Method, device and equipment for screening diatomic catalyst for nitrogen reduction reaction
By constructing a catalyst screening model based on key feature descriptors and utilizing machine learning algorithms and density functional theory calculations, the problem of low catalyst screening efficiency was solved, and highly active diatomic catalysts were discovered efficiently.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CNOOC GAS & POWER GRP
- Filing Date
- 2025-12-24
- Publication Date
- 2026-05-05
AI Technical Summary
Existing technologies for nitrogen reduction reaction (NRR) catalyst screening have low efficiency, resulting in wasted computational resources and time costs, and making it difficult to efficiently discover highly active catalysts.
By constructing a catalyst screening model based on key feature descriptors, and using machine learning algorithms and density functional theory calculations, diatomic catalysts with excellent catalytic activity are screened out, reducing unnecessary calculations.
This improved catalyst screening efficiency, reduced computational resources and time consumption, and enabled the efficient discovery of superior catalysts.
Smart Images
Figure CN121983173A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of catalyst technology, and in particular to a method, apparatus and equipment for screening diatomic catalysts for nitrogen reduction reactions. Background Technology
[0002] Ammonia is an important inorganic chemical product, widely used in the production of nitrogen fertilizers, nylon, and amines. Traditional industrial ammonia synthesis suffers from high energy consumption and high carbon emissions, making the search for low-energy, low-carbon alternative ammonia synthesis technologies an urgent need. Nitrogen reduction reaction (NRR) is a potential alternative technology, with readily available nitrogen sources and a relatively green reaction process. Currently, the main bottleneck restricting the application of NRR technology lies in the insufficient activation capacity of catalysts for N2 and low ammonia selectivity. Therefore, developing novel catalysts with high catalytic activity and high ammonia production selectivity is key to promoting the efficient development of NRR technology.
[0003] The concept of atomic-level catalysts has injected new vitality into catalysis research. Diatom catalysts, as a novel type of catalytic material, utilize two metal atoms as active sites for synergistic catalysis. They not only possess similar high atom utilization and excellent selectivity as single-atom catalysts but also overcome the physical limitations of single-atom catalysts, thus improving catalytic efficiency. Therefore, diatomic catalysts show great potential in the electrocatalytic N2 reduction reaction and can serve as a novel, highly efficient electrocatalytic catalyst. However, due to the diversity of active sites in diatomic catalysts, existing high-throughput screening and experimental screening processes face a large amount of inefficient computation, resulting in low catalyst screening efficiency and wasted computational resources and time. Summary of the Invention
[0004] This invention provides a method, apparatus, and equipment for screening diatomic catalysts for nitrogen reduction reactions. It addresses the shortcomings of existing technologies, such as high-throughput screening and experimental screening, which suffer from numerous invalid calculations, low catalyst screening efficiency, and wasted computational resources and time. The invention's solution uses key feature descriptors related to reaction activity and trains a catalyst to obtain a catalyst screening model. This model can efficiently screen catalysts with excellent catalytic activity, avoiding numerous invalid calculations in high-throughput computations, reducing the waste of computational resources and research time, and thus effectively improving catalyst screening efficiency. This provides an innovative technical path for the efficient discovery of catalysts.
[0005] This invention provides a method for screening diatomic catalysts for nitrogen reduction reactions, comprising the following steps.
[0006] A variety of candidate diatomic catalysts for nitrogen reduction reaction were obtained, and all of the candidate diatomic catalysts were divided into a variety of training catalysts and a variety of catalysts to be screened. All the catalysts to be screened are input into the catalyst screening model to obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
[0007] According to the present invention, a method for screening diatomic catalysts for nitrogen reduction reactions is provided, wherein the catalyst screening model is obtained by training based on the following steps: For each of the training catalysts, determine the target feature data corresponding to the training catalyst; The model training dataset is determined based on all the target feature data; The initial machine learning model is iteratively trained based on the model training dataset and the preset machine learning algorithm. When the preset iteration termination condition is met, the catalyst screening model is obtained.
