Methods, apparatus, equipment, media, and program products for generating multi-module collaborative schemes for rare earth solid materials.

By integrating a multi-module collaborative scheme generation method, the problems of reliance on manual intervention and inconsistent design frameworks in rare earth material research and development have been solved, achieving efficient and automated experimental scheme generation and improving the efficiency of rare earth material research and development.

CN123120137APending Publication Date: 2026-11-17CHINA RARE (SHENZHEN) RESEARCH INSTITUTE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611275857.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-21
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the current technology, the research and development of rare earth materials relies on human intervention and experience judgment, resulting in long research and development cycles, large parameter space, high trial and error costs, scattered knowledge accumulation and difficulty in sedimentation and reuse. Moreover, the research and development of functional materials and metallic materials lacks a unified design framework, making it difficult to directly output experimental schemes.

Method used

A multi-module collaborative scheme generation method is adopted, which integrates multi-source data management, simulation calculation and performance prediction platform. By acquiring R&D target information, constructing database, calling simulation module to perform simulation calculation, and combining performance prediction model to screen candidate schemes, an executable experimental scheme is generated.

Benefits of technology

It has enabled the efficient generation of R&D goals into executable candidate solutions, improving the efficiency and automation of rare earth material R&D, and ensuring the accuracy and reliability of the solutions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123120137A_ABST
    Figure CN123120137A_ABST
Patent Text Reader

Abstract

This application provides a method, apparatus, equipment, medium, and program product for generating multi-module collaborative schemes for rare earth solid-phase materials, relating to the field of scheme design technology. The method includes: acquiring R&D target information, including target material system, target performance requirements, and process boundary conditions; constructing a database of candidate schemes based on multi-source data; determining the corresponding target process route based on the R&D target information; calling the corresponding simulation module to perform simulation calculations on the target material system and obtaining simulation results; constructing a set of candidate schemes; using a performance prediction model to predict the performance of the candidate schemes; and screening the schemes based on the performance prediction results, target performance requirements, and process boundary conditions to obtain the target experimental scheme. This application embodiment can achieve efficient generation from R&D targets to executable candidate schemes, thereby improving the efficiency of rare earth material R&D and increasing the degree of automation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of scheme design technology, and more specifically, to a method, apparatus, equipment, medium, and program product for generating multi-module collaborative schemes for rare earth solid phase materials. Background Technology

[0002] Rare earth materials have broad application prospects in new energy and other fields. Traditional rare earth material research and development mainly relies on manual literature search, expert experience to set composition and process, and experimental verification. This approach generally suffers from problems such as long research and development cycles, large parameter space, high trial and error costs, and scattered knowledge accumulation that is difficult to consolidate and reuse.

[0003] Existing technologies include materials database management platforms, simulation calculation integration platforms, and machine learning-based materials performance prediction systems. However, these platforms typically implement their respective functions in isolation, or are only geared towards a single approach, a single type of material, or a single software tool. Therefore, the process from inputting R&D goals to outputting executable experimental schemes still relies heavily on manual intervention and experience-based judgment, resulting in low efficiency and automation in rare earth materials R&D. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, equipment, medium, and program product for generating multi-module collaborative solutions for rare earth solid materials, so as to improve the efficiency of rare earth material research and development and increase the degree of automation.

[0005] In a first aspect, embodiments of this application provide a method for generating multi-module collaborative schemes for rare earth solid materials, including: Obtain research and development target information; wherein, the research and development target information includes target material system, target performance requirements, and process boundary conditions; Based on pre-collected multi-source data, a ready-to-use database is constructed for the R&D target information; The corresponding target process routes are determined based on the aforementioned R&D target information; wherein, the target process routes include rare earth solid-phase functional material routes and rare earth metal material routes; The simulation module corresponding to the target process route is invoked to perform simulation calculations on the target material system, and the corresponding simulation results are obtained. Based on the target material system, the process boundary conditions, and the historical sample distribution in the database to be used, a set of candidate solutions is constructed; wherein each candidate solution includes composition parameters and process parameters; The performance prediction model is used in conjunction with the simulation results to predict the performance of each candidate scheme, and the performance prediction results of each candidate scheme are obtained. Based on the performance prediction results, the target performance requirements, and the process boundary conditions, combined with the experimental line equipment capability boundaries and allowable parameter ranges, each candidate scheme is screened to obtain the target experimental scheme.

[0006] In this embodiment of the application, by integrating platform functions such as multi-source data management, simulation calculation and performance prediction into a collaborative design engine, it is possible to efficiently generate from R&D goals to executable candidate solutions, thereby improving the efficiency of rare earth material R&D and increasing the degree of automation.

[0007] In some embodiments, determining the corresponding target process route based on the R&D target information further includes: Based on the aforementioned R&D objective information, multi-dimensional route information for the current R&D task is determined; wherein, the multi-dimensional route information includes target performance type, sample morphology, and expected process path; Based on the degree of matching between the multi-dimensional route information and the preset route determination criteria, the route confidence level of the target process route is determined.

[0008] In this embodiment of the application, the route confidence level is determined by the degree of matching between the target performance type, sample morphology and expected process path and the preset judgment criteria. This can quantify the reliability of route identification and provide a reliable basis for the use of subsequent simulation results.

[0009] In some embodiments, the step of using a performance prediction model in conjunction with the simulation results to predict the performance of each candidate solution includes: The feature fusion weights of the simulation results relative to each of the candidate schemes are adjusted based on the route confidence; wherein the feature fusion weights are positively correlated with the route confidence.

[0010] In this embodiment, the accuracy and reliability of model prediction are further improved by adjusting the feature fusion weights of simulation results in performance prediction based on the positive correlation of route confidence.

[0011] In some embodiments, the step of calling the simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtain corresponding simulation results includes: The target material system is simulated by calling the cross-scale simulation module corresponding to the target process route, and the corresponding simulation results are obtained. Wherein, when the target process route is a rare earth solid-phase functional material route, the cross-scale simulation module includes at least one of a phase equilibrium model, a diffusion model, a sintering densification model, and a grain growth model; where the target process route is a rare earth metal material route, the cross-scale simulation module includes at least one of a thermodynamic model, a solidification model, a precipitation model, and a microstructure evolution model.

[0012] In this embodiment of the application, corresponding cross-scale simulation model sets are configured for functional materials and metallic materials respectively, thereby covering the simulation requirements for the physical properties of the two types of rare earth materials.

[0013] In some embodiments, the step of invoking the cross-scale simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtain the corresponding simulation results includes: When the target process route is a rare earth solid-phase functional material route, the formation energy and diffusion coefficient obtained based on low-scale simulation calculations are used as input parameters for the phase equilibrium model and the diffusion model. The stable phase type, phase fraction, and phase transformation temperature output by the phase equilibrium model are used as the temperature boundary conditions and initial phase composition conditions for the grain growth model and the sintering densification model. Based on the concentration distribution output by the diffusion model, the average grain size output by the grain growth model, and the shrinkage rate and relative density output by the sintering densification model, the corresponding simulation results are obtained.

[0014] In the embodiments of this application, the effectiveness of simulation results is further improved by establishing data transfer from low-scale formation energy and diffusion coefficient to phase equilibrium, grain growth and sintering densification under the functional materials route.

[0015] In some embodiments, the step of invoking the cross-scale simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtain the corresponding simulation results includes: When the target process route is a rare earth metal material route, the formation energy and diffusion barrier obtained from low-scale simulation calculations are used as input parameters for the thermodynamic model, the solidification model, and the precipitation model. The phase stability range and phase composition output by the thermodynamic model are used as the temperature window and initial phase field conditions for the solidification model and the precipitation model. The solidification path and dendrite parameters output by the solidification model are used as the initial microstructure input for the precipitation model. Based on the precipitate type and size distribution output by the precipitation model, the phase composition output by the thermodynamic model, and the dendrite parameters output by the solidification model, the corresponding simulation results are obtained.

