of setting the range of input conditions may include determining the range of input conditions based on the upper and lower limits of the input conditions throughout the entire operation of the electrodialysis equipment.

[0011] The generating tank may include an alkali tank and an acid tank, and the output may include lithium (Li) output and sulfur (S) output in the alkali tank and sulfur and lithium output in the acid tank.

[0012] The primary variables may include a first primary variable for determining the lithium production in the alkali tank, a second primary variable for determining the sulfur production in the alkali tank, a third primary variable for determining the sulfur production in the acid tank, and a fourth primary variable for determining the lithium production in the acid tank.

[0013] The production results may include multiple production results, which may include sulfur (S) concentration in the base tank, lithium concentration in the acid tank, current efficiency, and lithium conversion rate.

[0014] The second step may include inputting first to fourth input conditions that are randomly selected corresponding to the first to fourth primary variables respectively, calculating the output for each of the first to fourth primary variables, and calculating the production result by combining the calculated outputs for each of the first to fourth primary variables.

[0015] In some embodiments, the lithium manufacturing apparatus includes: a primary variable calculation module, which calculates primary variables for determining the output of a generation tank within an electrodialysis device using an artificial intelligence model; an operating condition generation module, which sets a range of input conditions for the calculated primary variables, inputs input conditions selected within the range into the artificial intelligence model, and calculates the output and production results corresponding to the input conditions as operating conditions; and an operating condition optimization module, which determines the calculated operating conditions as final operating conditions when the calculated operating conditions meet specific conditions.

[0016] When the production result meets the preset constraints, the operation condition generation module can determine the output and the input conditions as candidate operation conditions. The operation condition optimization module can use the operation condition generation module to determine the candidate operation condition that is closest to the preset target condition among the multiple generated candidate operation conditions as the final operation condition.

[0017] The main variable calculation module may include: a cleaning unit, which cleans the data of multiple initial variables used in the lithium-generating electrodialysis equipment to obtain multiple first variables; a first preprocessing unit, which calculates the variable importance of each first variable by inputting the first variables into an enhancement-type artificial intelligence model, and obtains multiple second variables based on the variable importance through preprocessing; and a second preprocessing unit, which inputs the second variables into the enhancement-type artificial intelligence model to obtain multiple main variables that meet or exceed a first criterion in terms of prediction fit with a minimum number of variables.

[0018] The importance of the variables can be expressed as permutation importance. The first preprocessing unit can remove first variables whose permutation importance is lower than a preset second standard from the first variables.

[0019] The second preprocessing unit can calculate the prediction fit using the mean absolute percentage error (MAPE).

[0020] The operating condition generation module can determine the range of the input conditions based on the upper and lower limits of the input conditions throughout the entire operation of the electrodialysis equipment.

[0021] The generating tank may include an alkali tank and an acid tank, and the output may include lithium (Li) output and sulfur (S) output in the alkali tank and sulfur and lithium output in the acid tank.

[0022] The primary variables may include a first primary variable for determining the lithium production in the alkali tank, a second primary variable for determining the sulfur production in the alkali tank, a third primary variable for determining the sulfur production in the acid tank, and a fourth primary variable for determining the lithium production in the acid tank.

[0023] The production results may include multiple production results, which may include sulfur (S) concentration in the base tank, lithium concentration in the acid tank, current efficiency, and lithium conversion rate.

[0024] The operation condition generation module can input the first to fourth input conditions, which are randomly selected and correspond to the first to fourth main variables respectively, calculate the output for each of the first to fourth main variables, and calculate the production result by combining the calculated outputs for each of the first to fourth main variables.

[0025] (III) Beneficial Effects According to an embodiment of the present invention, a lithium manufacturing method and apparatus apply an enhanced artificial intelligence model during the preprocessing stage, thereby enabling the determination of optimal operating conditions for the calculated main variables.

[0026] According to an embodiment of the present invention, a lithium manufacturing method and apparatus can produce optimal operating conditions that satisfy the optimal yield and production results by preprocessing based on artificial intelligence for the calculated main variables. Attached Figure Description

[0027] Figure 1 This is a diagram illustrating the structure of an electrodialysis apparatus according to an embodiment of the present invention.

[0028] Figure 2 This is a diagram showing one end of an electrodialysis apparatus according to an embodiment of the present invention.

[0029] Figure 3 This is a block diagram of a lithium manufacturing apparatus according to an embodiment of the present invention.

[0030] Figure 4 This is a flowchart illustrating the steps for calculating the main variables according to an embodiment of the present invention.

[0031] Figure 5 and Figure 6 This is a flowchart of a lithium manufacturing method according to an embodiment of the present invention.

[0032] Figures 7 to 10 This is a graph illustrating the output calculation results for the main variables according to an embodiment of the present invention.

[0033] Figure 11 This illustrates optimal operating conditions generated according to a lithium manufacturing method, as shown in an embodiment of the present invention.