[0008] According to the present invention, a method for screening diatomic catalysts for nitrogen reduction reactions, wherein determining the target feature data corresponding to the training catalyst includes: The training catalyst is modeled to obtain the first structure file corresponding to the training catalyst; The input file corresponding to the first structure file is calculated based on first-principles calculations. Density functional theory calculations are performed on the input file to obtain the corresponding result file. The target feature data corresponding to the training catalyst is determined based on the result file and the key feature descriptor.
[0009] According to the present invention, a method for screening diatomic catalysts for nitrogen reduction reactions, wherein determining the target feature data corresponding to the training catalyst based on the result file and the key feature descriptor includes: Based on the key feature descriptor, first feature data is extracted from the result file, and / or, based on the key feature descriptor, second feature data corresponding to the result file is queried in a preset database; The target feature data corresponding to the training catalyst is determined based on the first feature data and / or the second feature data.
[0010] According to the present invention, a method for screening diatomic catalysts for nitrogen reduction reactions, wherein determining the model training dataset based on all the target feature data includes: Data preprocessing is performed on all the target feature data to obtain model training data corresponding to each target feature data; the data preprocessing includes one or more of the following: handling missing values, removing duplicate data, handling outliers, unifying data format, data validation, feature engineering, normalization, and standardization. The model training dataset is determined based on all the model training data.
[0011] According to the present invention, a method for screening diatomic catalysts for nitrogen reduction reactions, wherein obtaining multiple candidate diatomic catalysts for nitrogen reduction reactions includes: A variety of initial diatomic catalysts were constructed based on preset active sites, preset substrates, preset coordination numbers, and preset geometric configurations. A variety of candidate diatomic catalysts were randomly selected from all the initial diatomic catalysts.
[0012] The present invention also provides a diatomic catalyst screening device for nitrogen reduction reaction, comprising the following modules: The acquisition module is used to acquire a variety of candidate diatomic catalysts for nitrogen reduction reaction, and to classify all the candidate diatomic catalysts into a variety of training catalysts and a variety of catalysts to be screened. The screening module is used to input all the catalysts to be screened into the catalyst screening model and obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the diatomic catalyst screening method for nitrogen reduction reaction as described above.
[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the diatomic catalyst screening method for nitrogen reduction reaction as described above.
[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the diatomic catalyst screening method for nitrogen reduction reaction as described above.
[0016] This invention provides a method, apparatus, and equipment for screening diatomic catalysts for nitrogen reduction reactions. The method acquires multiple candidate diatomic catalysts for nitrogen reduction reactions and categorizes all candidate catalysts into multiple training catalysts and multiple catalysts to be screened. All catalysts to be screened are input into a catalyst screening model, and the catalyst screening results output by the model are obtained. The catalyst screening model is trained based on all training catalysts and key feature descriptors. This invention's technical solution, based on key feature descriptors related to reaction activity and training catalysts, trains a catalyst screening model that can efficiently screen catalysts with excellent catalytic activity. This avoids a large amount of invalid computation in high-throughput calculations, reducing the waste of computational resources and research time, thereby effectively improving catalyst screening efficiency and providing an innovative technical path for efficient catalyst discovery. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0018] Figure 1 This is a schematic flowchart of the diatomic catalyst screening method for nitrogen reduction reaction provided by the present invention.
[0019] Figure 2 This is a schematic diagram of the structure of the diatomic catalyst and the active site elements provided by the present invention.
[0020] Figure 3 This is a schematic diagram of the reaction path for the nitrogen reduction reaction provided by the present invention.
[0021] Figure 4 This is a schematic diagram of the structure of the diatomic catalyst screening device for nitrogen reduction reaction provided by the present invention.
[0022] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation
[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0024] To address the aforementioned problems in the prior art, this invention provides a method for screening diatomic catalysts for nitrogen reduction reactions. Figure 1 This is one of the flowcharts illustrating the method for screening diatomic catalysts for nitrogen reduction reactions provided by this invention, such as... Figure 1 As shown, the method includes the following steps 110 to 120.
