Data-driven research system for catalyst

Abstract

The invention discloses a data-driven research system for a catalyst, and relates to the technical field of computer and material science crossing, and the system comprises a catalyst database retrieval subsystem, a catalyst high-throughput calculation and multi-scale simulation subsystem and a catalyst artificial intelligence subsystem, the integrated research and development process of catalyst data retrieval, structural design, computational simulation and performance prediction is realized, the research and development cost of the catalyst is effectively reduced, and the research and development efficiency of the catalyst is improved.

Description

Technical Field

[0001] The present application relates to the interdisciplinary field of computer and material science, and in particular to a data-driven research system for photo / electric / thermal catalysts. Background Art

[0002] Catalysts are the core elements of the modern chemical industry. They are widely used in many industries such as petroleum refining, polymer synthesis, drug production and environmental governance. According to statistics, about 20%-30% of the global GDP is related to catalysis-related industries. With the rapid expansion of the chemical industry, the demand for high-performance catalysts continues to rise, and market competition has also intensified. However, the patents of catalysts that many key chemical raw materials rely on are in the hands of international chemical giants, which has brought challenges to the development of domestic chemical companies in the field of catalysts.

[0003] The traditional catalyst R&D model relies on expert experience and laboratory synthesis, a trial-and-error approach that is costly and has a long R&D cycle.

[0004] Therefore, there is an urgent need for an intelligent system that is low-cost and can improve the efficiency of catalyst research and development. Summary of the invention

[0005] The purpose of this application is to provide a data-driven research system for catalysts, which can reduce the cost of catalyst research and development and improve the efficiency of catalyst research and development.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] The present application provides a data-driven research system for catalysts, comprising:

[0008] Catalyst database retrieval subsystem, catalyst high-throughput calculation and multi-scale simulation subsystem, and catalyst artificial intelligence subsystem;

[0009] The catalyst database retrieval subsystem is connected to the catalyst high-throughput calculation and multi-scale simulation subsystem and the catalyst artificial intelligence subsystem;

[0010] The catalyst high-throughput computing and multi-scale simulation subsystem is connected to the catalyst artificial intelligence subsystem;

[0011] The catalyst database retrieval subsystem is used to store a number of catalyst information and retrieve target catalyst information according to the retrieval conditions, wherein the catalyst information includes catalyst models and attribute data, and the attribute data includes structural data, physical property data, calculated property data and experimental characterization data;

[0012] The catalyst high-throughput calculation and multi-scale simulation subsystem is used to generate a catalyst structure according to the target catalyst information, and perform quantum chemical calculation, density functional calculation and molecular dynamics simulation on the generated catalyst structure to obtain catalyst calculation simulation data;

[0013] The catalyst artificial intelligence subsystem is used to predict and classify the performance of the target catalyst and to fit the potential energy function of the catalyst system based on the target catalyst information and the catalyst computational simulation using a machine learning algorithm, wherein the catalyst system includes a target catalyst, reactants, intermediates, and products.

[0014] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0015] The present application provides a data-driven research system for catalysts, which includes a catalyst database retrieval subsystem, a catalyst high-throughput computing and multi-scale simulation subsystem, and a catalyst artificial intelligence subsystem, realizing an integrated R&D process of catalyst data retrieval, structural design, computational simulation, and performance prediction, effectively reducing the cost of catalyst R&D and improving the efficiency of catalyst R&D. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is a schematic diagram of the architecture of a data-driven research system for a catalyst in an embodiment of the present application;

[0018] Figure 2 This is a schematic diagram of a user login page in an embodiment of the present application;

[0019] Figure 3 This is a schematic diagram of a personal information management page in an embodiment of the present application;

[0020] Figure 4 This is a schematic diagram of an administrator user management subsystem page in an embodiment of the present application;

[0021] Figure 5 Schematic diagram of the catalyst database retrieval subsystem page in the embodiment of this application

[0022] Figure 6 This is a schematic diagram of a catalyst search page in the embodiments of the present application;

[0023] Figure 7A schematic diagram of a search result list in an embodiment of the present application;

[0024] Figure 8 This is a schematic diagram of a search result details page in an embodiment of the present application;

[0025] Fig. 9 This is a schematic diagram of the catalyst high-throughput calculation and multi-scale simulation subsystem page in the embodiment of this application;

[0026] Fig.10 This is a schematic diagram of the catalyst artificial intelligence subsystem page in an embodiment of the present application. DETAILED DESCRIPTION

[0027] Research has found that countries around the world are actively deploying data-driven new material research and development strategies, viewing it as a key way to enhance competitiveness in the field of high-tech materials. Given that this research and development paradigm relies on a large number of high-quality data sets, the construction and development of domestic and foreign material databases has become particularly urgent.