[0034] Figure 12 This is a diagram illustrating a computing device according to an embodiment of the present invention. Detailed Implementation

[0035] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, so that those skilled in the art can readily implement the invention. However, the present invention can be implemented in various different forms and is not limited to the embodiments described herein. Furthermore, for the purpose of clearly illustrating the invention in the drawings, parts unrelated to the description have been omitted, and similar reference numerals are used throughout the specification for similar parts.

[0036] When a portion of the specification and claims "includes" a constituent element, unless specifically stated otherwise, it means that other constituent elements may also be included, not that one constituent element is excluded. Terms including "first," "second," etc., can be used to describe various constituent elements, but the constituent elements are not limited by the terms. The terms are used only to distinguish one constituent element from another.

[0037] The terms “...part,” “...device,” “module,” etc. used in this specification may refer to a unit capable of performing at least one function or action described in this specification, which may be implemented by hardware or circuits, software, or a combination of hardware or circuits and software.

[0038] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0039] Figure 1 This is a diagram illustrating the structure of an electrodialysis apparatus according to an embodiment of the present invention. Figure 2 This is a diagram showing one end of an electrodialysis apparatus according to an embodiment of the present invention.

[0040] exist Figure 1 and Figure 2 In this context, the electrodialysis equipment can be a bipolar electrodialysis (BPED) device. That is, the electrodialysis equipment can be a bipolar electrodialysis device. The BPED device can be a device that converts an aqueous solution of lithium sulfate into lithium hydroxide and sulfuric acid. Here, lithium sulfate is Li₂SO₄, lithium hydroxide is LiOH, and sulfuric acid is H₂SO₄. The BPED device can also be an aqueous solution treatment device that uses an electrodialysis membrane in an electric field to simultaneously perform water splitting / ion separation.

[0041] Reference Figure 1 and Figure 2 Electrodialysis equipment (BPED) may include cation exchange membranes (CEM), anion exchange membranes (AEM), and bipolar membranes (BPM).

[0042] Cation exchange membranes (CEMs) have internal anion-exchange groups, thus allowing only cations (e.g., Li+) to pass through. Anion exchange membranes (AEMs) have internal cation-exchange groups, allowing only anions (e.g., SO42-) to pass through. 2- The bipolar membrane (BPM) contains a water-splitting catalyst sandwiched in the middle and is composed of overlapping cation and anion membranes. The BPM can split water in an electric field, thereby generating hydrogen ions (H+). + ) and hydroxide ions (OH) - ).

[0043] That is, a BPED (Bio-Plasma Dialysis) device can use an electrodialysis membrane (cation exchange membrane, anion exchange membrane, bipolar membrane) in an electric field to simultaneously perform water splitting (decomposing into H+). + OH - ) / ion separation (Li + SO4 2- Aqueous solution treatment equipment for ion separation.

[0044] exist Figure 1 In a BPED electrodialysis unit, through a three-step process (press) including stages one through three, the LS solution can transfer Li and SO4 ions to the LH and sulfuric acid (H2SO4) solutions. For example, in a BPED unit, deionized water (DI water) is contacted counterflow with the LS solution, thereby converting it into LH and sulfuric acid solutions. Here, the LS solution is lithium sulfate (Li2SO4), and the LH solution is lithium hydroxide (LiOH).

[0045] The first to third stages can respectively include a salt chamber, an acid chamber, a base chamber, a salt tank, an acid tank, and a base tank.

[0046] In each stage, the salt chamber can be supplied with lithium sulfate (Li₂SO₄) to produce desalted water after the reaction. The acid chamber can be supplied with deionized water (DI water) to produce sulfuric acid (H₂SO₄) after the reaction. The alkali chamber can be supplied with deionized water (DI water) to produce lithium hydroxide (LiOH) after the reaction.

[0047] Demineralized water generated in a brine tank. Sulfuric acid generated in an acid tank. Lithium hydroxide generated in an alkali tank.

[0048] The output of a BPED (Biodialysis Electrodialysis) unit is determined by the discharge flow rates of sulfuric acid and lithium hydroxide. The output can be determined by controlling the influent flow rates of water (H₂O), lithium sulfate, rectifier current and voltage, and the management of pH, conductivity, circulation flow rate, and circulation pressure in each chamber. The influent flow rates, rectifier current, voltage, pH, conductivity, circulation flow rate, and circulation pressure, corresponding to the control and management elements, can be detected by sensors installed inside each chamber. The control and management elements can correspond to variables that determine the concentrations of the generated lithium hydroxide and sulfuric acid.

[0049] As production results for determining the optimal operating conditions of the electrodialysis equipment (BPED), production results may include lithium production (kg / hr), current efficiency (%), sulfur (S) concentration ratio in the base solution, lithium (Li) concentration ratio in the acid solution, and lithium (Li) conversion rate (%).

[0050] The lithium concentration in the alkaline solution in the base tank, the sulfur concentration in the alkaline solution in the base tank, the sulfur concentration in the acid solution in the acid tank, and the lithium concentration in the acid solution in the acid tank of the BPED electrodialysis equipment can be components of the calculation of the production results.

