CN112182965B - Method for designing infrared phase change material based on deep learning - Google Patents
Method for designing infrared phase change material based on deep learning Download PDFInfo
- Publication number
- CN112182965B CN112182965B CN202011022757.2A CN202011022757A CN112182965B CN 112182965 B CN112182965 B CN 112182965B CN 202011022757 A CN202011022757 A CN 202011022757A CN 112182965 B CN112182965 B CN 112182965B
- Authority
- CN
- China
- Prior art keywords
- layer
- network
- layers
- gete
- determining
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 24
- 239000012782 phase change material Substances 0.000 title claims abstract description 19
- 238000013135 deep learning Methods 0.000 title claims abstract description 15
- 239000000463 material Substances 0.000 claims abstract description 53
- 238000013461 design Methods 0.000 claims abstract description 29
- 238000001228 spectrum Methods 0.000 claims abstract description 22
- 230000004044 response Effects 0.000 claims abstract description 17
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 16
- 238000012549 training Methods 0.000 claims abstract description 14
- 238000004519 manufacturing process Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims abstract description 4
- 238000012546 transfer Methods 0.000 claims abstract description 4
- 229910005900 GeTe Inorganic materials 0.000 claims description 26
- 239000002184 metal Substances 0.000 claims description 12
- 229910052751 metal Inorganic materials 0.000 claims description 12
- 239000008188 pellet Substances 0.000 claims description 4
- 229910052737 gold Inorganic materials 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- 229910052709 silver Inorganic materials 0.000 claims description 2
- 230000003595 spectral effect Effects 0.000 abstract description 11
- 230000008569 process Effects 0.000 abstract description 8
- 238000004364 calculation method Methods 0.000 abstract description 5
- 230000002068 genetic effect Effects 0.000 abstract description 5
- 230000007547 defect Effects 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 5
- 238000005457 optimization Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 239000002086 nanomaterial Substances 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000005094 computer simulation Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 210000000349 chromosome Anatomy 0.000 description 1
- 239000011258 core-shell material Substances 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000008303 genetic mechanism Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Geometry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Aiming, Guidance, Guns With A Light Source, Armor, Camouflage, And Targets (AREA)
- Photometry And Measurement Of Optical Pulse Characteristics (AREA)
Abstract
The invention discloses a method for designing an infrared phase change material based on deep learning, which comprises the following steps: step a, determining the types of inner layer materials and outer layer materials, and determining the number of layers and thickness range of the inner layer and the outer layer; step b, respectively manufacturing data sets corresponding to different material combinations by using a Mie scattering transfer matrix algorithm, wherein the thickness of each layer of each material combination is randomly distributed; step c, determining DNN network super parameters, and performing DNN network training on the data set based on the GPU; step d, determining a proper material for the next design according to the response characteristics of the spectrum in the classification result; step e, training and optimizing a sub-network corresponding to the proper material, and designing structural parameters according to the sub-network; and f, obtaining specific parameters of materials and structures. The method for designing the infrared phase change material based on the deep learning provided by the invention can be used for quickly designing a structure according to the required spectral response at any time, and overcomes the defect of redundancy in the iterative calculation process of the genetic algorithm.
Description
Technical Field
The invention relates to a method for designing an infrared phase change material based on deep learning, and belongs to the technical field of micro-nano optical structure design.
Background
In photonics, the forward calculation from the structure to the spectrum can be well calculated and understood by using maxwell's equations, but solving the inverse problem is often a complex problem, because the design usually involves mapping in a high-dimensional space, and it is difficult to integrate a plurality of design parameters at one time. The design of a traditional micro-nano optical structure is generally to perform simulation modeling on a nano structure, materials, surrounding media and an action wave band by using finite element software, and whether the spectral response of the nano structure can achieve an ideal effect or not is compared transversely by changing a parameter.
In the prior art, genetic algorithms are often used as optimization algorithms for reverse design. The genetic algorithm is a calculation model of the biological evolution process simulating the natural selection and genetic mechanism of the Darwin biological evolution theory, and is a method for searching the optimal solution by simulating the natural evolution process. The algorithm converts the solving process of the problem into processes like crossing, mutation and the like of chromosome genes in biological evolution by using a computer simulation operation in a mathematical mode. When solving the complex combination optimization problem, the speed advantage can be still presented compared with some conventional optimization algorithms.