[0028] Currently, there are many materials databases that are open for access around the world, such as MaterialsProject, AFLOW and OQMD in the United States, NOMAD, Materials Cloud and AiiDA in Europe, MatCloud, Atomly.net and Materials Informatics Platform in China, etc. These databases are very mature, but few of them focus specifically on catalysts, and most of them only provide material data retrieval and sharing functions. In addition, the high-throughput design, calculation and multi-scale simulation of catalysts are heavily dependent on repetitive manual operations and task submission, which greatly affects research efficiency; the popularization of the data-driven materials research and development paradigm faces another challenge: it requires researchers to have certain computer programming skills and a basic application foundation for computing software.

[0029] Moreover, in the context of the digital age, a new R&D paradigm - data-driven materials R&D - has begun to attract attention. This approach uses big data and computer technology to accelerate the design and discovery process of new materials, and has achieved remarkable results in the fields of superconductivity, quantum, topology, and optoelectronics. Applying the data-driven R&D paradigm to the field of catalysts can help domestic companies break through technical barriers, shorten R&D cycles, and reduce costs, thereby promoting the innovation and development of catalyst technology.

[0030] Based on this, the present embodiment provides a data-driven research system for catalysts, which integrates functions such as catalyst data retrieval, high-throughput design of catalyst structure, high-throughput calculation and multi-scale simulation, catalyst performance prediction and classification. The system focuses on the field of catalysts, integrates fragmented catalyst-related data, and continues to accumulate, providing data support for catalyst research, and covers the full process of catalyst data retrieval, automated design, and computational simulation research, realizing "one-stop" catalyst research and development. Each module in the system uses a graphical operating interface, which lowers the research and development threshold. Even if the researcher does not have computer programming skills, it can still be easily based on the system. Data-driven catalyst research and development, thereby promoting continuous progress and innovation in the field of catalyst research.

[0031] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0032] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0033] This embodiment builds a front-end visualization page based on React, Ts, Vite, and mui, builds a back-end based on Python and FastAPI framework, and builds a database based on PosgreSQL. The front-end, back-end, and database form a data-driven research system for catalysts. The system includes:

[0034] Catalyst database retrieval subsystem, catalyst high-throughput calculation and multi-scale simulation subsystem, and catalyst artificial intelligence subsystem.

[0035] The catalyst database retrieval subsystem is connected to the catalyst high-throughput computing and multi-scale simulation subsystem and the catalyst artificial intelligence subsystem.

[0036] The catalyst high-throughput computing is connected to the multi-scale simulation subsystem and the catalyst artificial intelligence subsystem.

[0037] The catalyst database retrieval subsystem is used to store a number of catalyst information and retrieve target catalyst information according to retrieval conditions, wherein the retrieval conditions are chemical elements or chemical formulas, the catalyst information includes catalyst models and attribute data, the attribute data includes structural data, physical property data, calculated property data and experimental characterization data, and the experimental characterization data includes XRD data, nuclear magnetic resonance data, atomic absorption spectroscopy data, specific surface area, SEM data, XPS data and Fourier transform infrared spectroscopy data.

[0038] The catalyst high-throughput calculation and multi-scale simulation subsystem is used to generate a catalyst structure according to the target catalyst information, and perform quantum chemical calculation, density functional calculation and molecular dynamics simulation on the generated catalyst structure to obtain catalyst calculation simulation data.

[0039] The catalyst artificial intelligence subsystem is used to predict and classify the performance of the target catalyst and to fit the potential energy function of the catalyst system based on the target catalyst information and the catalyst computational simulation using a machine learning algorithm, wherein the catalyst system includes a target catalyst, reactants, intermediates, and products.

[0040] The catalyst database retrieval subsystem includes an electrocatalyst retrieval module, a photocatalyst retrieval module, and a thermal catalyst retrieval module.

[0041] The electrocatalyst search module is used to search for a target electrocatalyst according to a search condition and display information of the target electrocatalyst.

[0042] The photocatalyst search module is used to search for a target photocatalyst according to a search condition and display the target photocatalyst information.