[0051] That is, for the BPED electrodialysis equipment, the primary variable for lithium concentration in the alkaline solution (LiOH) in the base tank, the secondary variable for sulfur concentration in the alkaline solution, the tertiary variable for sulfur concentration in the acid solution (H2SO4) in the acid tank, and the tertiary variable for lithium concentration in the acid solution can be the primary variables determining the production result. The first to fourth primary variables can be calculated respectively using an enhanced artificial intelligence model.

[0052] Figure 3 This is a block diagram of a lithium manufacturing apparatus according to an embodiment of the present invention.

[0053] The lithium manufacturing unit 1000 generates optimal operating conditions to achieve a target function (target conditions) that includes lithium yield (kg / hour (hr)), current efficiency (%), sulfur concentration ratio in alkaline solution, lithium concentration ratio in acidic solution, and lithium conversion rate (%).

[0054] The lithium manufacturing apparatus 1000 can generate optimal operating conditions using the aforementioned first to fourth primary variables. Since these primary variables are used as components constituting the production results representing the optimal operating conditions, calculations are performed using a highly fitted artificial intelligence model. The lithium manufacturing apparatus 1000 can generate a prediction model as a sub-model for each of the aforementioned first to fourth primary variables.

[0055] The lithium manufacturing apparatus 1000 includes a main variable calculation module 100, an operating condition generation module 200, and an operating condition optimization module 300.

[0056] The primary variable calculation module 100 can calculate the primary variables using an enhanced artificial intelligence model. More specifically, the primary variable calculation module 100 can calculate the primary variables using an enhanced artificial intelligence model that includes sub-models that respectively calculate the first to fourth primary variables. The primary variable calculation module 100 includes a cleaning unit 110, a first preprocessing unit 120, a second preprocessing unit 130, and a prediction verification unit 140.

[0057] The cleaning unit 110 can obtain multiple first variables by preprocessing multiple initial variables used in the BPED electrodialysis equipment. For example, the number of initial variables can be 525. The cleaning unit 110 can remove missing values, outliers, duplicate data, non-operational data (data from non-operational sensors), and process data outside the BPED process from the initial variables, including all variables of the BPED process, through data cleaning preprocessing. The cleaning unit 110 can also remove synchronously changing data as duplicate data based on correlation analysis between the data. The cleaning unit 110 obtains fewer first variables than the number of initial variables through data cleaning. For example, the number of first variables can be 400.

[0058] The first preprocessing unit 120 can input the first variable into the boosting-type artificial intelligence model to calculate the variable importance of each first variable.

[0059] Boosting AI models are AI models that employ boosting algorithms. Boosting is a machine learning algorithm that combines weak learners to build a strong learner. Boosting methods can be used for classification and regression problems. Boosting algorithms train models sequentially. At each step, the boosting algorithm adds a new model that corrects the error of the previous model. The boosting algorithm combines all the models to generate the final prediction.

[0060] Boosting-type AI models can be selected from a group consisting of AdaBoost, RandomForest, Catboost, Gradient Boosting Model, Lightweight Gradient Boosting Machine (Light GBM), and Extreme Gradient Boosting (XGBoost). Extreme Gradient Boosting (XGBoost) is preferred among boosting-type AI models.

[0061] Adaptive Boosting (AdaBoost) is an initial boosting algorithm that learns by assigning weights to each data point and giving higher weights to misclassified data points.

[0062] Gradient Boosting Machines (GBMs) are a method of training a model in the direction of reducing error. The error at each step is calculated using the gradient of the loss function.

[0063] Extreme Gradient Boosting (XGBoost) is an extended version of GBM, which includes features such as regularization, parallel processing, and missing value handling.

[0064] Lightweight Gradient Boosting Machine (Light GBM) is a boosting algorithm specifically designed for large-scale datasets, characterized by its fast training speed and high memory efficiency. Similar to Extreme Gradient Boosting (XGBoost), it includes regularization and parallel processing features.

[0065] Catboost is a boosting algorithm specifically designed for categorical data, offering automatic categorical feature transformation and faster training speed.

[0066] In improvement-type AI models, for regression problems such as concentration prediction, data is prepared through preprocessing, the model is trained, and performance is evaluated, along with feature importance analysis. Once a satisfactory model is obtained, it is applied to real-world environments to predict concentrations in real time and for generation control and optimization.

[0067] In this process, improved artificial intelligence models can analyze the importance of features or variables to determine which variables have the greatest impact on the generation concentration prediction.

[0068] That is, the first pretreatment unit 120 can improve the analysis of the importance of variables for the first variable by using an artificial intelligence-like model, and understand which variables among the first variables have a greater impact on the yield prediction, including the concentration of lithium and sulfur.

[0069] The first preprocessing unit 120 can use permutation importance to determine the importance of variables. Permutation importance is a metric used to assess model performance (accuracy, F1 score, R-squared, etc.) by determining how much more important a feature is removed from the AI ​​model. 2 A method to estimate the importance of a feature by measuring the degree of decrease in its value (e.g., [missing information]).