[0025] Step 110: Obtain a variety of candidate diatomic catalysts for nitrogen reduction reaction, and classify all the candidate diatomic catalysts into a variety of training catalysts and a variety of catalysts to be screened.
[0026] Specifically, multiple candidate diatomic catalysts for nitrogen reduction reactions can be pre-designed. After obtaining multiple candidate diatomic catalysts, all candidate diatomic catalysts can be divided into two parts: training catalysts and catalysts to be screened.
[0027] In one embodiment, obtaining a variety of candidate diatomic catalysts for the nitrogen reduction reaction includes: A variety of initial diatomic catalysts were constructed based on preset active sites, preset substrates, preset coordination numbers, and preset geometric configurations. A variety of candidate diatomic catalysts were randomly selected from all the initial diatomic catalysts.
[0028] Specifically, the pre-defined active sites should possess good electronegativity and ionization to ensure high reactivity, and should also be economical and environmentally friendly to meet the requirements of sustainable development. The pre-defined substrate should have good conductivity and stability to ensure efficient reaction; simultaneously, high specific surface area and good binding with the active sites can improve catalytic effect, while also emphasizing economy and environmental friendliness to ensure sustainable use of the material. Pre-defined coordination numbers and pre-defined geometries are selected within the coordination environment to enhance reactivity and improve selectivity. It is easy to understand that pre-defined active sites include multiple active sites, pre-defined substrates include multiple substrates, pre-defined coordination numbers include multiple coordination numbers, and pre-defined geometries include multiple geometries.
[0029] Furthermore, preset active sites, preset substrates, preset coordination numbers, and preset geometric configurations can be arbitrarily combined to construct a variety of initial diatomic catalysts. Subsequently, multiple candidate diatomic catalysts can be randomly selected from all initial diatomic catalysts.
[0030] In the above embodiments, constructing an initial diatomic catalyst based on preset active sites, a preset substrate, a preset coordination number, and a preset geometric configuration ensures the catalyst's reactivity, economic and environmental characteristics, good conductivity, and stability, thus guaranteeing the initial diatomic catalyst's good reactivity. Furthermore, randomly selecting candidate diatomic catalysts from the initial diatomic catalyst reduces computational load and improves computational efficiency.
[0031] Step 120: Input all the catalysts to be screened into the catalyst screening model to obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
[0032] Specifically, an initial machine learning model can be trained using all the training catalysts to obtain a catalyst screening model. Then, all the catalysts to be screened can be input into the catalyst screening model to obtain the catalyst screening results output by the model. These results identify the catalysts with the best overall catalytic performance from all the catalysts to be screened.
[0033] In one embodiment, the catalyst screening model is trained based on the following steps: For each of the training catalysts, determine the target feature data corresponding to the training catalyst; The model training dataset is determined based on all the target feature data; The initial machine learning model is iteratively trained based on the model training dataset and the preset machine learning algorithm. When the preset iteration termination condition is met, the catalyst screening model is obtained.
[0034] Specifically, for each training catalyst, the target feature data corresponding to that catalyst can be determined, and then the model training dataset can be determined based on all target feature data. Further, the initial machine learning model can be iteratively trained based on the model training dataset and a preset machine learning algorithm. When a preset iteration termination condition is reached, a catalyst selection model is obtained. The preset iteration termination condition can be, for example, reaching a preset number of iterations; this embodiment of the invention does not specifically limit this. The preset machine learning algorithm can include, for example, random forest regression, K-nearest neighbor regression, and decision tree regression, and can also include other machine learning algorithms; this embodiment of the invention does not specifically limit this. Optionally, any one of the preset machine learning algorithms can be selected to iteratively train the initial machine learning model. Alternatively, a model test dataset can be determined based on the target feature data, and the initial machine learning model can be iteratively trained using multiple preset machine learning algorithms to obtain multiple candidate catalyst selection models. Then, the model test dataset can be input into the multiple candidate catalyst selection models, and the output results of the multiple candidate catalyst selection models can be compared to evaluate the optimal candidate catalyst selection model as the catalyst selection model.
