CN113990415A - Three-dimensional catalytic material catalytic characteristic screening system based on neural network - Google Patents
Three-dimensional catalytic material catalytic characteristic screening system based on neural network Download PDFInfo
- Publication number
- CN113990415A CN113990415A CN202111637075.7A CN202111637075A CN113990415A CN 113990415 A CN113990415 A CN 113990415A CN 202111637075 A CN202111637075 A CN 202111637075A CN 113990415 A CN113990415 A CN 113990415A
- Authority
- CN
- China
- Prior art keywords
- module
- catalytic
- catalytic material
- catalyst material
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000000463 material Substances 0.000 title claims abstract description 98
- 230000003197 catalytic effect Effects 0.000 title claims abstract description 71
- 238000012216 screening Methods 0.000 title claims abstract description 26
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 21
- 239000003054 catalyst Substances 0.000 claims abstract description 35
- 238000005070 sampling Methods 0.000 claims abstract description 23
- 239000013598 vector Substances 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 claims 2
- 238000006555 catalytic reaction Methods 0.000 claims 1
- 230000004927 fusion Effects 0.000 claims 1
- 238000003775 Density Functional Theory Methods 0.000 abstract description 3
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000000034 method Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C60/00—Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computing Systems (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Catalysts (AREA)
Abstract
The invention discloses a three-dimensional catalytic material catalytic characteristic screening system based on a neural network, which comprises a catalyst material shallow characteristic sampling module, a catalyst material structure information embedding module, a catalytic material property learning module, a catalytic material hidden layer information representing module and a catalytic material multi-task screening module, wherein the catalyst material shallow characteristic sampling module comprises central atom projection and adjacent structure sampling, and the catalytic material property learning module comprises a N-layer Graph transform neural network learning catalytic material structure information, and the system has the beneficial effects that: the time spent for molecular property screening by the present system is significantly reduced, and the computational cost spent is significantly reduced, compared to predicting molecular properties by traditional DFT methods.
Description
Technical Field
The invention relates to the technical field of material property detection, in particular to a three-dimensional catalytic material catalytic characteristic screening system based on a neural network.
Background
The material science is the premise of modern industry, and the catalyst is an important field of the material science and is a hot point of research. Traditional catalyst screening methods are time consuming and expensive. Catalytic material processes such as those based on the DFT process often require months to calculate material properties. With the increasing attention and great success of the data-driven and Artificial Intelligence (AI) combined method in various applications, the system provides a three-dimensional chemical material screening system based on a neural network based on a fourth exemplary form of the innovation.
Disclosure of Invention
The invention aims to provide a three-dimensional catalytic material catalytic characteristic screening system based on a neural network, so as to solve the problems in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme: a three-dimensional catalytic material catalytic characteristic screening system based on a neural network comprises a catalyst material shallow characteristic sampling module, a catalyst material structure information embedding module, a catalytic material property learning module, a catalytic material hidden layer information representing module and a catalytic material multitask screening module, wherein the catalyst material shallow characteristic sampling module comprises a central atom projection module and an adjacent structure sampling module, the catalyst material structure information embedding module comprises a basic structure information splicing module, an atom basic property embedding module, a molecule global structure embedding module and a relative structure embedding module, the catalytic material property learning module comprises a model transform neural network learning catalytic material structure information through N layers, the catalytic material hidden layer information representing module decodes hidden layer catalytic material property information output by the catalytic material property learning module, and the decoding information is fused and represented, and the catalytic material multitask screening module represents the hidden layer property of the catalytic material output by the catalytic material hidden layer information representation module and fuses basic shallow layer structure information.
As a further scheme of the invention: the catalyst material shallow characteristic sampling module is responsible for mapping the molecular three-dimensional structure chart to a two-dimensional structure chart, sequentially sampling each atom as a center and atoms directly connected with the atom, and mapping and sampling the atoms into a plurality of substructures with the same number of parts as the number of the atoms.
As a further scheme of the invention: the catalyst material structure information embedding module is responsible for preprocessing original data into data which can be recognized by a computer, inputting the substructure processed by the catalyst material shallow characteristic sampling module, and splitting and splicing the substructure by the basic structure information splicing module to obtain SEG identification as interval identification.
As a further scheme of the invention: the catalytic material property learning module is used for carrying out neural network training on the data output by the catalyst material structure information embedding module and outputting a corresponding number of vectors to the catalytic material hidden layer information representation module.
As a further scheme of the invention: the catalytic material multitask screening module can match different system tasks by replacing different training targets.
Compared with the prior art, the invention has the beneficial effects that: the time taken for molecular property detection by the present system is significantly increased compared to the prediction of molecular properties by traditional DFT methods.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of the present invention.