[0016] In this embodiment of the application, the effectiveness of the simulation results is further improved by establishing data transfer from low-scale formation energy and diffusion barrier to thermodynamics, solidification, and precipitation models under the metallic material route.

[0017] In some embodiments, the step of using a performance prediction model in conjunction with the simulation results to perform performance predictions on each of the candidate solutions, and obtaining performance prediction results for each of the candidate solutions, includes: The performance prediction model of the multi-task prediction architecture is used to predict the performance of each of the candidate solutions. The input parameters of the performance prediction model include component features and process features extracted based on the candidate schemes, simulation features extracted based on the simulation results, and structural features, characterization image features, and historical performance features extracted from the database to be used. The multi-task prediction architecture includes a unified feature input layer, a task prediction layer, and a result aggregation layer. The task prediction layer is used to establish corresponding prediction tasks for different prediction objectives. The prediction tasks include phase composition classification tasks, performance regression tasks, tissue parameter regression tasks, and experimental risk discrimination tasks.

[0018] In this embodiment, a performance prediction model employing a multi-task prediction architecture performs multi-task prediction and aggregation based on multi-dimensional features such as composition, process, simulation, structure, image, and historical performance, thereby improving the accuracy of performance prediction.

[0019] In some embodiments, the step of screening each candidate scheme based on the performance prediction results, the target performance requirements, and the process boundary conditions, combined with the experimental line equipment capability boundaries and allowable parameter ranges, to obtain the target experimental scheme includes: Based on a preset objective function and constraints, at least one preliminary selection scheme is obtained by screening each candidate scheme; wherein, the objective function includes a performance prediction value weighting term and a model uncertainty penalty term; the constraints include phase stability constraints, impurity phase content constraints, and safety constraints; The preliminary selection schemes are screened based on the preset experimental line equipment capability boundaries and allowable parameter ranges to obtain the target experimental scheme.

[0020] In this embodiment, by first screening preliminary schemes based on constraints such as objective function and phase stability, and then performing secondary screening based on the capability boundary of experimental line equipment, it is possible to comprehensively screen out executable candidate schemes that take into account both performance and compliance with constraints, thereby further improving the accuracy and reliability of the output scheme.

[0021] Secondly, embodiments of this application provide a multi-module collaborative scheme generation device for rare earth solid materials, comprising: The information acquisition module is used to acquire R&D target information; wherein, the R&D target information includes target material system, target performance requirements, and process boundary conditions; The data construction module is used to build a ready-to-use database based on pre-collected multi-source data, which is oriented towards the R&D target information; The route determination module is used to determine the corresponding target process route based on the R&D target information; wherein, the target process route includes rare earth solid-phase functional material route and rare earth metal material route; The simulation calculation module is used to call the simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtain the corresponding simulation results. A candidate construction module is used to construct a set of candidate schemes based on the target material system, the process boundary conditions, and the historical sample distribution in the database to be used; wherein each candidate scheme includes composition parameters and process parameters; The performance prediction module is used to perform performance prediction on each of the candidate solutions using the performance prediction model and the simulation results, and to obtain the performance prediction results of each of the candidate solutions. The scheme screening module is used to screen each of the candidate schemes based on the performance prediction results, the target performance requirements, and the process boundary conditions, combined with the experimental line equipment capability boundaries and allowable parameter ranges, to obtain the target experimental scheme.

[0022] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the method described in any embodiment of the first aspect.

[0023] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the method described in any embodiment of the first aspect.

[0024] Fifthly, embodiments of this application provide a computer program product, the computer program product including a computer program, wherein when the computer program is executed by a processor, it can implement the method described in any embodiment of the first aspect. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A flowchart illustrating a method for generating multi-module collaborative schemes for rare earth solid materials, provided in an embodiment of this application; Figure 2 This is a schematic diagram of a multi-module collaborative solution generation device for rare earth solid materials provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0028] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0029] It should be noted that current rare earth material research and development still largely relies on the traditional model of manual literature retrieval, expert experience in setting composition and processes, and experimental verification. This model generally suffers from problems such as long development cycles, large parameter spaces, high trial-and-error costs, and fragmented knowledge accumulation that is difficult to consolidate and reuse. Especially in the field of rare earth solid-phase materials, functional materials and metallic materials are usually developed along different process routes: functional materials often adopt solid-phase synthesis and densification routes such as mixing, pressing, and sintering, while metallic materials often adopt routes such as batching, smelting, heat treatment, sample preparation, and characterization. The two types of materials differ significantly in terms of data structure, mechanism models, process windows, and evaluation indicators, making it difficult for existing general-purpose materials informatics platforms to directly support rare earth material research and development decisions and process optimization.

[0030] Existing technologies include materials database management platforms, materials genome intelligent design platforms, simulation calculation integration platforms, and machine learning-based materials performance prediction systems. However, most solutions implement these functional modules, such as data management, simulation calculation, and machine learning prediction, in a fragmented manner, or are only geared towards a single route, a single type of material, or a single software tool, making it difficult to form a collaborative design engine that can directly output experimental schemes.

[0031] Specifically, existing material development solutions have the following main problems: 1. Data from multiple sources is scattered and lacks uniformity in standards: The sources of literature, patents, experimental records, instrument data, simulation results, etc. are heterogeneous, and the field naming, units of measurement, sample identification, and process step recording methods are inconsistent, making it difficult to reuse directly.

[0032] 2. A single model is difficult to balance mechanistic constraints and engineering efficiency: relying solely on empirical formulas or data-driven models often makes it difficult to balance interpretability, generalization ability, and reliability under small sample conditions.

[0033] 3. There is a lack of a unified design framework between functional materials and metallic materials routes: the former focuses on proportioning, mixing uniformity, compact density, sintering temperature regime, and phase composition evolution; the latter focuses on composition design, melting heat history, heat treatment regime, and coupling of precipitated phases with microstructure and properties. Existing design systems are difficult to simultaneously accommodate both types of routes.

[0034] 4. Design results are difficult to directly translate into experimentally feasible solutions: Existing platforms mostly output performance prediction values ​​or candidate material lists, without fully considering the equipment capabilities, adjustable process windows, and characterization feedback requirements of subsequent experimental lines.

[0035] To address at least one of the problems existing in the prior art, this application provides a method for generating multi-module collaborative schemes for rare earth solid materials, aiming to improve the efficiency of rare earth material research and development.

[0036] like Figure 1 As shown in the embodiments of this application, a method for generating multi-module collaborative schemes for rare earth solid materials is provided, which may include the following steps: S1. Obtain R&D target information; where R&D target information includes target material system, target performance requirements, and process boundary conditions.

[0037] Specifically, researchers can input research and development target information through the system's human-computer interaction interface. This information mainly includes three aspects: the target material system, the target performance requirements, and the process boundary conditions. Taking the research and development of rare earth solid electrolytes as an example, the target material system is a tantalum-doped garnet-type solid electrolyte Li... 6.4 La3Zr 1.4 Ta 0.6 O 12 The target performance requirements include cubic phase stability, a relative density of not less than 90%, and a room temperature ionic conductivity of not less than 1×10⁻⁶. -4 S / cm; The process boundary conditions can be set according to literature and historical experimental experience as follows: Ta doping amount 0.4 to 0.8 (in terms of the number of atoms per molecular formula), Li2CO3 excess ratio 5% to 15%, sintering temperature 1150℃ to 1220℃, and holding time 8 hours to 12 hours, etc.