[0035] For example, the root mean square error can be calculated based on the degree of dispersion of the output results. and the coefficient of determination of the model's approximation of the objective function To evaluate the merits of candidate catalyst screening models, and It can be expressed by the following formula: in, Represents the actual value. Indicates the output result. Indicates the number of samples. Indicates the index of the sample. Indicates the output result The average value. The closer the RMSE is to 0, the higher the model accuracy. The closer the model is to 1, the higher its accuracy.
[0036] In the above embodiments, a dataset is constructed based on the training catalyst, and then the initial machine learning model is trained to obtain a catalyst screening model, which lays the foundation for the screening of efficient diatomic catalysts.
[0037] In one embodiment, determining the target feature data corresponding to the training catalyst includes: The training catalyst is modeled to obtain the first structure file corresponding to the training catalyst; The input file corresponding to the first structure file is calculated based on first-principles calculations. Density functional theory calculations are performed on the input file to obtain the corresponding result file. The target feature data corresponding to the training catalyst is determined based on the result file and the key feature descriptor.
[0038] Specifically, a modeling tool can be used to model the training catalyst, obtaining a first structure file corresponding to the training catalyst. This first structure file can contain the coordinate information of all atoms and lattice parameters, and can be in POSCAR format, for example. Then, the input file corresponding to the first structure file can be calculated based on first-principles calculations. This input file can include, for example, calculation parameters, k-point grid, pseudopotential file, crystal structure, and other possible input files. The calculation parameters can further include calculation type, calculation accuracy, self-consistent iterative convergence criteria, etc.
[0039] Furthermore, density functional theory (DFT) calculations can be performed on the input file to obtain a corresponding result file. This result file can include the calculation results, optimized crystal structure, density of states, and band structure. During the DFT calculation process, parallel computing can be designed based on available computing resources and task requirements to fully utilize computing resources and improve computational efficiency. The calculation process can also monitor the execution status and progress of the computation task, including checking task status, viewing computation log files, and monitoring the load of computation nodes, to ensure smooth execution of the computation task and to promptly identify and resolve any problems that may arise during the calculation.
[0040] After obtaining the results file, the target feature data corresponding to the training catalyst can be determined based on the results file and key feature descriptors. These key feature descriptors can be pre-determined based on stability, adsorption, activation capacity, and reaction mechanism. Reasonable key feature descriptors are a crucial foundation for ensuring the accuracy of catalyst screening. The structural stability of a catalyst is a prerequisite for its actual existence and catalytic activity. Therefore, it is necessary to analyze the performance of this type of catalyst in the nitrogen reduction reaction from multiple perspectives, mainly including catalyst stability, adsorption, activation capacity, and reaction mechanism. These factors are essential for evaluating the effectiveness of the catalyst in actual reactions. By comprehensively considering these key factors, potential catalysts can be screened more accurately. Based on literature review and related research, constraints on the key feature descriptors can be determined. These constraints provide a theoretical basis for catalyst screening and help limit the range of candidate catalysts, thereby ensuring the accuracy and effectiveness of the screening process.
[0041] In the above embodiments, key feature descriptors help limit the range of candidate catalysts, thereby ensuring the accuracy and effectiveness of the screening process. The combination of the results file and key feature descriptors ensures the accuracy of training catalyst feature extraction.
[0042] In one embodiment, determining the target feature data corresponding to the training catalyst based on the result file and the key feature descriptor includes: Based on the key feature descriptor, first feature data is extracted from the result file, and / or, based on the key feature descriptor, second feature data corresponding to the result file is queried in a preset database; The target feature data corresponding to the training catalyst is determined based on the first feature data and / or the second feature data.
[0043] Specifically, first feature data can be extracted from the result file based on key feature descriptors, and / or, second feature data corresponding to the result file can be queried from a preset database based on the key feature descriptors. For example, the key feature descriptors include formation energy and adsorption energy. Formation energy and adsorption energy can be directly extracted from the result file as first feature data, or they can be queried from a preset database as second feature data. Alternatively, formation energy can be extracted from the result file as first feature data, and adsorption energy can be queried from a preset database as second feature data.