[0070] The first preprocessing unit 120 can obtain multiple second variables based on the calculated variable importance through first preprocessing. For example, the first preprocessing unit 120 can remove first variables whose permutation importance is lower than a preset specific standard from the first variables. That is, the first preprocessing unit 120 can assume that, according to the permutation importance of each first variable, the lower the permutation importance of a variable, the smaller its impact on the performance of the artificial intelligence model. By removing variables whose permutation importance is lower than a specific standard, the noise in the data can be reduced, thereby improving the output prediction performance of the artificial intelligence model. The second variables may include the remaining variables after removing variables whose permutation importance is lower than a specific standard from the first variables. For example, the number of second variables may be 100.

[0071] The second preprocessing unit 130 can obtain multiple third variables that meet or exceed the first criterion of prediction fit by inputting the second variable into the improved artificial intelligence model. For example, the number of third variables can be determined differently for each sub-model, such as 4, 10, 33, or 100.

[0072] Enhancement-type artificial intelligence models improve their predictive performance by generating models using preprocessed data and evaluating their performance. In one embodiment, the second preprocessing unit 130 evaluates the model's performance using a second variable and obtains a third variable that has a predictive fit exceeding a specific criterion for the evaluation result. For example, the second preprocessing unit 130 can use evaluation metrics such as mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE).

[0073] Mean Absolute Error (MAE) is the average of the absolute values ​​of all prediction errors. Root Mean Square Error (RMSE) is the square root of the average of the squares of all prediction errors. Mean Absolute Percentage Error (MAPE) is the average of the absolute values ​​of all prediction errors divided by the actual values. In other words, it is the percentage of the difference between the actual and predicted values ​​divided by the actual value. The result can be expressed as a percentage. As a tool for evaluating prediction fit, Mean Absolute Percentage Error (MAPE) is a commonly used value in regression problems / models because it represents a relative error ratio, thus providing a direct interpretation of the prediction results.

[0074] The second preprocessing unit 130 can use the mean absolute percentage error (MAPE) to calculate the predictive fit of the artificial intelligence model using the second variable, and obtain a third variable from a set of multiple variables whose calculated predictive fit meets a preset specific criterion. This third variable can be a primary variable. The primary variables can be determined separately for each sub-model. The primary variables can be the first to fourth primary variables mentioned above.

[0075] The prediction validation unit 140 can calculate the variable importance for each acquired third variable. The prediction validation unit 140 can compare the prediction fit when only the variable with the highest variable importance among the third variables is input into the improved AI model with the prediction fit when all third variables are input. When the prediction fit when all third variables are input is higher than the prediction fit when only the variable with the highest variable importance is input, the prediction validation unit 140 can determine the third variable as the final variable.

[0076] The third variable can be any one of the conductivity, input flow rate, circulation flow rate, and circulation pressure of any of the following: lithium sulfate aqueous solution, sulfuric acid, and lithium hydroxide.

[0077] The operation condition generation module 200 can set the range of input conditions for the main variables.

[0078] The operating condition generation module 200 can determine the range of input conditions based on the upper and lower limits of the input conditions throughout the entire operation of the electrodialysis equipment. The operating condition generation module 200 confirms the lower and upper limits of the input conditions for each variable throughout the entire operation. Table 1 shows the lower and upper limits of the input conditions for a subset of 100 variables.

[0079] Table 1 For example, the first to fifth flow rates of variables 1 to 5 represent the lower and upper limits of the flow rates detected (Tag) by the first to fifth sensors, respectively. The first temperature value, number 25, indicates that the lower limit of the temperature (TT) detected (Tag) by the first temperature sensor during the entire operation period is 25 degrees Celsius and the upper limit is 37 degrees Celsius. The third conductivity value, number 100, indicates that the lower limit of the conductivity (CT) detected by the third current-voltage sensor is 35 and the upper limit is 339. The operation condition generation module 200 can set the range of the lower to upper limits of the input conditions for each confirmed variable as the range of input conditions for each variable. The operation condition generation module 200 can randomly select input conditions within the set range. For example, the operation condition generation module 200 can set the temperature input condition for the first temperature value to 30 degrees Celsius, randomly selected from the range of 25 to 37 degrees Celsius. Simultaneously, the input condition for the conductivity value of the third conductivity value can be set to 250, randomly selected from the range of 35 to 339. The operation condition generation module 200 can generate multiple input conditions by repeatedly and randomly selecting input conditions.

[0080] The operation condition generation module 200 inputs the selected input conditions into the improvement-type artificial intelligence model, thereby calculating the corresponding output and production results.

[0081] For example, the operating condition generation module 200 can input the first input condition, randomly selected as described above, into the calculated first primary variable in a sub-model for the lithium concentration (first primary variable) in the alkaline solution in the base tank, thereby calculating the lithium concentration corresponding to the input condition. The operating condition generation module 200 can also input randomly selected second to third input conditions for the second to third primary variables respectively, thereby calculating the sulfur or lithium concentration as the yield.