[0038] S2. Based on pre-collected multi-source data, construct a standby database for R&D target information.

[0039] Specifically, the system can retrieve and access pre-collected multi-source data based on the target material system. This multi-source data can include data from sources such as papers, patents, standards, experimental records, online instrument logs, first-principles calculations, thermodynamic database calculations, phase-field simulations, and finite element simulations, and a unified field and tagging system is established. Based on this, heterogeneous multi-source data related to the target material system, such as historical literature data, patent data, experimental record data, and equipment log data, can be extracted from the multi-source data and subjected to unified standardization processing, including noise reduction, unit conversion standardization, missing value handling, and field mapping alignment.

[0040] Missing values ​​can be handled in a tiered manner based on field importance and data type: For essential fields such as sample number, composition, key process parameters, and target performance, if they are missing, they are marked as to be supplemented or enter the rejection process; for continuous numerical fields, they are filled using the median or KNN (K-Nearest Neighbors) imputation method based on the same material system and process stage; for categorical fields, they are retained with the "not recorded" label; fields with a missing rate exceeding a preset threshold (e.g., 40%) are not included in subsequent model training.

[0041] After standardization, the system can automatically construct a structured database according to a multivariate correlation model of "material system-composition-structure-process-performance-characterization-source," and generate a unique sample primary key for each data entry. This sample primary key consists of the material system number, formulation number, experimental batch number, sample serial number, and data source number. (Using the aforementioned Li...) 6.4 La3Zr 1.4 Ta 0.6 O 12 Taking the material system as an example, the system retrieves all historical sample records under the system from the database, extracts process parameters such as the excess ratio of Li2CO3, ball milling time, ball milling speed, pressing pressure, sintering temperature, holding time, and sintering atmosphere, as well as the corresponding XRD (X-ray Diffraction) phase composition data, SEM (Scanning Electron Microscope) microstructure images, and measured ionic conductivity data, forming a structured database for use in this research and development goal.

[0042] S3. Determine the corresponding target process routes based on the R&D target information; among which, the target process routes include rare earth solid-phase functional material routes and rare earth metal material routes.

[0043] It should be noted that, based on the target performance requirements, sample morphology, and expected process path input in step S1, the system can automatically identify and determine the target process route. For example, the process route determination logic is as follows: if the target performance is ionic conductivity, magnetic properties, luminescent properties, or electrical functional properties, and the sample morphology is powder, compact, or sintered body, then the system determines the current R&D task as a rare earth solid-phase functional material route; if the target performance is hardness, strength, plasticity, heat resistance, or microstructure stability, and the sample morphology is alloy ingot, bulk, or metallographic sample, then it is determined as a rare earth metal material route. Using the aforementioned Li... 6.4 La3Zr 1.4 Ta 0.6 O 12 Taking solid electrolytes as an example, since the target performance is ionic conductivity and the expected sample morphology is sintered, the system automatically identifies it as a rare earth solid functional material route.

[0044] S4. Call the simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtain the corresponding simulation results.

[0045] Specifically, based on the target process route type determined in step S3, the system can automatically call the corresponding simulation module to perform simulation calculations on the target material system within the candidate composition range and process boundary conditions. The purpose of the simulation calculation is to obtain physicochemical response information of the target material system under different composition and process conditions, including but not limited to quantitative results on phase stability, thermodynamic behavior, microstructure evolution trends, and process feasibility. After the simulation calculation is completed, the system organizes the results output by each simulation module according to a unified data format and unit system, and stores them in association using the sample primary key as an index, for subsequent steps to use as auxiliary input features for performance prediction.

[0046] For example, the tantalum-doped garnet-type solid electrolyte Li 6.4 La3Zr 1.4 Ta 0.6 O 12 For example, after identifying it as a rare earth solid-phase functional material route, the system calls the corresponding cross-scale simulation module to perform simulation calculations on multiple composition-temperature combinations within the range of Ta doping concentration (0.4 to 0.8) and sintering temperature (1150℃ to 1220℃). The simulation results for each combination point are obtained, including phase stability, phase transition temperature, and sintering densification behavior. Each simulation result can be associated with the corresponding candidate components and process parameters through a unified primary key and written into a database for subsequent steps.

[0047] S5. Based on the target material system, process boundary conditions, and historical sample distribution in the database to be used, construct a set of candidate solutions; where each candidate solution includes composition parameters and process parameters.

[0048] It should be noted that, based on the target material system and process boundary conditions input in step S1, and combined with the historical sample distribution in the database to be used constructed in step S2, the system automatically generates several sets of candidate component and process parameter combinations within the parameter space defined by the process boundary conditions, forming a candidate scheme set. Each candidate scheme in the candidate scheme set includes at least component parameters and process parameters. Furthermore, when generating candidate schemes, the system can also refer to the feasible interval indicated by the simulation results in step S4, using the simulation results as one of the constraints for generating candidate schemes, thus avoiding wasting computational resources on physically impossible parameter combinations.

[0049] For example, with Li 6.4 La3Zr 1.4 Ta 0.6 O 12 Taking solid-state electrolytes as an example, the system generates several candidate schemes within the physically feasible parameter space based on the process boundary conditions of Ta doping amount (0.4 to 0.8%), Li₂CO₃ excess ratio (5% to 15%), sintering temperature (1150℃ to 1220℃), and holding time (8 to 12 hours). This is combined with the distribution density of existing experimental samples in the database within the aforementioned parameter space. For example, historical samples are concentrated in the range of Ta doping amount (0.5 to 0.6%) and sintering temperature (1170℃ to 1190℃). Each candidate scheme includes clearly defined compositional parameters (Ta doping amount, Li₂CO₃ excess ratio) and process parameters (including ball milling time, pressing pressure, sintering temperature, holding time, and sintering atmosphere).

[0050] S6. Using the performance prediction model and simulation results, the performance of each candidate scheme is predicted to obtain the performance prediction results of each candidate scheme.

[0051] Specifically, the composition parameters and process parameters of each candidate scheme are used as the basis for the model input features. At the same time, the simulation results obtained in step S4 (including quantitative information on phase stability, thermodynamic behavior, microstructure evolution trend and process feasibility) are used as auxiliary evaluation features and are jointly input into the pre-constructed performance prediction model for prediction.

[0052] For example, the performance prediction model is a model built based on machine learning algorithms. Its input features may include one or more of the following: component features, process features, simulation features, structural features, characterization image features, and historical performance features. The output of the model may include the target performance prediction value of each candidate scheme and the uncertainty (or confidence) evaluation of each candidate scheme.

[0053] For example, with Li 6.4 La3Zr 1.4 Ta 0.6O 12 Taking solid-state electrolytes as an example, for each of the generated candidate schemes, the system extracts parameters such as Ta doping amount, Li2CO3 excess ratio, ball milling time, pressing pressure, sintering temperature, and holding time as compositional and process features; it extracts simulation results such as phase stability, phase transition temperature, and densification trend from simulation calculations as simulation features; and it extracts XRD phase composition and ionic conductivity data of historical samples of the same material system from the database to be used as historical performance features. These multiple features are input into the performance prediction model, which outputs predicted values ​​for room temperature ionic conductivity, relative density, cubic phase formation probability, and the uncertainty corresponding to each prediction result for each candidate scheme. For example, the system predicts that a certain candidate scheme (Ta doping amount of 0.6%, Li2CO3 excess of 10 mol%, sintering temperature of 1180℃, and holding time of 10 hours) has a high probability of cubic phase formation, a predicted relative density of approximately 90% to 93%, and a predicted room temperature ionic conductivity of approximately 1.0 × 10⁻⁶. -4 Up to 3.0×10 -4 S / cm, uncertainty is rated as low.