[0043] The thermal catalyst search module searches for a target thermal catalyst according to a search condition and displays information of the target thermal catalyst.

[0044] The electrocatalyst search module, the photocatalyst search module, and the thermal catalyst search module all include: a query interface display component and an element periodic table interface selection component.

[0045] The periodic table selection component is used to select chemical elements according to the periodic table.

[0046] Users can directly select chemical elements as search conditions for fuzzy search through the periodic table, or enter chemical formulas for precise search.

[0047] The query display interface component is used to query and display the target catalyst information according to the search conditions.

[0048] The query display interface component includes a search input box, a search result list and a search result details interface.

[0049] The search input box is used to input the search condition.

[0050] The search result list is used to display rough information of target catalysts that meet the search conditions, wherein the rough information of the catalyst includes ID, chemical formula, Smiles code and adsorption energy.

[0051] The search result details interface is used to display detailed information of the target catalyst that meets the search conditions, wherein the target catalyst detailed information includes the attribute data of the target catalyst.

[0052] The catalyst database retrieval subsystem also includes: a data processing module and a catalyst data knowledge spectrum module.

[0053] The data processing module is used to provide data cleaning services, data extraction services and data analysis services.

[0054] The catalyst data knowledge spectrum module is used to organize and classify the attribute data in the catalyst information to form a catalyst data knowledge spectrum.

[0055] The catalyst high-throughput calculation and multi-scale simulation subsystem is used to design catalyst structures, and perform quantum chemical calculations and molecular dynamics simulations based on the generated catalyst structures.

[0056] The catalyst high-throughput calculation and multi-scale simulation subsystem specifically includes: a catalyst structure input control module and a quantum chemistry and molecular dynamics calculation module.

[0057] The catalyst structure input control module is connected with the quantum chemistry and molecular dynamics calculation modules.

[0058] The catalyst structure input control module is used to generate a catalyst structure according to the target catalyst information.

[0059] The catalyst structure input control module provides catalyst structure information for calculations of the quantum chemistry and molecular dynamics calculation modules.

[0060] The quantum chemistry and molecular dynamics calculation module is used to perform quantum chemistry calculation, density functional calculation and molecular dynamics simulation according to the generated catalyst structure to obtain catalyst calculation simulation data.

[0061] Quantum chemical calculations include molecular structure optimization, IRC (Intrinsic Reaction Coordinate) calculations, transition state calculations and static calculations. The optimized structure file mol can be obtained. The HOMO / LUMO energy level distribution, transition state activation energy, electrostatic potential, charge distribution, etc. can be obtained by analyzing the optimized output wave function file.

[0062] Density functional calculations include structural optimization and static calculations. Structural optimization can obtain the energy-minimized structure, and the output file of static calculations can obtain information such as band structure, Bader charge, state density and partial density, crystal orbital layout analysis, etc.

[0063] Molecular dynamics simulation can obtain energy trajectory, force trajectory and structure trajectory files.

[0064] The catalyst structure input control module includes a catalyst structure high-throughput generation component, a catalyst structure online design component, a catalyst structure import component and a catalyst calculation simulation task batch automatic submission component.

[0065] The catalyst structure high-throughput generation component is used to automatically generate a plurality of catalyst structures according to the target catalyst information and set conditions, wherein the catalyst structure includes a molecular structure and a periodic structure.

[0066] The catalyst structure online design component is used to customize the design of the catalyst structure according to the target catalyst information and user requirements.

[0067] The catalyst structure import component is used to upload the catalyst structure locally.

[0068] The catalyst calculation simulation task batch automatic submission component is used to submit calculation tasks and simulation tasks related to catalysts in batches.

[0069] Specifically, the catalyst structure high-throughput generation component is a self-compiled program that can realize automatic batch generation of catalyst structures according to set conditions. The catalyst structure online design component can manually construct catalyst structures according to user customization. The catalyst structure import component allows users to upload catalyst structures from local computers to the module.

[0070] The quantum chemistry and molecular dynamics calculation module includes a calculation parameter setting component, a VASP calculation component, a GAUSSIAN calculation component, and a LAMMPS calculation component.

[0071] The calculation parameter setting component is used to set relevant parameters of quantum chemistry and molecular dynamics calculation software, and can realize simulation related to quantum chemistry, first principles and molecular dynamics of catalysts.