[0044] Furthermore, the target feature data corresponding to the training catalyst can be determined based on the first feature data and / or the second feature data.
[0045] The preset database is used to query the characteristic data of the catalyst. It may include the Materials Project (MP) database, as well as other academic databases. In addition, the preset database may also include various academic literature, such as scientific journals and conference papers. This embodiment of the invention does not make specific limitations here.
[0046] In the above embodiments, feature data is extracted based on key feature descriptors, which allows catalysts to be screened according to requirements, further improving the catalyst screening efficiency.
[0047] In one embodiment, determining the model training dataset based on all the target feature data includes: Data preprocessing is performed on all the target feature data to obtain model training data corresponding to each target feature data; the data preprocessing includes one or more of the following: handling missing values, removing duplicate data, handling outliers, unifying data format, data validation, feature engineering, normalization, and standardization. The model training dataset is determined based on all the model training data.
[0048] Specifically, data preprocessing can be performed on all target feature data to obtain model training data corresponding to each target feature data. Data preprocessing includes one or more of the following processes: (1) Handling missing values: Check whether there are missing values in all target feature data and process them as needed; (2) Removing duplicate data: Check whether there are duplicate samples or duplicate features in all target feature data, merge or delete duplicate data to reduce the redundancy of the dataset; (3) Handling outliers: Use statistical methods and visualization methods to identify and process outliers; (4) Unifying data format: Unify the format and unit of all target feature data to ensure data consistency and comparability; (5) Data validation: Validate the cleaned data to ensure the accuracy and integrity of the data. Statistical methods and visualization methods can be used to validate the distribution and characteristics of the data; (6) Feature engineering processing; (7) Normalization processing; (8) Standardization processing.
[0049] Furthermore, the model training dataset can be determined based on all model training data.
[0050] Optionally, the model training dataset and the model test dataset can be determined based on all model training data.
[0051] In the above embodiments, data preprocessing of the target feature data can reduce data anomalies and errors in the target feature data, and optimization of the training data improves the model accuracy, thereby improving the accuracy of catalyst screening.
[0052] This invention provides a method for screening diatomic catalysts for nitrogen reduction reactions. The method obtains multiple candidate diatomic catalysts for nitrogen reduction reactions and categorizes all candidate catalysts into multiple training catalysts and multiple catalysts to be screened. All catalysts to be screened are input into a catalyst screening model, and the model outputs the catalyst screening results. The catalyst screening model is trained based on all training catalysts and key feature descriptors. This invention's technical solution, based on key feature descriptors related to reaction activity and training catalysts, trains a catalyst screening model that can efficiently screen catalysts with excellent catalytic activity. This avoids a large amount of invalid computation in high-throughput calculations, reducing the waste of computational resources and research time, thereby effectively improving catalyst screening efficiency and providing an innovative technical path for efficient catalyst discovery.
[0053] By way of example, the present application is illustrated below through a complete embodiment: Studies have shown that nitrogen-doped graphite (NG) exhibits excellent stability and tunable active site coordination structures as catalyst substrates. In particular, the trinitrogen coordination structure demonstrates good catalytic performance and electron transfer characteristics for diatomic catalysts. Therefore, the experiment selected a microstructure where a single metal atom is coordinated by three nitrogen atoms (N6G). Figure 2 This is a schematic diagram of the structure and active site elements of the diatomic catalyst provided by this invention. Fourteen transition metal atoms (Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Zr, Nb, Mo, Ru, Pd) were selected as active sites to construct the TM1-TM2@N6G model for catalytic performance studies. Figure 2 As shown. Given the structural diversity and complexity of TM1-TM2@N6G, the respective combinations of these 14 metal atoms yielded a total of 91 heteronuclear diatomic catalysts and 14 homonuclear catalysts, resulting in a total of 105 initial diatomic catalysts. From these, 21 catalyst groups were randomly selected as candidate diatomic catalysts.