[0054] S7. Based on the performance prediction results, target performance requirements, and process boundary conditions, combined with the experimental line equipment capacity boundary and allowable parameter range, the candidate schemes are screened to obtain the target experimental scheme.

[0055] Specifically, firstly, based on the performance prediction results of each candidate scheme, and combined with the target performance requirements and process boundary conditions input in step S1, the candidate schemes are initially screened. For example, the screening criteria are: the performance prediction value of the candidate scheme must meet or exceed the target performance requirements, and its composition and process parameters must be within the set process boundary conditions.

[0056] For candidate schemes that pass the initial screening, the system further obtains the equipment capability boundaries and allowable parameter ranges for subsequent automated experimental lines, including but not limited to discrete constraints such as maximum operating temperature, maximum pressing pressure, weighing accuracy, atmosphere control capability, and characterization throughput. Schemes that do not meet the equipment's execution capabilities are eliminated, or the parameters are modified for feasibility. Finally, one or more target experimental schemes that meet all constraints are obtained, and a standardized scheme output is generated and sent to the downstream experimental verification system.

[0057] For example, for rare earth solid-phase functional materials, the fields in the standardized protocol sheet include: material system, target performance, raw material type and ratio, dopant element and doping amount, mixing method and time, pressing pressure, sintering temperature, holding time, sintering atmosphere, suggested characterization items, predicted phase composition, predicted performance value, and priority verification order. For rare earth metal materials, the fields in the standardized protocol sheet include: material system, target performance, matrix metal, rare earth element type and amount added, alloying element ratio, melting temperature, holding time, protective atmosphere, cooling method, heat treatment temperature and time, suggested characterization items, predicted microstructure characteristics, predicted performance value, and priority verification order.

[0058] For example, still using Li 6.4 La3Zr 1.4 Ta 0.6 O 12 Taking solid electrolytes as an example, the predicted results of each candidate scheme are compared with the target performance requirements (relative density not less than 90%, room temperature ionic conductivity not less than 1×10⁻⁶). -4 After comparing the S / cm ratio, a batch of candidate schemes that meet the performance requirements were initially screened. The system further obtained the equipment capability boundaries of the subsequent experimental line. For example, the highest operating temperature of the sintering furnace available for this experimental line is 1250℃, which meets the sintering temperature requirement of 1180℃ to 1200℃ of the above schemes; the maximum pressure of the pressing equipment is 300MPa, which meets the pressing requirement of 200MPa; however, the existing atmosphere furnace only supports air and argon atmospheres and does not support oxygen or reducing atmospheres, so the schemes involving special atmosphere requirements in the original schemes were eliminated. After a second screening based on the equipment boundaries of the experimental line, the system finally outputs a set of target experimental schemes. For example, the specific content of a set of target experimental schemes is: the target composition is Li 6.4 La3Zr 1.4 Ta 0.6 O 12 The sintering conditions were: 10 mol% Li₂CO₃ excess, ball milling speed 300 r / min, ball milling time 6 h, pressing pressure 200 MPa, sintering temperature 1180℃, holding time 10 h, and sintering atmosphere air; the predicted probability of cubic phase formation was high, the predicted relative density was approximately 90% to 93%, and the predicted room temperature ionic conductivity was approximately 1.0 × 10⁻⁶. -4 Up to 3.0×10 - 4 S / cm; Recommended characterization methods include XRD phase composition analysis, SEM microstructure observation, and electrochemical impedance spectroscopy. The system can mark this scheme as a priority validation scheme and generate a standardized scheme output to the downstream experimental validation system.

[0059] Furthermore, after subsequent experimental results are refluxed, if XRD shows a high impurity phase content or an ionic conductivity lower than the target value, the system can automatically generate optimization suggestions for the next round based on the deviation. For example, when the relative density is less than 90%, it is recommended to increase the sintering temperature from 1180℃ to 1200℃ or extend the holding time from 10 h to 12 h; when grains grow abnormally or the sample cracks, it is recommended to reduce the sintering temperature to 1150℃ and shorten the holding time. This achieves a closed-loop design from R&D target input, route identification, simulation calculation, machine learning screening to experimental scheme output and result reflux.

[0060] This application embodiment integrates platform functions such as multi-source data management, simulation calculation, and performance prediction into a collaborative design engine, which can achieve efficient generation from R&D goals to executable candidate solutions, thereby improving the efficiency of rare earth material R&D and increasing the degree of automation.

[0061] For example, the following describes the implementation process of the method in this application under the rare earth metal material route, taking the research and development design of Mg-Gd-Y-Zn-Zr rare earth magnesium alloy as an example: First, researchers can input their research goal as obtaining alloy samples with a relatively uniform distribution of precipitates and an average hardness of not less than 85 HV (Vickers Hardness). The system aggregates existing alloy composition, melting temperature, holding time, cooling method, solution treatment, aging treatment, microstructure images, and hardness test data to establish a database linking "material system - composition - melting - heat treatment - microstructure - properties - source". Based on the target properties being hardness and microstructure stability, the sample morphology being alloy blocks and metallographic specimens, and the process path being melting and heat treatment, the system can automatically identify it as a rare earth metal material route.

[0062] The system utilizes a cross-scale simulation module to calculate the phase stability range of candidate components using a thermodynamic model, predict the main precipitated phase types and precipitation behavior under different heat treatment regimes using a precipitation model, and predict the microstructure evolution trend under different aging regimes using a microstructure evolution model. Simulation results and historical hardness data serve as inputs to the machine learning co-design module, searching for candidate solutions within the range of Gd content 7%-9%, Y content 2%-4%, Zn content 0.5%-1.5%, Zr content 0.3%-0.7%, solution temperature 500-530℃, aging temperature 180-220℃, and aging time 16-24h.

[0063] After objective function evaluation and equipment boundary screening, the system outputs the first experimental scheme: the alloy composition is Mg-8Gd-3Y-1Zn-0.5Zr (unit: mass percentage); the melting temperature is 760℃, under argon protection, and the melting holding time is 25min; the feeding sequence is Mg-Gd, Mg-Y, Mg-Zn, and Mg-Zr master alloys added sequentially after the matrix Mg melts; the casting method is to pour into a metal mold under argon protection and cool to form; the heat treatment regime is solution treatment at 520℃ for 8h, followed by aging treatment at 200℃ for 16h; the recommended characterization items include metallographic microstructure observation, precipitate distribution analysis, and Vickers hardness testing; the system predicts that the hardness of this scheme is close to the target value, but there is a risk of insufficient aging strengthening, therefore the aging time is marked as a key optimization parameter.

[0064] Based on this, after the first round of experimental results were refluxed, if the microstructure showed insufficient uniformity of precipitate distribution and an average hardness of 78 HV, failing to reach the target of 85 HV, the system determined that the main source of deviation was insufficient aging precipitation strengthening and generated a second round of adjustment plan: keeping the Mg-8Gd-3Y-1Zn-0.5Zr alloy composition, melting temperature of 760℃, melting holding time of 25 min, and solution treatment regime of 520℃ for 8 h unchanged, while extending the aging time from 200℃ for 16 h to 200℃ for 24 h. The second round of experimental results showed improved uniformity of precipitate distribution and an average hardness increase to 87 HV, achieving the set target.