[0072] The VASP computing component, GAUSSIAN computing component, and LAMMPS computing component are used for submitting computing tasks of VASP, GAUSSIAN, and LAMMPS, respectively. Specifically, the VASP computing component is used to call the VASP program to perform density functional calculations according to the generated catalyst structure; the GAUSSIAN computing component is used to call GAUSSIAN to perform quantitative calculations according to the generated catalyst structure; and the LAMMPS computing component is used to call LAMMPS to perform molecular dynamics calculations according to the catalyst structure.

[0073] The catalyst artificial intelligence subsystem is used to predict and classify the performance of the target catalyst and to fit the potential energy function of the catalyst system based on the target catalyst information and the catalyst computational simulation using a machine learning algorithm, wherein the catalyst system includes a target catalyst, reactants, intermediates, and products.

[0074] Users can freely choose the data for machine learning training. Of course, the initial feature data is huge and complex. In the feature extraction process, principal component analysis is mainly used to extract features with a relatively large proportion. The weight of each feature will be visualized in the module. Users can analyze the weights to remove features with smaller weights and retain features with larger weights to reduce the dimension of features. In addition, these feature data can be summarized and uploaded by users themselves. Users can add or delete feature columns according to their own definitions. Feature data can also be obtained from other catalyst database retrieval subsystems and catalyst high-throughput calculation and multi-scale simulation subsystems. If the features entered by the user already exist in the system, the training list will be automatically displayed. For example, if the user enters the element C, the list will automatically display the corresponding values ​​of features such as electronegativity, number of valence electrons, electron affinity, atomic radius, first ionization energy, formation enthalpy, atomic mass, etc. These values ​​are stored in the catalyst database retrieval subsystem.

[0075] When performing performance prediction and performance classification, the characteristic input data used include: the intrinsic properties of the elements that make up the catalyst itself and the electronic structure properties of the catalyst itself calculated using quantum chemistry or first principles methods. The intrinsic properties include the type of element, electronegativity, number of valence electrons, electron affinity, atomic radius, first ionization energy, formation enthalpy and atomic mass. The electronic structure properties include energy band structure, state density distribution, charge population and HOMO / LUMO orbital energy level distribution.

[0076] The performance prediction and classification target values ​​used in machine learning can be the activation energy, adsorption free energy or decomposition temperature of the catalyst's decomposition transition state for a molecule or intermediate.

[0077] The training data for the catalyst machine learning potential energy function includes the trajectory files obtained in the catalyst high-throughput calculation and multi-scale simulation subsystem. The trajectory files contain a series of information such as the element type, atomic arrangement, energy and force of the catalyst during the calculation process. The catalyst artificial intelligence subsystem can fit the potential energy function based on the information extracted from the trajectory files.

[0078] The catalyst artificial intelligence subsystem specifically includes: a machine learning modeling module and a potential energy function development module.

[0079] The machine learning modeling module is used to predict the performance of the target catalyst according to a performance prediction model, and to classify the performance of the target catalyst according to a performance classification model, wherein both the performance prediction model and the performance classification model are models established using a machine learning algorithm.

[0080] The potential energy function development module is used to fit the potential energy function of the catalyst system using a neural network algorithm.

[0081] The machine learning modeling module includes a data preprocessing component, a feature engineering component, a test / training set division component, a machine learning algorithm selection component, and a data import component.

[0082] The data preprocessing component is used to preprocess the machine learning data, including data standardization, normalization, logarithmic transformation, category data conversion and data dimension reduction (using principal component analysis technology and linear discriminant analysis), etc. The tools used by the data preprocessing component are the machine learning library scilit-lean and the pandas library in Python.

[0083] The feature engineering component is used to perform feature engineering, feature selection, feature combination, etc., to extract valuable features from the preprocessed machine learning data, and to combine and transform features to improve model performance.

[0084] The test / training set partitioning component is used to partition the data set, and the data ratio used as the training set and the test set can be selected. Data sampling uses two methods: random sampling and stratified sampling.

[0085] The machine learning algorithm component is used to select machine learning algorithms, including random forests, decision trees, extreme gradient boosting trees, light gradient boosting machines, and neural networks. Users can select the listed algorithms for machine learning modeling.

[0086] The data import component is used to import the data required for machine learning modeling. The import methods include: data import based on the database retrieval module and data import from the user's local PC, and machine learning modeling is realized by dragging and dropping to form a flowchart.