In the figure: 1. a catalyst material shallow characteristic sampling module; 2. a catalyst material structure information embedding module; 3. a catalytic material property learning module; 4. a catalytic material hidden layer information representation module; 5. a catalytic material multitask screening module; 6. a central atom projection module; 7. an adjacent structure sampling module; 8. an infrastructure information stitching module; 9. an atomic basis property embedding module; 10. a molecular global structure embedding module; 11. the counter structure is embedded in the module.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, in an embodiment of the present invention, a three-dimensional catalytic material catalytic characteristic screening system based on a neural network includes a catalyst material shallow characteristic sampling module 1, a catalyst material structure information embedding module 2, a catalytic material property learning module 3, a catalytic material hidden layer information representing module 4, and a catalytic material multitask screening module 5, where the catalyst material shallow characteristic sampling module 1 includes a central atom projection module 6 and an adjacent structure sampling module 7, the catalyst material structure information embedding module 2 includes a basic structure information splicing module 8, an atom basic property embedding module 9, a molecule global structure embedding module 10, and a relative structure embedding module 11, the catalytic material property learning module 3 includes a Graph Transformer neural network learning catalytic material structure information through N layers, and the catalytic material hidden layer information representing module 4 learns catalytic material property information output by the catalytic material property learning module 3 Decoding, and fusion-representing the decoding information, wherein the catalytic material multitask screening module 5 represents the properties of the hidden layer of the catalytic material output by the catalytic material hidden layer information representation module 4 and fuses basic multi-layer structure information.
Example (b):
projecting molecules into an undirected graph by using a catalyst material shallow characteristic sampling module 1, wherein atoms contained in the undirected graph are graph nodes, molecular bonds are graph edges, and after the projection is finished, the molecules split molecular structures by using each contained atom as a center, and each substructure comprises a central atom, a plurality of directly connected molecular bonds and adjacent atoms;
in the catalyst material structure information embedding module 2, atoms in a split substructure are converted into 64-dimensional vectors and represented by I, one substructure is converted into a central atom vector and a plurality of adjacent atom vectors, then the Absolute Position information of each atom in a molecular graph is calculated based on an original molecular graph WL-Absolute Position algorithm and is coded into a 64-dimensional vector II, then the direct pair bond energy of the central atom and the adjacent atoms is coded into a 64-dimensional vector III, and finally the three vectors are added to an embedding information representation matrix of a material molecular structure and are integrally transmitted to the catalyst material property learning module 3;
the catalytic material property learning module 3 learns the structural information of the catalytic material through a Graph transform neural network of an N layer, excavates the high-dimensional relation between the molecular structure and the attribute and transmits the high-dimensional relation to the catalytic material hidden layer information representation module 4 in a hidden layer expression mode;
the catalytic material hidden layer information representation module 4 decodes the information transmitted by the catalytic material property learning module 3, fuses basic layer structure information, processes the information through a residual error network, and finally transmits the information to the catalytic material multitask screening module 5, the catalytic material multitask screening module 5 outputs a result of network prediction through a neural network, and different activation functions can be changed to match multiple tasks: catalytic performance prediction, molecular representation recovery.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.
Claims (5)
1. The utility model provides a three-dimensional catalytic material catalysis characteristic screening system based on neural network, includes catalyst material shallow layer characteristic sampling module (1), catalyst material structure information embedding module (2), catalytic material nature learning module (3), catalytic material hidden layer information representation module (4) and catalytic material multitask screening module (5), its characterized in that: the catalyst material shallow layer characteristic sampling module (1) comprises a central atom projection module (6) and an adjacent structure sampling module (7), the catalyst material structure information embedding module (2) comprises a basic structure information splicing module (8), an atom basic attribute embedding module (9), a molecule global structure embedding module (10) and a relative structure embedding module (11), the catalyst material property learning module (3) learns the catalyst material structure information through a Graph transform neural network of N layers, and the catalyst material hidden layer information representation module (4) decodes hidden layer catalyst material property information output by the catalyst material property learning module (3) and performs fusion representation on the decoded information.
2. The neural network-based three-dimensional catalytic material catalytic property screening system of claim 1, wherein: the catalyst material shallow characteristic sampling module (1) is responsible for mapping a molecular three-dimensional structure diagram to a two-dimensional structure diagram, sampling each atom as a center and atoms directly connected with the atom in sequence and mapping the sampling into a plurality of substructures with the same number of parts as the number of the atoms.
3. The neural network-based three-dimensional catalytic material catalytic property screening system of claim 1, wherein: the catalyst material structure information embedding module (2) is responsible for preprocessing original data into data which can be recognized by a computer, inputting the substructure processed by the catalyst material shallow characteristic sampling module (1), and splitting and splicing the substructure by the basic structure information splicing module (8) to obtain SEG marks serving as interval marks.