[0065] The above embodiments demonstrate that the proposed solution can adapt to the R&D needs of rare earth metal materials. Based on the data characteristics, simulation requirements, and experimental line capabilities of the material system, it can automatically complete route identification, model invocation, candidate scheme screening, and standardized experimental scheme output, providing executable input for subsequent autonomous iterative experimental systems.

[0066] In some embodiments, the method may further include the step of: S8, receiving data returned by the subsequent experimental system and incrementally updating the database and model.

[0067] Specifically, after the target experimental scheme is handed over to the subsequent automated experimental verification system for execution, the system can continuously receive the measured data and abnormal information returned by the experimental line, including but not limited to the actual process parameters recorded during the actual sintering or melting heat treatment process, XRD phase composition analysis results, microstructure images, measured values ​​of properties such as relative density or hardness, as well as equipment operation logs and sample abnormality markers.

[0068] Then, the system can determine whether to start the incremental update process based on preset trigger conditions. The trigger conditions can be set as follows: the number of newly added valid samples reaches a preset threshold (e.g., 3 to 5 new valid samples), the deviation between the measured value and the predicted value exceeds a preset threshold (e.g., the relative density deviation exceeds 3% or the ionic conductivity deviation exceeds half an order of magnitude), data distribution drift occurs, multiple consecutive rounds of schemes fail to achieve the target, or equipment abnormality markers appear, etc.

[0069] When any of the above triggering conditions are met, the system can automatically initiate the incremental update process for the database and model. For example, the incremental update process may include: first, standardizing the experimental data returned from the experimental line, including denoising, unit conversion, missing value handling, and field mapping, and generating a corresponding sample primary key for each new set of data; second, writing the processed new data into the pending database to expand and update the database; then, based on the expanded database, incrementally training or fine-tuning the performance prediction model, updating the model's weight parameters to enable the model to absorb the latest experimental feedback information; finally, deploying the updated model to the next round of the design process for use in performance prediction of subsequent candidate solutions. For samples marked as anomalous (e.g., data generated during equipment failure or anomalous data caused by operational errors), the system will enter a review process, and after review and confirmation, they will be downweighted or removed, and will not be directly used as high-confidence training samples.

[0070] In some embodiments, determining the corresponding target process route based on R&D target information further includes: Based on the R&D objective information, determine the multi-dimensional route information for the current R&D task; among which, the multi-dimensional route information includes the target performance type, sample form and expected process path; Based on the degree of matching between multi-dimensional route information and preset route judgment criteria, the route confidence level of the target process route is determined.

[0071] Specifically, when determining the target process route based on R&D target information, the confidence level of the determined route can be further calculated. For example, the system can extract route basis information from the R&D target information in three dimensions: target performance type, sample form, and expected process path, and calculate the matching degree between these three dimensions and preset route determination criteria. For instance, if the target performance type is ionic conductivity, magnetic properties, or electrical functional properties, and the sample form is powder, compact, or sintered body, then it is determined to have a high matching degree with the rare earth solid-phase functional material route; conversely, if the target performance type is hardness, strength, or plasticity, and the sample form is alloy ingot, bulk, or metallographic sample, then it is determined to have a high matching degree with the rare earth metal material route.

[0072] Then, by assigning different weight coefficients to the matching degree of each of the three dimensions mentioned above and summing them by weight, the final route confidence score can be obtained. For example, the route confidence score can be expressed as a value between 0 and 100%, with a higher confidence score indicating a higher degree of certainty from the system regarding the route determination.

[0073] Based on this, by determining the route confidence level according to the matching degree between the target performance type, sample morphology, and expected process path and the preset judgment criteria, the reliability of route identification can be quantified, thereby providing a reliable basis for the use of subsequent simulation results.

[0074] In some embodiments, the performance of each candidate scheme is predicted using a performance prediction model combined with simulation results, including: The feature fusion weights of the simulation results relative to each candidate scheme are adjusted based on the route confidence; the feature fusion weights are positively correlated with the route confidence.

[0075] It should be noted that, based on the currently determined route confidence level, the system can adjust the feature fusion weights of the simulation results in subsequent performance predictions. There is a positive correlation between feature fusion weights and route confidence level. Specifically, when the route confidence level is high, it indicates that the system is relatively confident in its judgment of the current process route. In this case, the simulation module is based on the correct physical model, and its output simulation results have high reference value. Therefore, the fusion weight of the simulation results in the input features of the performance prediction model is correspondingly increased, allowing physical simulation information to play a greater constraining and guiding role in prediction. When the route confidence level is low, it indicates that the R&D target information may be relatively ambiguous or located in the boundary region between two types of routes. The system's confidence in its judgment of the process route is insufficient. In this case, the simulation results may have the risk of model selection bias. Therefore, the system correspondingly reduces the fusion weight of the simulation results in the performance prediction model, while relatively increasing the weights of other features (such as component features and process features extracted based on candidate schemes, and historical experimental data features extracted based on the database to be used), making the model prediction rely more on existing experimental data rather than potentially biased simulation results.

[0076] Based on this, the accuracy and reliability of model prediction can be further improved by adjusting the feature fusion weights of simulation results in performance prediction based on the positive correlation of route confidence.

[0077] In some embodiments, a simulation module corresponding to the target process route is invoked to perform simulation calculations on the target material system, and the corresponding simulation results are obtained, including: The cross-scale simulation module corresponding to the target process route is invoked to perform simulation calculations on the target material system, and the corresponding simulation results are obtained. Specifically, when the target process route is a rare earth solid-phase functional material route, the cross-scale simulation module includes at least one of the following: phase equilibrium model, diffusion model, sintering densification model, and grain growth model; when the target process route is a rare earth metal material route, the cross-scale simulation module includes at least one of the following: thermodynamic model, solidification model, precipitation model, and microstructure evolution model.

[0078] It should be noted that the cross-scale simulation module is equipped with different sets of simulation models for the rare earth solid-phase functional materials route and the rare earth metal materials route, in order to cover the different physical mechanisms and process characteristics of the two types of materials.

[0079] Specifically, when the target process route is a rare earth solid-phase functional material route, the system calls at least one of the following models from a pre-set simulation model library: phase equilibrium model, diffusion model, sintering densification model, and grain growth model. The phase equilibrium model takes composition, temperature, and atmosphere conditions as input to calculate the types and fractions of stable phases in the target material system under different conditions, used to determine the formation conditions of the target phase and the risk of impurity phases. The diffusion model takes diffusion coefficient, temperature, and time as input to calculate the long-range migration behavior of atoms or vacancies driven by concentration gradients, outputting the concentration distribution or diffusion distance, used to determine the degree of material migration during sintering or heat treatment. The sintering densification model takes initial grain size, green density, and sintering regime as input to calculate the shrinkage rate and relative density at different sintering temperatures and holding times, used to evaluate the degree of densification of the sample. The grain growth model takes initial grain size, temperature, and time as input to calculate the average grain size under different sintering or heat treatment conditions, used to evaluate the grain growth trend and determine the risk of abnormal grain growth. Taking tantalum-doped garnet-type solid electrolyte as an example, the system can call the phase equilibrium model to calculate the stable phase types and phase fractions under different Ta doping amounts and temperature conditions, call the sintering densification model to calculate the shrinkage rate and relative density under different sintering regimes, and call the grain growth model to predict the grain size under different holding times.