[0054] Furthermore, the structural stability of the catalyst is a prerequisite for its effective existence; therefore, we will consider the formation energy of the catalyst. As the first key feature descriptor, the chemisorption of N2 by the catalyst is crucial in the NRR reaction. Effective adsorption can activate the N≡N triple bond, lower the activation energy, reduce the reaction barrier, and increase the effective collision frequency between the reactants and the catalyst. Therefore, we consider the adsorption energy of N2 as... As the second key feature descriptor.
[0055] Figure 3 This is a schematic diagram of the reaction path for the nitrogen reduction reaction provided by the present invention. Figure 3 This demonstrates three main reaction pathways in the NRR process: the distal pathway, the alternating pathway, and the enzymatic pathway. In all reaction mechanisms, the first step involves hydrogenation (…). N2+H= N2H) and the final step of hydrogenation desorption ( The combination of NH₂ and H₂ (NH₃) always exists, in which... This symbol represents a free radical and exists as a reaction intermediate. Due to the inertness of the N≡N triple bond, the first step of hydrogenation requires a relatively large activation energy. Theoretically, this is the rate-limiting step in NRR, therefore the activation energy is used as the third key characteristic descriptor to reflect the ease or difficulty of the reaction. The final step, hydrogenation desorption, is also crucial to the reaction. NH2— The formation and desorption of NH3 on diatomic catalysts require a relatively large ascent free energy. Therefore, we selected NH2 to... Energy changes of NH3 As the fourth key feature descriptor. Studies have shown that in the NRR reaction, and These values are typically large and positive; therefore, using these two free energy changes as activity descriptors helps reflect catalyst performance. Through these four key characteristic descriptors, screening efficiency can be significantly improved without reducing screening precision. Furthermore, for the final electrocatalytic hydrogen evolution reaction (HER), we select catalysts with N2 adsorption capacity greater than their electrocatalytic hydrogen evolution capacity. As the fifth key feature descriptor, and The fifth key feature descriptor is used to remove catalysts with high HER activity.
[0056] Based on the above analysis, the following screening steps can be designed: First, […]. and As the first screening criterion, catalysts with low stability and low activation degree to N2 were excluded; the second step was to... and The third step involves screening for highly active catalysts. Screening for low HER hydrogen evolution catalysts. In order to select the best-performing diatomic catalysts, the standard settings for each key feature descriptor are as follows: (1) It should be less than -0.5 electron volts (eV) to ensure better stability; (2) It should be less than -1 eV, corresponding to the degree of chemisorption and activation of N2; (3) It should not be greater than 0 eV. For NRR, the breaking of the N≡N triple bond will be a spontaneous process, which greatly reduces the difficulty of the reaction; (4) The value should not exceed 0.70 eV, which, according to the Arrhenius equation, indicates that the generated NH3 can be rapidly removed. It should be noted that the types and number of key feature descriptors can be adjusted according to research needs.
[0057] Eight different physicochemical properties can be selected to depict the key intermediate reactions in the NRR catalysis process of TM1-TM2@N6G. The electronic structure of the TM metal atoms is characterized by parameters such as metal electronegativity (p), electron affinity (EA), first ionization energy (IE), and d electron number (Nd). Furthermore, considering the unique structure of TM1-TM2@N6G and the interaction between TM atoms and N6G, we selected the van der Waals radius (R) of the TM atom, the atomic mass (M) of the TM atom, and the sum of the atomic radii of the two TM atoms, ultimately combining 23 input features as target feature data.
[0058] Initial machine learning models were built using three machine learning regression algorithms: random forest regression, K-nearest neighbor regression, and decision tree regression. The generalization performance of the different initial models was cross-validated using a 4-fold cross-validation method. Each model achieved its optimal performance after extensive parameter tuning. , The model was evaluated, and to prevent overfitting during tuning, performance evaluations were added for both the training and testing datasets. Ultimately, the model trained using the random forest regression algorithm achieved the best accuracy.