4. The neural network-based three-dimensional catalytic material catalytic property screening system of claim 1, wherein: the catalytic material property learning module (3) is used for carrying out neural network training on the data output by the catalyst material structure information embedding module (2) and outputting vectors with corresponding quantity to the catalytic material hidden layer information representation module (4).
5. The neural network-based three-dimensional catalytic material catalytic property screening system of claim 1, wherein: the catalytic material multitask screening module (5) can be matched with different system tasks by changing different training targets.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111637075.7A CN113990415A (en) | 2021-12-30 | 2021-12-30 | Three-dimensional catalytic material catalytic characteristic screening system based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111637075.7A CN113990415A (en) | 2021-12-30 | 2021-12-30 | Three-dimensional catalytic material catalytic characteristic screening system based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113990415A true CN113990415A (en) | 2022-01-28 |
Family
ID=79734920
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111637075.7A Pending CN113990415A (en) | 2021-12-30 | 2021-12-30 | Three-dimensional catalytic material catalytic characteristic screening system based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113990415A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108079934A (en) * | 2017-11-30 | 2018-05-29 | 山东大学 | A kind of composite material and preparation method thereof |
| CN111177915A (en) * | 2019-12-25 | 2020-05-19 | 北京化工大学 | High-throughput calculation method and system for catalytic material |
| CN113241128A (en) * | 2021-04-29 | 2021-08-10 | 天津大学 | Molecular property prediction method based on molecular space position coding attention neural network model |
| CN113299354A (en) * | 2021-05-14 | 2021-08-24 | 中山大学 | Small molecule representation learning method based on Transformer and enhanced interactive MPNN neural network |
-
2021
- 2021-12-30 CN CN202111637075.7A patent/CN113990415A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108079934A (en) * | 2017-11-30 | 2018-05-29 | 山东大学 | A kind of composite material and preparation method thereof |
| CN111177915A (en) * | 2019-12-25 | 2020-05-19 | 北京化工大学 | High-throughput calculation method and system for catalytic material |
| CN113241128A (en) * | 2021-04-29 | 2021-08-10 | 天津大学 | Molecular property prediction method based on molecular space position coding attention neural network model |
| CN113299354A (en) * | 2021-05-14 | 2021-08-24 | 中山大学 | Small molecule representation learning method based on Transformer and enhanced interactive MPNN neural network |
Non-Patent Citations (2)
| Title |
|---|
| 卢新宇: "基于语义表示的催化剂材料筛选机制研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
| 秦琦枫等: "深度神经网络在化学中的应用研究", 《江西化工》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110188167B (en) | End-to-end dialogue method and system integrating external knowledge | |
| CN111126218A (en) | Human behavior recognition method based on zero sample learning | |
| CN111523420A (en) | Header classification and header list semantic identification method based on multitask deep neural network | |
| Shi et al. | Network embedding via community based variational autoencoder | |
| Fang et al. | Collaborative filtering recommendation algorithm based on deep neural network fusion | |
| Khalyasmaa et al. | Review of the digital twin technology applications for electrical equipment lifecycle management | |
| CN112397155A (en) | Single-step reverse synthesis method and system | |
| CN112836278A (en) | Intelligent factory multi-source data fusion method and device based on BIM | |
| CN116629977A (en) | Electronic commerce decision method and system based on big data | |
| Liu et al. | The smart architect: Scalable ontology-based modeling of ancient Chinese architectures | |
| CN116433223A (en) | Substation equipment fault early warning method and equipment based on double-domain sparse transducer model | |
| CN113990415A (en) | Three-dimensional catalytic material catalytic characteristic screening system based on neural network | |
| CN103646194A (en) | Credible sorting method of synchronous data flow procedures based on formal verification | |
| CN111159424B (en) | Method and device for labeling knowledge graph entity, storage medium and electronic equipment | |
| CN116108143A (en) | Digital economic monitoring method and system based on block chain technology | |
| CN112612884A (en) | Entity label automatic labeling method based on public text | |
| CN108830030B (en) | Atom type definition system and atom type matching method thereof | |
| Kovalyov | Design and development of a power system digital twin: A model-based approach | |
| CN116129251A (en) | Intelligent manufacturing method and system for office desk and chair | |
| CN116469171A (en) | Self-supervision skeleton action recognition method based on cross-view consistency mining | |
| CN116403278A (en) | Human body action recognition method based on text supervision | |
| CN112991744B (en) | Automatic driving decision-making method and system suitable for long-distance urban road | |
| Ren et al. | Knowledge base enabled semantic communication: A generative perspective | |
| Guo et al. | 3d digital twin of intelligent transportation system based on road-side sensing | |
| CN116049427A (en) | Breadth-first search-based collaborative editing method for small-sample knowledge graph |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220128 |