[0080] When the target process route is a rare earth metal material route, the system calls at least one of the following models: thermodynamic model, solidification model, precipitation model, and microstructure evolution model. The thermodynamic model takes composition and temperature as input to calculate the phase stability range and equilibrium phase composition, used to determine the solidus and liquidus temperatures and precipitation temperature windows of the alloy. The solidification model takes cooling rate and compositional undercooling as input to simulate the solidification path and dendrite parameters during melt cooling, used to predict as-cast microstructure characteristics and elemental segregation. The precipitation model takes solution temperature and aging regime as input to simulate the nucleation, growth, and coarsening behavior of precipitates during aging, outputting the types and size distribution of precipitates, used to evaluate the age-strengthening potential. The microstructure evolution model takes grain size and precipitate distribution as input, combining the calculation results of the aforementioned models to output the microstructure evolution results, used to predict the grain size, phase distribution, and uniformity of precipitate distribution of the final material. Taking Mg-Gd-Y-Zn-Zr rare earth magnesium alloy as an example, the system calls the thermodynamic model to calculate the phase stability range of different Gd and Y contents, calls the precipitation model to predict the types and size distribution of the main precipitates under different aging regimes, and calls the microstructure evolution model to predict the grain size and uniformity of precipitate distribution under different heat treatment regimes.

[0081] Based on this, corresponding cross-scale simulation model sets are configured for functional materials and metallic materials respectively, thereby covering the simulation requirements of physical properties of the two types of rare earth materials.

[0082] In some embodiments, a cross-scale simulation module corresponding to the target process route is invoked to perform simulation calculations on the target material system, and the corresponding simulation results are obtained, including: When the target process route is a rare earth solid functional material route, the formation energy and diffusion coefficient obtained from low-scale simulation calculations are used as input parameters for the phase equilibrium model and diffusion model. Based on the stable phase type, phase fraction and phase transformation temperature output by the phase equilibrium model, the temperature boundary conditions and initial phase composition conditions of the grain growth model and the sintering densification model are used. Based on the concentration distribution output by the diffusion model, the average grain size output by the grain growth model, and the shrinkage rate and relative density output by the sintering densification model, the corresponding simulation results are obtained.

[0083] It should be noted that when calling the cross-scale simulation module for simulation calculations, data is transferred between simulations at different scales in the order of "low-scale parameter extraction - mesoscale process simulation - macroscopic performance prediction".

[0084] Specifically, when the target process route is a rare earth solid-phase functional material route, the system performs cross-scale simulation calculations according to the following data transfer order: First, the formation energy and diffusion coefficient obtained from low-scale first-principles calculations are used as input parameters for the phase equilibrium model and the diffusion model. After the formation energy is input into the phase equilibrium model, the types, phase fractions, and phase transition temperatures of stable phases under different temperatures and compositions are calculated. Similarly, after the diffusion coefficient is input into the diffusion model, the concentration distribution or diffusion distance under different sintering conditions is calculated. Second, the phase transition temperature and stable phase types output from the phase equilibrium model are used as the temperature boundary conditions and initial phase composition conditions for the grain growth model and the sintering densification model. The grain growth model calculates the average grain size based on these conditions, and the sintering densification model calculates the shrinkage rate and relative density. Finally, the concentration distribution output from the diffusion model, the average grain size output from the grain growth model, and the shrinkage rate and relative density output from the sintering densification model are collectively used as auxiliary input features for the performance prediction model.

[0085] Based on this, by establishing data transfer from low-scale formation energy and diffusion coefficient to phase equilibrium, grain growth and sintering densification under the functional materials route, the effectiveness of simulation results can be further improved.

[0086] In some embodiments, a cross-scale simulation module corresponding to the target process route is invoked to perform simulation calculations on the target material system, and the corresponding simulation results are obtained, including: When the target process route is a rare earth metal material route, the formation energy and diffusion barrier obtained from low-scale simulation calculations are used as input parameters for the thermodynamic model, solidification model and precipitation model. The phase stability range and phase composition output by the thermodynamic model are used as the temperature window and initial phase field conditions for the solidification model and precipitation model. The solidification path and dendrite parameters output by the solidification model are used as the initial microstructure input for the precipitation model. Based on the types and size distribution of precipitated phases output by the precipitation model, the phase composition output by the thermodynamic model, and the dendrite parameters output by the solidification model, the corresponding simulation results are obtained.

[0087] Specifically, when the target process route is a rare earth metal material route, the system performs cross-scale simulation calculations according to the following data transfer order: First, the formation energy and diffusion barrier obtained from low-scale first-principles calculations are used as input parameters for the thermodynamic model, solidification model, and precipitation model. The thermodynamic model, using the formation energy as input, calculates the phase stability range and phase composition under different composition and temperature conditions. The phase stability range and phase composition output by the thermodynamic model are further used as the temperature window and initial phase field conditions for the solidification and precipitation models. Based on this, the solidification model simulates the solidification path and dendrite parameters. The solidification path and dendrite parameters output by the solidification model serve as the initial microstructure input for the precipitation model, which then outputs the type and size distribution of precipitated phases. Finally, the phase composition output by the thermodynamic model, the dendrite parameters output by the solidification model, and the type and size distribution of precipitated phases output by the precipitation model are collectively input into the microstructure evolution model for summarization. The grain size, phase distribution, and precipitated phase distribution output by the microstructure evolution model serve as auxiliary input features for the performance prediction model.

[0088] Based on this, by establishing data transfer from low-scale formation energy and diffusion barrier to thermodynamics, solidification, and precipitation models under the metallic material route, the effectiveness of simulation results can be further improved.

[0089] In some embodiments, the performance of each candidate scheme is predicted using a performance prediction model combined with simulation results, resulting in performance prediction results for each candidate scheme, including: The performance prediction model of the multi-task prediction architecture is used to predict the performance of each candidate solution. The input parameters of the performance prediction model include component features and process features extracted based on candidate schemes, simulation features extracted based on simulation results, and structural features, characterization image features, and historical performance features extracted from the database to be used. The multi-task prediction architecture includes a unified feature input layer, a task prediction layer, and a result aggregation layer. The task prediction layer is used to establish corresponding prediction tasks for different prediction objectives. The prediction tasks include phase composition classification tasks, performance regression tasks, tissue parameter regression tasks, and experimental risk discrimination tasks.

[0090] It should be noted that the performance prediction model adopts a multi-task prediction architecture, which specifically includes a three-layer architecture of "unified feature input layer - task prediction layer - result aggregation layer".

[0091] Specifically, the unified feature input layer is responsible for receiving input features from different sources and standardizing them into a unified tensor format. The input features include component features and process features extracted based on candidate schemes, simulation features extracted based on simulation results, and structural features, characterizing image features, and historical performance features extracted from the database to be used.

[0092] The task prediction layer establishes corresponding regression or classification tasks for different prediction objectives, including: a phase composition classification task to determine whether candidate schemes can form the target phase; a performance regression task to predict specific values ​​such as ionic conductivity or hardness; a microstructure parameter regression task to predict grain size or precipitate size distribution; and an experimental risk discrimination task to assess the risk of impurity phase formation or cracking. Each task shares the feature representation of a unified feature input layer, but each has independent task parameters and an output layer, thus achieving multi-task joint learning.

[0093] The results aggregation layer integrates the outputs of each task, outputting the comprehensive performance prediction value of each candidate solution and the corresponding uncertainty evaluation. The uncertainty evaluation can be obtained using methods such as Bayesian neural networks, deep ensembles, or MC-Dropout (Monte Carlo Dropout), reflecting the reliability of the model's prediction results for that candidate solution. The uncertainty evaluation is used in the objective function calculation during subsequent solution selection, ensuring that the system considers both the quality of prediction performance and the reliability of the prediction results during the selection process.

[0094] For example, the model training and application process mainly includes three types of input features: historical data features (composition features, process features, historical performance features, etc.) are mainly used to train the model and establish the mapping relationship between composition, process and performance; simulation result features (simulation features) are mainly used to supplement auxiliary evaluation information such as phase stability, sintering window or microstructure evolution, and provide physical constraints for prediction; material mechanism features (structural features, characterization image features, etc.) are mainly used as constraints and interpretation basis to eliminate candidate schemes that do not conform to the laws of materials.