[0059] Finally, the catalysts to be screened were input into the catalyst screening model to obtain the catalyst screening results. The results showed that four catalysts met the screening criteria.
[0060] The following describes the diatomic catalyst screening device for nitrogen reduction reaction provided by the present invention. The diatomic catalyst screening device for nitrogen reduction reaction described below can be referred to in correspondence with the diatomic catalyst screening method for nitrogen reduction reaction described above.
[0061] Figure 4 This is a schematic diagram of the structure of the diatomic catalyst screening device for nitrogen reduction reaction provided by the present invention, as shown below. Figure 4 As shown, the diatomic catalyst screening device 400 for nitrogen reduction reaction includes the following modules: The acquisition module 410 is used to acquire a variety of candidate diatomic catalysts for nitrogen reduction reaction, and to classify all the candidate diatomic catalysts into a variety of training catalysts and a variety of catalysts to be screened. The screening module 420 is used to input all the catalysts to be screened into the catalyst screening model and obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
[0062] In one embodiment, the diatomic catalyst screening device 400 for nitrogen reduction reaction further includes a model training module, which is specifically used for: For each of the training catalysts, determine the target feature data corresponding to the training catalyst; The model training dataset is determined based on all the target feature data; The initial machine learning model is iteratively trained based on the model training dataset and the preset machine learning algorithm. When the preset iteration termination condition is met, the catalyst screening model is obtained.
[0063] In one embodiment, the model training module is further used for: The training catalyst is modeled to obtain the first structure file corresponding to the training catalyst; The input file corresponding to the first structure file is calculated based on first-principles calculations. Density functional theory calculations are performed on the input file to obtain the corresponding result file. The target feature data corresponding to the training catalyst is determined based on the result file and the key feature descriptor.
[0064] In one embodiment, the model training module is further used for: Based on the key feature descriptor, first feature data is extracted from the result file, and / or, based on the key feature descriptor, second feature data corresponding to the result file is queried in a preset database; The target feature data corresponding to the training catalyst is determined based on the first feature data and / or the second feature data.
[0065] In one embodiment, the model training module is further used for: Data preprocessing is performed on all the target feature data to obtain model training data corresponding to each target feature data; the data preprocessing includes one or more of the following: handling missing values, removing duplicate data, handling outliers, unifying data format, data validation, feature engineering, normalization, and standardization. The model training dataset is determined based on all the model training data.
[0066] In one embodiment, the acquisition module 410 is specifically used for: A variety of initial diatomic catalysts were constructed based on preset active sites, preset substrates, preset coordination numbers, and preset geometric configurations. A variety of candidate diatomic catalysts were randomly selected from all the initial diatomic catalysts.
[0067] This invention provides a diatomic catalyst screening device for nitrogen reduction reactions. It acquires multiple candidate diatomic catalysts for nitrogen reduction reactions and categorizes all candidate catalysts into multiple training catalysts and multiple catalysts to be screened. All catalysts to be screened are input into a catalyst screening model, which outputs the catalyst screening results. The catalyst screening model is trained based on all training catalysts and key feature descriptors. This invention's technical solution, based on key feature descriptors related to reaction activity and training catalysts, trains a catalyst screening model that can efficiently screen catalysts with excellent catalytic activity. This avoids a large amount of invalid computation in high-throughput calculations, reducing the waste of computational resources and research time, thereby effectively improving catalyst screening efficiency and providing an innovative technical path for efficient catalyst discovery.
[0068] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 can call logical instructions in the memory 530 to execute a diatomic catalyst screening method for nitrogen reduction reaction, the method including: A variety of candidate diatomic catalysts for nitrogen reduction reaction were obtained, and all of the candidate diatomic catalysts were divided into a variety of training catalysts and a variety of catalysts to be screened. All the catalysts to be screened are input into the catalyst screening model to obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
[0069] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part 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 the present invention. 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.