[0087] The catalyst database retrieval subsystem, catalyst high-throughput calculation and multi-scale simulation subsystem and catalyst artificial intelligence subsystem in the present embodiment are independent and interconnected, and each submodule has independent functions, and the modules are data-interactive. The database retrieval subsystem can provide the catalyst structure file (catalyst structure data includes structure file and composition, composed of molecular formula or element composition) necessary for computational simulation for the catalyst high-throughput calculation and multi-scale simulation subsystem, and can also provide the catalyst artificial intelligence subsystem with the necessary characteristic attributes required for machine learning training; At the same time, the catalyst structure and attribute data calculated by the catalyst high-throughput calculation and multi-scale simulation subsystem can be stored in the database retrieval subsystem as new catalyst data, and the data calculated by the subsystem can also be imported into the catalyst artificial intelligence subsystem as the necessary features or target values ​​for machine learning training, such as adsorption energy, electronic structure information, etc.

[0088] The data-driven research system for a catalyst provided in this embodiment can realize an integrated research and development model of photo / electric / thermal catalyst data retrieval, structural design, computational simulation and performance prediction, thereby improving the research and development efficiency of catalysts and enabling rapid development of efficient photo / electric / thermal catalysts.

[0089] like Figure 1As shown, a data-driven research system for a catalyst provided in this embodiment can be divided into an interface visualization part, a module service part, and a hardware infrastructure support part, wherein the Web interface visualization is collaboratively implemented by FastAPI, PythonScript, Ts language, and react framework; the module service part can be divided into a database retrieval module (i.e., a catalyst database retrieval subsystem), a high-throughput computing and multi-scale simulation module (i.e., a catalyst high-throughput computing and multi-scale simulation subsystem), and an artificial intelligence module (i.e., a catalyst artificial intelligence subsystem). Among them, the database retrieval module can provide users with data cleaning services, data extraction services, data analysis services (the service uses the Panda library to read the uploaded data, identify and mark problem data, check the integrity and consistency of the data, fill in the missing and modify the wrong data), catalyst data knowledge spectrum (automatically classify the data, including catalyst structure data, physical property data, calculated property data and experimental characterization data), etc. The high-throughput calculation and multi-scale simulation module can provide users with structural high-throughput design services (based on RDkit, ASE, pymatgen libraries in pytho for batch construction of molecular structures and periodic structures), parameter preset services (provide parameter setting models in the calculation software to achieve tasks such as structure optimization, static calculation, transition state calculation and molecular dynamics simulation), batch submission of calculation tasks (using python and shell language in linux system to compile batch calculations) The system can submit scripts in batches to realize the automatic submission of multiple tasks), batch extraction and visualization services of calculation result data (using a combination of Python and shell language to accurately locate target data and extract data to text files), the artificial intelligence module can provide users with data set preparation services (using the data preprocessing component to preprocess data and divide data sets), feature extraction services (using the feature engineering component to extract feature data), model online training services (using machine learning algorithms to select components to perform machine learning regression and classification model search), potential energy function fitting services (using machine learning algorithms to select neural networks in components, reading atomic types, coordinates, energy, forces and other data in the calculation output files, and fitting potential energy curves); the hardware infrastructure is a high-performance server cluster composed of GPU, CPU, memory, solid-state drive, IB network card and other hardware, which provides hardware support for the operation of the system.

[0090] like Figure 2 As shown, users can register and log in on the login page. After passing the verification, they can enter the catalyst data-driven research system. Depending on the permissions of the logged-in user, the user management subsystem will have different functions, such as Figure 3 As shown in the figure, when a normal user logs in, the user management subsystem has the function of modifying personal information, such as Figure 4As shown, when an administrator logs in, the user management subsystem adds a user management function that can realize the functions of adding, deleting, and controlling permissions of users.

[0091] The catalyst data retrieval subsystem is intended to store the models and properties of millions of catalysts for user retrieval. Figure 5 As shown in the figure, the page displays the electrocatalyst search module, photocatalyst search module and thermal catalyst search module. Users can select the catalyst type they need to know to search for data. Click to enter the catalyst search page, such as Figure 6 As shown, users can select elements in the periodic table to search for catalysts. The search methods include fuzzy search based on elements and precise search based on chemical formulas. The chemical periodic table selection component is used to directly select chemical elements as search conditions through the chemical periodic table for search. This interface displays all chemical elements contained in the conventional entire chemical periodic table. You can click to select each chemical element, which is convenient for searching related materials without memorizing chemical elements, and preventing query deviation caused by memory bias.