[0082] The operating condition generation module 200 can calculate the production results by combining the outputs for the first to fourth main variables. As mentioned above, the production results can be represented by production results including lithium output (kg / hour (hr)), current efficiency (%), sulfur concentration ratio in alkaline solution, lithium concentration ratio in acidic solution, and lithium conversion rate (%).

[0083] When the production result meets the preset constraints, the operation condition generation module 200 can generate the output and the first to fourth input conditions as candidate operation conditions. The operation condition generation module 200 can generate multiple candidate operation conditions. That is, the operation condition generation module 200 can repeatedly generate randomly selected input conditions within the range of input conditions for the main variable, and based on this, generate multiple candidate operation conditions. The operation condition generation module 200 can repeat the process of generating candidate operation conditions a preset number of times m.

[0084] The operation condition optimization module 300 can determine one of the candidate operation conditions that is closest to the preset target condition for calculating the output and production result from multiple candidate operation conditions generated by the operation condition generation module 200, and use it as the final operation condition.

[0085] Figure 4 This is a flowchart illustrating the steps for calculating the principal variables according to an embodiment of the present invention. The steps for calculating the principal variables can be performed by the principal variable calculation module 100 (see...). Figure 3 ) Execution. Main variable calculation module 100 (refer to Figure 3 Multiple primary variables can be calculated using the same steps as calculating primary variables.

[0086] The steps for calculating the main variables include a cleaning step (step S100), a first preprocessing step (step S200), a second preprocessing step (step S300), and a verification step (step S400).

[0087] exist Figure 4 First, the cleaning step (step S100) may include a cleaning step (step S110) to prepare and clean the dataset of BPED operational data. The cleaning step (step S100) obtains the first variable by cleaning the initial variable (step S120). Data cleaning may include removing missing values, outliers, duplicate data, non-operational data (non-operational sensor data), and process data outside the BPED process.

[0088] Subsequently, the first preprocessing step (step S200) can preprocess the first variable using Extreme Gradient Boosting (XGBoost), a boosting-type artificial intelligence model. The first preprocessing step (step S200) can input the first variable into Extreme Gradient Boosting (XGBoost) (step S210). The first preprocessing step (step S200) determines the variable importance of each first variable by permutation importance (step S220). The first preprocessing step (step S200) obtains the second variable by removing sub-factors of the permutation importance (step S230).

[0089] Subsequently, the second preprocessing step (step S300) inputs the second variable into Extreme Gradient Boosting (XGBoost) and obtains performance feedback based on the goodness of fit (step S310). The second preprocessing step (step S300) identifies factors based on goodness of fit such as MAPE and searches for the condition that minimizes the number of variables that meet specific goodness of fit criteria (e.g., 90% or 95%) and variable importance (e.g., ranking importance) (step S320). The second preprocessing step (step S300) derives the main factors that meet the above conditions as the minimum variables (main variables) that have an impact on the process and quantifies them (step S330).

[0090] Subsequently, the validation step (step S400) trains and evaluates the artificial intelligence model using the derived principal variables (step S410). The training and evaluation step (step S400) can validate the influence of the principal variables based on the predictive fit of the evaluation results (step S420). Figures 7 to 10 The charts can be used to verify the impact of the main variables.

[0091] Figure 5 and Figure 6 This is a flowchart of a lithium manufacturing method according to an embodiment of the present invention. The lithium manufacturing method can be carried out using a lithium manufacturing apparatus 1000 (see reference 1000). Figure 3 )implement.

[0092] The lithium manufacturing unit 1000 can target lithium production through... Figure 4 The steps involve calculating the primary variables and setting upper / lower limits for the search range of input conditions for each primary variable (step S610). The upper / lower limits of the search range can be set based on tag data measured throughout the entire operation period.

[0093] The lithium manufacturing apparatus 1000 can randomly generate input conditions within the search range of each primary variable (step S620). For example, the lithium manufacturing apparatus 1000 can target multiple sub-models (e.g., four prediction models (see reference)). Figure 6 The main variables are randomly generated for each input condition.

[0094] The lithium manufacturing apparatus 1000 can input the generated input conditions into the Extreme Gradient Boosting (XGBoost) prediction model for each major variable, calculate the values ​​of lithium production and production results, and confirm whether the production results satisfy the constraints (step S630). The lithium manufacturing apparatus 1000 can determine the input conditions and lithium production that satisfy the constraints as candidate operating conditions.

[0095] After repeatedly executing the steps (steps S610 to S630), the lithium manufacturing apparatus 1000, for each major variable, may replace the global optimal solution if the final optimal solution is better than the global optimal solution (step S640). That is, by repeating the steps (steps S610 to S630), the lithium manufacturing apparatus 1000 determines, from among a plurality of determined candidate operating conditions, a candidate operating condition for calculating the output and production result closest to the preset target condition, and uses this as the final operating condition.

[0096] Figure 6 A detailed embodiment is shown. Figure 5 A diagram illustrating the lithium manufacturing process.