[0070] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is capable of executing the diatomic catalyst screening method for nitrogen reduction reaction provided by the above methods, the method comprising: A variety of candidate diatomic catalysts for nitrogen reduction reaction were obtained, and all of the candidate diatomic catalysts were divided into a variety of training catalysts and a variety of catalysts to be screened. All the catalysts to be screened are input into the catalyst screening model to obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
[0071] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the diatomic catalyst screening method for nitrogen reduction reaction provided by the methods described above, the method comprising: A variety of candidate diatomic catalysts for nitrogen reduction reaction were obtained, and all of the candidate diatomic catalysts were divided into a variety of training catalysts and a variety of catalysts to be screened. All the catalysts to be screened are input into the catalyst screening model to obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
[0072] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0073] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for screening diatomic catalysts for nitrogen reduction reactions, characterized in that, include: A variety of candidate diatomic catalysts for nitrogen reduction reaction were obtained, and all of the candidate diatomic catalysts were divided into a variety of training catalysts and a variety of catalysts to be screened. All the catalysts to be screened are input into the catalyst screening model to obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
2. The method for screening diatomic catalysts for nitrogen reduction reactions according to claim 1, characterized in that, The catalyst screening model was trained based on the following steps: For each of the training catalysts, determine the target feature data corresponding to the training catalyst; The model training dataset is determined based on all the target feature data; The initial machine learning model is iteratively trained based on the model training dataset and the preset machine learning algorithm. When the preset iteration termination condition is met, the catalyst screening model is obtained.
3. The method for screening diatomic catalysts for nitrogen reduction reactions according to claim 2, characterized in that, Determining the target feature data corresponding to the training catalyst includes: The training catalyst is modeled to obtain the first structure file corresponding to the training catalyst; The input file corresponding to the first structure file is calculated based on first-principles calculations. Density functional theory calculations are performed on the input file to obtain the corresponding result file. The target feature data corresponding to the training catalyst is determined based on the result file and the key feature descriptor.
4. The method for screening diatomic catalysts for nitrogen reduction reactions according to claim 3, characterized in that, The step of determining the target feature data corresponding to the training catalyst based on the result file and the key feature descriptor includes: Based on the key feature descriptor, first feature data is extracted from the result file, and / or, based on the key feature descriptor, second feature data corresponding to the result file is queried in a preset database; The target feature data corresponding to the training catalyst is determined based on the first feature data and / or the second feature data.
5. The method for screening diatomic catalysts for nitrogen reduction reactions according to claim 2, characterized in that, The process of determining the model training dataset based on all the target feature data includes: Data preprocessing is performed on all the target feature data to obtain model training data corresponding to each target feature data; the data preprocessing includes one or more of the following: handling missing values, removing duplicate data, handling outliers, unifying data format, data validation, feature engineering, normalization, and standardization. The model training dataset is determined based on all the model training data.
6. The method for screening diatomic catalysts for nitrogen reduction reactions according to any one of claims 1 to 5, characterized in that, The acquisition of multiple candidate diatomic catalysts for nitrogen reduction reactions includes: A variety of initial diatomic catalysts were constructed based on preset active sites, preset substrates, preset coordination numbers, and preset geometric configurations. A variety of candidate diatomic catalysts were randomly selected from all the initial diatomic catalysts.
7. A diatomic catalyst screening device for nitrogen reduction reaction, characterized in that, include: The acquisition module is used to acquire a variety of candidate diatomic catalysts for nitrogen reduction reaction, and to classify all the candidate diatomic catalysts into a variety of training catalysts and a variety of catalysts to be screened. The screening module is used to input all the catalysts to be screened into the catalyst screening model and obtain the catalyst screening results output by the catalyst screening model; the catalyst screening model is trained based on all the training catalysts and key feature descriptors.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the diatomic catalyst screening method for nitrogen reduction reaction as described in any one of claims 1 to 6.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the diatomic catalyst screening method for nitrogen reduction reaction as described in any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the diatomic catalyst screening method for nitrogen reduction reaction as described in any one of claims 1 to 6.