[0095] Based on this, a performance prediction model using a multi-task prediction architecture is adopted to perform multi-task prediction and aggregation based on multi-dimensional features such as composition, process, simulation, structure, image and historical performance, thereby improving the accuracy of performance prediction.

[0096] In some embodiments, based on performance prediction results, target performance requirements, and process boundary conditions, combined with the experimental line equipment capability boundaries and allowable parameter ranges, candidate schemes are screened to obtain the target experimental scheme, including: Based on the preset objective function and constraints, at least one preliminary selection scheme is obtained by screening each candidate scheme; wherein, the objective function includes a performance prediction value weighting term and a model uncertainty penalty term; the constraints include phase stability constraints, impurity phase content constraints and safety constraints. Based on the preset experimental line equipment capability boundaries and allowable parameter ranges, the initial selection schemes are screened to obtain the target experimental scheme.

[0097] Specifically, firstly, a multi-objective evaluation is performed on the performance prediction results of each candidate scheme based on a preset objective function and constraints. The objective function includes a performance prediction value weighting term and a model uncertainty penalty term. For example, the objective function can be expressed as:

[0098] in, This represents the overall evaluation score of the candidate solutions. This represents a combination vector of all adjustable variables in the candidate scheme. Indicates the first i There are 10 metrics (the total number of performance metrics that need to be optimized is 100). n The predicted performance value of ) This represents the weighting coefficient of each performance indicator. For the first i A normalized evaluation function for each performance index (used to map performance indices of different dimensions to comparable intervals). U(x) For model uncertainty, This represents the uncertainty penalty coefficient. It is understandable that the objective function is designed to maximize target performance while penalizing schemes with higher prediction uncertainty, thereby guiding the system to prioritize schemes that offer both superior performance and reliable prediction.

[0099] The constraints include phase stability constraints (e.g., the probability of target phase formation is not lower than a preset threshold to ensure that the candidate scheme has the thermodynamic conditions for forming the target phase), impurity content constraints (e.g., the impurity fraction is not higher than a preset threshold to avoid the impurities adversely affecting performance), and safety constraints (e.g., the process temperature does not exceed the material decomposition temperature or the upper limit of the equipment's safe temperature). The mathematical form of these constraints can be expressed as:

[0100]

[0101] in, Indicates the first j There are 1 constraint condition (the total number of constraint conditions is 1). m ), used to quantify candidate solutions x Whether a certain physical or engineering constraint is met, including constraints such as phase stability, impurity content, and safety. and They represent the first k The lower and upper limits of each parameter. Indicates the first candidate solution k The specific value of each adjustable parameter. p This represents the total number of adjustable parameters in the candidate scheme.

[0102] Based on this, the system can select the candidate scheme that satisfies all constraints and optimizes the objective function from among the candidate schemes, and use it as the initial selection scheme.

[0103] Then, the system obtains the equipment capability boundaries and allowable parameter ranges of the subsequent automated experimental verification line, which can be expressed as:

[0104] in, and These represent the lower and upper limits of the parameters allowed by the equipment. Equipment capability boundaries include, but are not limited to, discrete constraints such as the equipment's maximum operating temperature, maximum pressing pressure, weighing accuracy, heating rate limit, atmosphere control capability, and characterization throughput. For example, the system will reject any initial selection scheme that exceeds the equipment's allowable parameter range; similarly, it will reject schemes involving conditions not supported by the equipment (e.g., requiring the use of a special atmosphere not available on the experimental line or exceeding the equipment's maximum capacity).

[0105] After two-step screening, the system outputs at least one set of target experimental schemes that meet all constraints and generates a standardized scheme sheet.

[0106] Based on this, by first screening preliminary schemes based on constraints such as objective function and phase stability, and then conducting secondary screening based on the capability boundary of experimental line equipment, it is possible to comprehensively screen out executable candidate schemes that take into account both performance and compliance with constraints, thereby further improving the accuracy and reliability of the output scheme.

[0107] It should be noted that, compared with the prior art, the embodiments of this application have the following beneficial effects: 1. Improve the design efficiency of rare earth materials and shorten the time from target proposal to candidate solution output.

[0108] 2. Improve the quality of the solution by reducing the proportion of blind trial and error through the coupling of simulation and data-driven approaches.

[0109] 3. Strengthen the ability to accumulate and reuse knowledge, and achieve continuous accumulation of data, models and rules.

[0110] 4. Provide standardized and executable inputs for subsequent automated verification lines, improving overall closed-loop R&D efficiency.

[0111] Please refer to Figure 2 , Figure 2 This paper illustrates a block diagram of a multi-module collaborative scheme generation apparatus for rare earth solid materials, provided in some embodiments of this application. It should be understood that this multi-module collaborative scheme generation apparatus for rare earth solid materials is similar to the one described above. Figure 1Corresponding to the method embodiments, it is able to execute the various steps involved in the above method embodiments. The specific functions of the multi-module collaborative scheme generation device for rare earth solid phase materials can be found in the description above. To avoid repetition, detailed descriptions are appropriately omitted here.

[0112] Figure 2 The multi-module collaborative solution generation device for rare earth solid materials includes at least one software functional module that can be stored in a memory or embedded in the multi-module collaborative solution generation device for rare earth solid materials in the form of software or firmware. The multi-module collaborative solution generation device for rare earth solid materials includes: The information acquisition module 210 is used to acquire R&D target information, including target material system, target performance requirements, and process boundary conditions. Data construction module 220 is used to build a ready database for R&D target information based on pre-collected multi-source data; The route determination module 230 is used to determine the corresponding target process route based on the R&D target information; among which, the target process routes include rare earth solid-phase functional material routes and rare earth metal material routes. Simulation calculation module 240 is used to call the simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtain the corresponding simulation results. Candidate building module 250 is used to build a set of candidate schemes based on the target material system, process boundary conditions and historical sample distribution in the database to be used; wherein, each candidate scheme includes composition parameters and process parameters; The performance prediction module 260 is used to perform performance prediction on each candidate scheme by combining the performance prediction model with the simulation results, and obtain the performance prediction results of each candidate scheme. The scheme screening module 270 is used to screen each candidate scheme based on the performance prediction results, target performance requirements and process boundary conditions, combined with the experimental line equipment capacity boundary and allowable parameter range, to obtain the target experimental scheme.

[0113] It is understood that the above-described device embodiments correspond to the method embodiments of this application. The multi-module collaborative scheme generation device for rare earth solid materials provided by the embodiments of this application can realize the multi-module collaborative scheme generation method for rare earth solid materials provided by any one of the method embodiments of this application.

[0114] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the device described above can be referred to the corresponding process in the aforementioned method, and will not be elaborated further here.

[0115] like Figure 3As shown, some embodiments of this application provide an electronic device 300, which includes a memory 310, a processor 320, and a computer program stored in the memory 310 and executable on the processor 320. When the processor 320 reads the program from the memory 310 via a bus 330 and executes the program, it can implement any of the methods included in the above-described method for generating multi-module collaborative schemes for rare earth solid phase materials.

[0116] Processor 320 can process digital signals and may include various computing architectures. For example, it may be a complex instruction set computer architecture, a reduced instruction set computer architecture, or an architecture that implements multiple instruction set combinations. In some examples, processor 320 may be a microprocessor.