[0097] exist Figure 6 In this context, tag data can include variables, which can include operational variables and management variables. Operational variables include control elements, and management variables can include management elements. Control and management elements can include, respectively, the input flow rate of alkaline solutions and acid solutions, the rectifier current, voltage, pH, conductivity, circulation flow rate, and circulation pressure. The lithium concentration in Li2SO4 in the ICP data can be the lithium (Li) concentration in the stock tank containing the LS input solution (Li2SO4).

[0098] n particles can refer to variables with n distinct input conditions. The n input conditions can be randomly selected from the lower to the upper bound of each variable.

[0099] Objective expressions include target conditions that must be met to maximize lithium (Li) production while also satisfying other production outcomes. Constraint expressions include constraints that the equipment must at least meet.

[0100] The lithium manufacturing process detects optimal operating conditions (optimal particles) that meet both target and constraint criteria.

[0101] In the objective formulation, the lithium manufacturing method can utilize multiple (4) sub-models (predictive models) and input variables to meet the target conditions for the production outcome. In the objective formulation, the lithium manufacturing method can calculate the lithium concentration in the alkaline solution (LiOH) in the base tank, the sulfur concentration in the alkaline solution (LiOH) in the base tank, the sulfur concentration in the acid solution (H2SO4) in the acid tank, and the lithium concentration in the acid solution (H2SO4) in the acid tank using the predictive models.

[0102] The first prediction model calculates the lithium concentration in the alkaline solution within the base tank. The second prediction model calculates the sulfur concentration in the alkaline solution within the base tank. The third prediction model calculates the sulfur concentration in the acid solution within the acid tank. The fourth prediction model calculates the lithium concentration in the acid solution within the acid tank.

[0103] The first to fourth input conditions, which are input to the first to fourth principal variables corresponding to the first to fourth prediction models using the enhancement-type artificial intelligence model, can be randomly generated within upper / lower bounds. For example... Figure 4 As shown, the first to fourth main variables can be obtained through artificial intelligence-based preprocessing.

[0104] The current of the first to sixth rectifiers can be the current value of each of the first to sixth rectifiers. LS solution is lithium sulfate (Li2SO4), and LH solution is lithium hydroxide (LiOH).

[0105] Under the target conditions, the production results can be calculated by combining the main variables and input variables of the first to fourth prediction models.

[0106] Lithium production was calculated using the first prediction model and daily LH production. The sulfur concentration (g / L) in the first-stage alkaline solution was calculated using the first and second prediction models. The lithium concentration (g / L) in the acid solution was calculated using the third and fourth prediction models. Lithium conversion efficiency was calculated based on the first prediction model, the lithium (Li) concentration in the feed tank, and the feed flow rate of the LS feed solution. Current efficiency was calculated based on the first prediction model and the currents of the first through sixth rectifiers.

[0107] The lithium manufacturing method can verify whether the main variables and input variables that meet the target conditions satisfy the constraints. In one example, the constraints are: sulfur concentration in the alkaline solution is below 0.065 g / L; lithium concentration in the acid solution is below 0.1 g / L (maximum below 0.2 g / L); lithium conversion rate is above 85% (minimum above 65%); current efficiency is above 50% (minimum above 40%); and the input conditions for each variable are within the range of the lower to upper limits.

[0108] The lithium manufacturing method can be repeated a preset number of times (maxiter=m). The lithium manufacturing method can generate optimized operating conditions through repeated execution. That is, by repeating the process m times, the lithium production of the i-th particle and the input conditions (variable values) that produce a production result closer to the target conditions than the globally optimal particle can be used as the final optimal operating conditions.

[0109] Figures 7 to 10 This is a graph showing the output calculation results of the main variables of an embodiment of the present invention. Figures 7 to 10 The changes in lithium or sulfur concentrations during operation are shown separately.

[0110] Figure 7 This is a graph showing the lithium concentration prediction results for the first principal variable of the lithium concentration prediction model based on the alkaline solution in the base tank.

[0111] exist Figure 7 In the study, the actual lithium concentration change (Real) and the predicted lithium concentration change (Pred) showed similar trends. The mean absolute percentage error (MAPE) was 2.024, and the prediction goodness of fit was approximately 98%, which is above 90%. Therefore, the first principal variable used to validate the calculations consisted of a minimum number of variables with a goodness of fit above 90%.

[0112] Figure 8 This is a graph showing the sulfur concentration prediction results for the second principal variable based on the sulfur concentration prediction model in the alkaline solution within the alkaline tank.

[0113] exist Figure 8 In the study, the actual lithium concentration change (Real) and the predicted lithium concentration change (Pred) showed similar trends. The mean absolute percentage error (MAPE) was 5.223, and the prediction goodness of fit was approximately 95%, which is above 90%. That is, the calculated second principal variable consists of the minimum number of variables with a goodness of fit above standard.

[0114] Figure 9 This is a graph showing the predicted sulfur concentration results for the third principal variable in a model based on a sulfur concentration prediction model for acid solutions in an acid bath.