[0092] like Figure 7 As shown, the preliminary search results will be displayed in the search result list page, which displays the catalyst ID, chemical formula, Smiles code, adsorption energy and other information. Users can further search for catalysts that meet their needs based on the search results and enter the search result details page by clicking, such as Figure 8 As shown, the physical property data, calculated properties, experimental characterization information, references and other information of the catalyst are displayed. The properties of the catalyst material are mainly composed of numbers, text, and pictures. The pictures are pictures of the relevant properties of the material, including but not limited to png, jpg and other picture formats.

[0093] The high-throughput calculation and multi-scale simulation subsystem integrates quantum chemistry and molecular dynamics software such as VASP, GAUSSIAN, and LAMMPS, and embeds a self-compiled high-throughput catalyst structure generation program and structure import tool to initially establish a theoretical calculation scheme for catalyst materials with reliable accuracy and feasible efficiency, realize high-precision calculation of catalyst electronic structure, summarize and improve catalyst material active centers, catalytic mechanisms, reaction energy barriers and design theories, and provide users with accurate and intuitive chemical images. Secondly, improve and apply molecular dynamics simulation technology to study the real solvent environment of the catalyst.

[0094] like Fig. 9As shown, the high-throughput computing and multi-scale simulation module is divided into catalyst structure uploading and computing software selection and parameter configuration. There are two ways to upload the catalyst structure: uploading on the user's local PC and uploading the catalyst online design. After uploading, the computing software selection and parameter configuration module is used to select the computing software and configure the relevant parameters; the computing software selection and parameter configuration module includes the system-integrated VASP, GAUSSIAN, and LAMMPS software.

[0095] The catalyst artificial intelligence subsystem integrates machine learning regression and classification algorithms such as random forest, support vector machine, and gradient boosting, making it easy for users to achieve regression prediction of catalyst material performance and establish classification models. At the same time, it embeds neural network algorithms to facilitate users to develop neural network potential functions for catalyst systems. By encapsulating each algorithm as a function, users only need to drag and drop the encapsulated function modules to form a flow chart to achieve artificial intelligence modeling.

[0096] like Fig.10 As shown, the functions of the Catalyst AI subsystem include data input, data preprocessing, feature engineering, test / training set partitioning, integrated functions (such as regression algorithms), and a user modeling flowchart canvas.

[0097] Compared with the prior art, the data-driven research system for catalysts provided in this embodiment focuses on the catalyst field, integrates catalyst data retrieval, catalyst high-throughput design, high-throughput calculation and multi-scale simulation, performance prediction and classification, and covers the entire process of catalyst research and development; the data generated by users using the research and development of each module will be stored in the database for subsequent catalyst retrieval, greatly improving the utilization of data; and the system uses a graphical operation interface that is easy for human-computer interaction, which greatly reduces the technical threshold for catalyst research and development, shortens the catalyst research and development cycle and cost, and accelerates the catalyst research and development process.

[0098] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0099] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A data-driven research system for catalysts, characterized in that: The catalyst data-driven research system includes: Catalyst database retrieval subsystem, catalyst high-throughput calculation and multi-scale simulation subsystem, and catalyst artificial intelligence subsystem; The catalyst database retrieval subsystem is connected to the catalyst high-throughput calculation and multi-scale simulation subsystem and the catalyst artificial intelligence subsystem; The catalyst high-throughput computing and multi-scale simulation subsystem is connected to the catalyst artificial intelligence subsystem; The catalyst database retrieval subsystem is used to store a number of catalyst information and retrieve target catalyst information according to the retrieval conditions, wherein the catalyst information includes catalyst models and attribute data, and the attribute data includes structural data, physical property data, calculated property data and experimental characterization data; The catalyst high-throughput calculation and multi-scale simulation subsystem is used to generate a catalyst structure according to the target catalyst information, and perform quantum chemical calculation, density functional calculation and molecular dynamics simulation on the generated catalyst structure to obtain catalyst calculation simulation data; The catalyst artificial intelligence subsystem is used to predict and classify the performance of the target catalyst and to fit the potential energy function of the catalyst system based on the target catalyst information and the catalyst computational simulation using a machine learning algorithm, wherein the catalyst system includes a target catalyst, reactants, intermediates, and products.