[0117] The memory 310 can be used to store instructions executed by the processor 320 or data related to the execution of instructions. These instructions and / or data may include code used to implement some or all of the functions of one or more modules described in the embodiments of this application. The processor 320 of this embodiment can be used to execute the instructions in the memory 310 to implement the methods shown above. The memory 310 includes dynamic random access memory, static random access memory, flash memory, optical memory, or other memories well known to those skilled in the art.

[0118] Some embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, describes the method described in the method embodiments.

[0119] Some embodiments of this application also provide a computer program product that, when run on a computer, causes the computer to perform the method described in the method embodiments.

[0120] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For apparatus embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0121] It should be understood, in the several embodiments provided in this application, that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0122] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0123] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0124] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0125] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0126] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for generating multi-module collaborative schemes for rare earth solid materials, characterized in that, include: Obtain research and development target information; wherein, the research and development target information includes target material system, target performance requirements, and process boundary conditions; Based on pre-collected multi-source data, a ready-to-use database is constructed for the R&D target information; The corresponding target process routes are determined based on the aforementioned R&D target information; wherein, the target process routes include rare earth solid-phase functional material routes and rare earth metal material routes; The simulation module corresponding to the target process route is invoked to perform simulation calculations on the target material system, and the corresponding simulation results are obtained. Based on the target material system, the process boundary conditions, and the historical sample distribution in the database to be used, a set of candidate solutions is constructed; wherein each candidate solution includes composition parameters and process parameters; The performance prediction model is used in conjunction with the simulation results to predict the performance of each candidate scheme, and the performance prediction results of each candidate scheme are obtained. Based on the performance prediction results, the target performance requirements, and the process boundary conditions, combined with the experimental line equipment capability boundaries and allowable parameter ranges, each candidate scheme is screened to obtain the target experimental scheme.

2. The method of claim 1, wherein the method is a multi-module synergistic approach for rare earth solid state material facing, and The step of determining the corresponding target process route based on the R&D target information further includes: Based on the aforementioned R&D objective information, multi-dimensional route information for the current R&D task is determined; wherein, the multi-dimensional route information includes target performance type, sample morphology, and expected process path; Based on the degree of matching between the multi-dimensional route information and the preset route determination criteria, the route confidence level of the target process route is determined.

3. The method of claim 2, wherein the method further comprises: The process of using a performance prediction model in conjunction with the simulation results to predict the performance of each candidate solution includes: The feature fusion weights of the simulation results relative to each of the candidate schemes are adjusted based on the route confidence; wherein the feature fusion weights are positively correlated with the route confidence.

4. The method of claim 1, wherein the method is a multi-module synergistic approach for rare earth solid state material facing, and The step of calling the simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtaining the corresponding simulation results includes: The target material system is simulated by calling the cross-scale simulation module corresponding to the target process route, and the corresponding simulation results are obtained. Wherein, when the target process route is a rare earth solid-phase functional material route, the cross-scale simulation module includes at least one of a phase equilibrium model, a diffusion model, a sintering densification model, and a grain growth model; where the target process route is a rare earth metal material route, the cross-scale simulation module includes at least one of a thermodynamic model, a solidification model, a precipitation model, and a microstructure evolution model.

5. The method of claim 4, wherein the method further comprises: The step of calling the cross-scale simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtaining the corresponding simulation results includes: When the target process route is a rare earth solid-phase functional material route, the formation energy and diffusion coefficient obtained based on low-scale simulation calculations are used as input parameters for the phase equilibrium model and the diffusion model. The stable phase type, phase fraction, and phase transformation temperature output by the phase equilibrium model are used as the temperature boundary conditions and initial phase composition conditions for the grain growth model and the sintering densification model. Based on the concentration distribution output by the diffusion model, the average grain size output by the grain growth model, and the shrinkage rate and relative density output by the sintering densification model, the corresponding simulation results are obtained.

6. The method of claim 4, wherein the method further comprises: The step of calling the cross-scale simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtaining the corresponding simulation results includes: When the target process route is a rare earth metal material route, the formation energy and diffusion barrier obtained based on low-scale simulation calculations are used as input parameters for the thermodynamic model, the solidification model, and the precipitation model. The phase stability range and phase composition output by the thermodynamic model are used as the temperature window and initial phase field conditions for the solidification model and the precipitation model. The solidification path and dendrite parameters output by the solidification model are used as the initial microstructure input for the precipitation model. Based on the precipitate type and size distribution output by the precipitation model, the phase composition output by the thermodynamic model, and the dendrite parameters output by the solidification model, the corresponding simulation results are obtained.

7. The method of claim 1, wherein the method is a multi-module, synergistic approach to rare earth solid state material growth. The performance prediction of each candidate scheme is performed using the performance prediction model in conjunction with the simulation results, resulting in the performance prediction results for each candidate scheme, including: The performance prediction model of the multi-task prediction architecture is used to predict the performance of each of the candidate solutions. The input parameters of the performance prediction model include component features and process features extracted based on the candidate schemes, simulation features extracted based on the simulation results, and structural features, characterization image features, and historical performance features extracted from the database to be used. The multi-task prediction architecture includes a unified feature input layer, a task prediction layer, and a result aggregation layer. The task prediction layer is used to establish corresponding prediction tasks for different prediction objectives. The prediction tasks include phase composition classification tasks, performance regression tasks, tissue parameter regression tasks, and experimental risk discrimination tasks.

8. The method of claim 1, wherein the method is a multi-module, synergistic approach to rare earth solid state material facing, and Based on the performance prediction results, the target performance requirements, and the process boundary conditions, combined with the experimental line equipment capability boundaries and allowable parameter ranges, the candidate schemes are screened to obtain the target experimental scheme, including: Based on a preset objective function and constraints, at least one preliminary selection scheme is obtained by screening each candidate scheme; wherein, the objective function includes a performance prediction value weighting term and a model uncertainty penalty term; the constraints include phase stability constraints, impurity phase content constraints, and safety constraints; The preliminary selection schemes are screened based on the preset experimental line equipment capability boundaries and allowable parameter ranges to obtain the target experimental scheme.

9. A multi-module collaborative scheme generation apparatus for rare earth solid state materials, characterized in that, include: The information acquisition module is used to acquire R&D target information; wherein, the R&D target information includes target material system, target performance requirements, and process boundary conditions; The data construction module is used to build a ready-to-use database based on pre-collected multi-source data, which is oriented towards the R&D target information; The route determination module is used to determine the corresponding target process route based on the R&D target information; wherein, the target process route includes rare earth solid-phase functional material route and rare earth metal material route; The simulation calculation module is used to call the simulation module corresponding to the target process route to perform simulation calculations on the target material system and obtain the corresponding simulation results. A candidate construction module is used to construct a set of candidate schemes based on the target material system, the process boundary conditions, and the historical sample distribution in the database to be used; wherein each candidate scheme includes composition parameters and process parameters; The performance prediction module is used to perform performance prediction on each of the candidate solutions using the performance prediction model and the simulation results, and to obtain the performance prediction results of each of the candidate solutions. The scheme screening module is used to screen each of the candidate schemes based on the performance prediction results, the target performance requirements, and the process boundary conditions, combined with the experimental line equipment capability boundaries and allowable parameter ranges, to obtain the target experimental scheme.

10. An electronic device, comprising: The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the multi-module collaborative scheme generation method for rare earth solid phase materials as described in any one of claims 1-8.

11. A computer readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, performs the method for generating multi-module collaborative schemes for rare earth solid phase materials as described in any one of claims 1-8.

12. A computer program product, characterised in that, The computer program product includes a computer program that, when executed by a processor, implements the multi-module collaborative scheme generation method for rare earth solid phase materials as described in any one of claims 1-8.