[0115] exist Figure 9 In the study, the actual lithium concentration change (Real) and the predicted lithium concentration change (Pred) showed similar trends. The mean absolute percentage error (MAPE) was 4.325, and the prediction goodness of fit was approximately 96%, which is above 90%. That is, the calculated third principal variable consists of the minimum number of variables with a goodness of fit above standard.

[0116] Figure 10 This is a graph showing the lithium concentration prediction results for the fourth principal variable of the lithium concentration prediction model based on the acid concentration prediction model in the acid solution in the acid tank.

[0117] exist Figure 10 Similarly, the actual lithium concentration change (Real) and the predicted lithium concentration change (Pred) show similar trends. The mean absolute percentage error (MAPE) is 4.325, and the prediction goodness of fit is approximately 96%, which is above 90%. That is, the calculated fourth principal variable consists of the minimum number of variables with a goodness of fit above standard.

[0118] Figure 11 This illustrates optimal operating conditions generated according to a lithium manufacturing method, as shown in an embodiment of the present invention. Figure 11 Showing according to Figure 6 The best operating conditions generated by the examples.

[0119] Reference Figure 11The table displaying productivity targets includes the setpoints (constraints) and the resulting values ​​(production results) for each production outcome. Each production outcome represents the optimal production result that satisfies the constraints.

[0120] Operational and management variables represent the input conditions applied to the principal variables. Each principal variable is calculated using a preprocessing and prediction model based on an enhanced artificial intelligence model. For example, principal variables might include the input of raw material, the input flow rate, the rectifier voltage, and the circulation flow rate. Input conditions can be randomly selected for each principal variable and optimized through repeated calculations. For instance, for the principal variable of raw material input, the optimal input condition could be selected within a range of conductivity 1, from a maximum of 339 to a minimum of 0.

[0121] Figure 12 This is a diagram illustrating a computing device according to an embodiment of the present invention.

[0122] Reference Figure 12 The lithium manufacturing method and apparatus according to the embodiments can be implemented using a computing device 900.

[0123] The computing device 900 may include at least one of a processor 910, a memory 930, a user interface input device 940, a user interface output device 950, and a storage device 560 that communicate via a bus 920. The computing device 900 may also include a network interface 970 electrically connected to a network 90. ​​The network interface 970 can send or receive signals with other entities via the network 90.

[0124] The processor 910 can be implemented as a microcontroller unit (MCU), application processor (AP), central processing unit (CPU), graphics processing unit (GPU), neural processing unit (NPU), or any other type. It can be any semiconductor device that executes instructions stored in memory 930 or storage device 960. The processor 910 can be configured to implement... Figures 1 to 11 The aforementioned functions and methods.

[0125] The memory 930 and storage device 960 may include various forms of volatile or non-volatile storage media. For example, the memory may include read-only memory (ROM) 931 and random access memory (RAM) 932. In this embodiment, the memory 930 may be located inside or outside the processor 910, and the memory 930 may be connected to the processor 910 in various known ways.

[0126] In some embodiments, at least a portion of the structure or function of the lithium manufacturing method and apparatus according to the embodiments may be implemented in the form of a program or software executed in the computing device 900, and the program or software may be stored in a computer-readable medium.

[0127] In some embodiments, at least a portion of the structure or function of the lithium manufacturing method and apparatus according to the embodiments may be implemented by the hardware or circuitry of the computing device 900, or by a separate hardware or circuitry electrically connected to the computing device 900.

[0128] Although the embodiments of the present invention have been described in detail above, the claims of the present invention are not limited thereto. Various modifications and improvements made by those skilled in the art based on the basic concepts of the present invention as defined in the claims are all within the scope of the claims of the present invention.

[0129] [Explanation of reference numerals in the attached figures] 1000: Lithium manufacturing facility 100: Main Variable Calculation Module 200: Operation Condition Generation Module 300: Operation Condition Optimization Module 110: Cleaning Department 120: First Pre-processing Department 130: Second Pre-processing Department 140: Prediction Validation Department

Claims

1. A method for manufacturing lithium, characterized in that, include: The steps to determine the main variables of the output of the generation tank in an electrodialysis device by using an artificial intelligence model; The step of setting the range of input conditions for the calculated principal variable; as well as The steps involve inputting input conditions within the specified range into an artificial intelligence model to calculate the output and production results corresponding to the input conditions, and determining the final operating conditions based on the calculated output and production results when the calculated output and production results meet specific conditions.

2. The lithium manufacturing method according to claim 1, characterized in that, The steps for determining the final operating conditions include: The first step is to generate randomly selected input conditions within the defined range; The second step is to input the generated input conditions into the improvement-type artificial intelligence model to calculate the output and production results corresponding to the input conditions. The third step involves determining the output and the input conditions as candidate operating conditions when the production result meets preset constraints; and By repeating the first to the third steps, the step of determining the final operating condition from among the determined multiple candidate operating conditions is to calculate the output and production result that is closest to the preset target condition.