2. The data-driven research system for catalysts according to claim 1, characterized in that: The catalyst database retrieval subsystem specifically includes: a query display interface component; The query display interface component is used to query and display the target catalyst information according to the search condition; The query display interface component includes a search input box, a search result list and a search result details interface; The search input box is used to input the search condition, wherein the search condition is a chemical element or a chemical formula; The search result list is used to display rough information of target catalysts that meet the search conditions, wherein the rough information of the catalyst includes ID, chemical formula, Smiles code and adsorption energy; The search result details interface is used to display detailed information of the target catalyst that meets the search conditions, wherein the target catalyst detailed information includes the attribute data of the target catalyst.

3. The data-driven research system for catalysts according to claim 2, characterized in that: The catalyst database retrieval subsystem further includes: a periodic table selection component; The periodic table selection component is connected to the search input box; The periodic table selection component is used to select the chemical element according to the periodic table.

4. The data-driven research system for catalysts according to claim 1, characterized in that: The catalyst database retrieval subsystem further includes: a catalyst data knowledge spectrum module; The catalyst data knowledge spectrum module is used to organize and classify the attribute data in the catalyst information to form a catalyst data knowledge spectrum.

5. The data-driven research system for catalysts according to claim 1, characterized in that: The catalyst high-throughput calculation and multi-scale simulation subsystem specifically includes: a catalyst structure input control module and a quantum chemistry and molecular dynamics calculation module; The catalyst structure input control module is connected with the quantum chemistry and molecular dynamics calculation module; The catalyst structure input control module is used to generate a catalyst structure according to the target catalyst information; The quantum chemistry and molecular dynamics calculation module is used to perform quantum chemistry calculation, density functional calculation and molecular dynamics simulation according to the generated catalyst structure to obtain catalyst calculation simulation data.

6. The data-driven research system for catalysts according to claim 5, characterized in that: The catalyst structure input control module specifically includes: a catalyst structure high-throughput generation component, a catalyst structure online design component and a catalyst structure import component; The catalyst structure high-throughput generation component is used to automatically generate a plurality of catalyst structures according to the target catalyst information and set conditions, wherein the catalyst structure includes a molecular structure and a periodic structure; The catalyst structure online design component is used to customize the catalyst structure according to the target catalyst information and user requirements; The catalyst structure import component is used to upload the catalyst structure locally.

7. The data-driven research system for catalysts according to claim 5, characterized in that: The quantum chemistry and molecular dynamics calculation module specifically includes: a calculation parameter setting component, a VASP calculation component, a GAUSSIAN calculation component and a LAMMPS calculation component; The calculation parameter setting component is used to set the calculation parameters according to user needs; The VASP calculation component is used to call the VASP program to perform density functional calculation according to the generated catalyst structure; The GAUSSIAN calculation component is used to call GAUSSIAN to perform quantitative calculation according to the generated catalyst structure; The LAMMPS calculation component is used to call LAMMPS to perform molecular dynamics calculations according to the catalyst structure.

8. The data-driven research system for catalysts according to claim 1, characterized in that: The catalyst artificial intelligence subsystem specifically includes: a machine learning modeling module and a potential energy function development module; The machine learning modeling module is used to predict the performance of the target catalyst according to the performance prediction model, and classify the performance of the target catalyst according to the performance classification model, wherein the performance prediction model and the performance classification model are both models established using a machine learning algorithm; The potential energy function development module is used to fit the potential energy function of the catalyst system using a neural network algorithm.

9. The data-driven research system for catalysts according to claim 8, characterized in that: The machine learning modeling module specifically includes: a data preprocessing component, a feature engineering component, a test / training set division component, a machine learning algorithm selection component, and a data import component; The data preprocessing component is used to preprocess the machine learning data, wherein the machine learning data includes the category, electronegativity, number of valence electrons, electron affinity, atomic radius, first ionization energy, formation enthalpy, atomic mass, electronic structure information of the target catalyst element and related calculation information of the intermediate, the electronic structure information is the data in the catalyst calculation simulation data, and the electronic structure information includes atomic charge, energy band gap and magnetic moment, and the preprocessing includes data standardization, normalization, logarithmic transformation, category data conversion and data dimension reduction; The feature engineering component is used to extract features from the preprocessed machine learning data to obtain feature data; The test / training set division component is used to divide the data set into a training set and a test set, wherein the data set includes the feature data; The machine learning algorithm selection component is used to provide a variety of machine learning algorithms for selection, wherein the machine learning algorithms include random forest, decision tree, XGBoost, LightGBM and neural network; The data import component is used to import the required data of the performance prediction model and the performance classification model.

Images