3. The lithium manufacturing method according to claim 2, characterized in that, The steps for calculating the main variables include: The cleaning step involves cleaning the data of multiple initial variables used in the electrodialysis equipment to obtain multiple first variables. The first preprocessing step involves inputting the first variable into an enhancement-type artificial intelligence model, calculating the variable importance of each first variable, and obtaining multiple second variables based on the variable importance through preprocessing; and The second preprocessing step involves inputting the second variable into the improved artificial intelligence model to obtain multiple main variables that meet or exceed the first criterion for predictive fit with a minimum number of variables.

4. The lithium manufacturing method according to claim 3, characterized in that, The importance of the variables is determined by ranking their importance. The first preprocessing step includes removing the first variable whose ranking importance is lower than a preset second standard from the first variable.

5. The lithium manufacturing method according to claim 3, characterized in that, The second preprocessing step includes calculating the prediction fit using the mean absolute percentage error (MAPE).

6. The lithium manufacturing method according to claim 2, characterized in that, The step of setting the range of input conditions includes determining the range of input conditions based on the upper and lower limits of the input conditions throughout the entire operation of the electrodialysis equipment.

7. The lithium manufacturing method according to claim 2, characterized in that, The generating tank includes an alkali tank and an acid tank. The output includes lithium (Li) output and sulfur (S) output in the alkali tank, and sulfur output and lithium output in the acid tank.

8. The lithium manufacturing method according to claim 7, characterized in that, The main variables include a first main variable for determining the lithium production in the alkali tank, a second main variable for determining the sulfur production in the alkali tank, a third main variable for determining the sulfur production in the acid tank, and a fourth main variable for determining the lithium production in the acid tank.

9. The lithium manufacturing method according to claim 8, characterized in that, The production results include multiple production results. The various production results include sulfur (S) concentration in the alkali tank, lithium concentration in the acid tank, current efficiency, and lithium conversion rate.

10. The lithium manufacturing method according to claim 9, characterized in that, The second step includes: Input the first to fourth input conditions, which are randomly selected and correspond to the first to fourth primary variables respectively, and calculate the output for each of the first to fourth primary variables. The step of calculating the production result by combining the calculated output for each of the first to fourth principal variables.

11. A lithium manufacturing apparatus, characterized in that, include: The main variable calculation module uses an artificial intelligence model to calculate the main variables used to determine the output of the generation tank in the electrodialysis equipment; The operation condition generation module sets the range of input conditions for the calculated main variable, inputs the input conditions selected within the range into the artificial intelligence model, and calculates the output and production results corresponding to the input conditions as operation conditions. as well as The operation condition optimization module determines the calculated operation condition as the final operation condition when it meets a specific condition.

12. The lithium manufacturing apparatus according to claim 11, characterized in that, When the production result meets the preset constraints, the operation condition generation module determines the output and the input conditions as candidate operation conditions. The operation condition optimization module, through the operation condition generation module, determines the final operation condition as the candidate operation condition that is closest to the preset target condition in terms of output and production result among the multiple generated candidate operation conditions.

13. The lithium manufacturing apparatus according to claim 12, characterized in that, The main variable calculation module includes: The cleaning department obtains multiple first variables by cleaning the data of multiple initial variables used in the electrodialysis equipment for lithium generation; The first preprocessing unit calculates the importance of each first variable by inputting the first variable into an enhancement-type artificial intelligence model, and obtains multiple second variables based on the variable importance through preprocessing; and The second preprocessing unit obtains multiple main variables that meet or exceed the first criterion in a minimum number by inputting the second variable into the improved artificial intelligence model.

14. The lithium manufacturing apparatus according to claim 13, characterized in that, The importance of the variables is determined by ranking their importance. The first preprocessing unit removes the first variable whose arrangement importance is lower than a preset second standard from the first variable.

15. The lithium manufacturing apparatus according to claim 13, characterized in that, The second preprocessing unit calculates the prediction fit using the mean absolute percentage error (MAPE).

16. The lithium manufacturing apparatus according to claim 12, characterized in that, The operating condition generation module determines the range of the input conditions based on the upper and lower limits of the input conditions throughout the entire operation of the electrodialysis equipment.

17. The lithium manufacturing apparatus according to claim 12, characterized in that, The generating tank includes an alkali tank and an acid tank. The output includes lithium (Li) output and sulfur (S) output in the alkali tank, and sulfur output and lithium output in the acid tank.

18. The lithium manufacturing apparatus according to claim 17, characterized in that, The main variables include a first main variable for determining the lithium production in the alkali tank, a second main variable for determining the sulfur production in the alkali tank, a third main variable for determining the sulfur production in the acid tank, and a fourth main variable for determining the lithium production in the acid tank.

19. The lithium manufacturing apparatus according to claim 18, characterized in that, The production results include multiple production results. The various production results include sulfur (S) concentration in the alkali tank, lithium concentration in the acid tank, current efficiency, and lithium conversion rate.

20. The lithium manufacturing apparatus according to claim 19, characterized in that, The operation condition generation module receives input conditions, each corresponding to a randomly selected first to fourth primary variable, and calculates the output for each of the first to fourth primary variables. The production results are calculated by combining the outputs for each of the first to fourth principal variables.

Images