The use of genetic algorithms to solve the inverse design problem of a set of systems, determines an objective function every time a structure is optimized, and requires a significant amount of computational power and time that grows exponentially with the complexity of the parameters. In popular terms, if a multi-layer Ag-GeTe core-shell structure is to be optimized to achieve super scattering in the 2000nm band, the thickness of each layer is first considered as a single design, or as an individual, all groups of elements are involved in forming a population, and the adaptability of each individual in the population is evaluated to obtain the possibility that the individual becomes the father of the next generation. For each structure design, the more the number of layers of the multi-layer ball is, the longer the design time is, if the two-layer structure is within 10 minutes, the four-layer structure is within 30 minutes, and the defect of lengthy iterative calculation process exists.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, and provides a method for designing an infrared phase change material based on deep learning, which can quickly design a structure according to required spectral response at any time and overcome the lengthy defect of iterative calculation process of a genetic algorithm.
In order to solve the technical problems, the invention adopts the following technical scheme:
a method of designing an infrared phase change material based on deep learning, comprising the steps of:
and a, determining the types of the inner layer material and the outer layer material, and determining the number of layers of the inner layer and the outer layer and the thickness range of the layers. In the design of the micro-nano material, a material selection link is added, so that the embarrassment that the design cannot be continued due to the problem of material limitation can be avoided, and the selection of the material is increased, so that more possibilities can be provided for the design.
Step b, respectively manufacturing data sets corresponding to different material combinations by using a Mie scattering transfer matrix algorithm, wherein the thickness of each layer of each material combination is randomly distributed;
and c, determining DNN network super parameters, and performing DNN network training on the data set based on the GPU. The DNN network with deep learning can be designed faster without long design time like other optimization algorithms, and design efficiency is improved.
Step d, determining a proper material for the next design according to the response characteristics of the spectrum in the classification result;
step e, training and optimizing a sub-network corresponding to the proper material, and designing structural parameters according to the sub-network;
and f, obtaining specific parameters of materials and structures.
In the step a, the inner and outer layer material structure is 2 layers, 4 layers or 6 layers, and each layer is defined as 30-200 nm.
In step b, two dielectric constants of the outer layer phase change material at low and high temperatures are selected as data set data.
In the step c, DNN network input layers are discrete value points of spectrums, wave bands are infrared wave bands of 1000-3000nm, each 5nm is used as an interval, 401 discrete points are shared, and the output classification layers respectively represent different inner layer materials. The design wave band of the spectrum is flexible to select, the whole spectrum input can be used for forecasting the structure, and the design can be carried out through an objective function.
In the step d, firstly, the spectral response is pre-trained as an input and the material is used as an output classification network, the output layer adopts a single-heat coding mode, then, after the network is obtained, the target spectral response is added into the input layer, and the most suitable metal layer material is obtained after the network is screened in the output layer.
In step e, the sub-network input is spectrum Y 0 The output is the thickness value of each layer, and the structure result predicted by the final network is brought into the forward network to obtain the correct predicted spectrum Y 1 Then the mean square error formula is used for obtaining Y 1 And Y is equal to 0 The loss value of the layer is determined, and the structural parameters of specific layers can obtain smaller loss.
The invention has the beneficial effects that: according to the method for designing the infrared phase change material based on the deep learning, the deep learning algorithm is adopted, and although the early network training is time-consuming, once the network is formed, the training speed is within a few seconds, and compared with other optimization algorithms, the training speed is obviously improved; the most suitable material can be automatically selected according to the target spectrum, the correct rate of the material selection is 83%, and even if the network prediction is not completely correct, the suitability of various materials for the target spectrum is shown; the phase change material which can well generate the ultra-scattering infrared wave band to the low scattering can be regulated and controlled to a specific wave band; the network can be related to materials with other functions according to a specific objective function, so that flexible design can be realized, and after the objective function is changed, the design can be completed rapidly within a few seconds.
Drawings
FIG. 1 is a schematic diagram of a multi-layer micro-nano-pellet phase change material according to the present invention;
FIG. 2 is a flow chart of a method for designing an infrared phase change material based on deep learning according to the present invention;
FIG. 3 shows the scatter spectrum of two layers of Ag-GeTe particles for different target peaks.
Detailed Description
The present invention will be further described with reference to the accompanying drawings, and the following examples are only for more clearly illustrating the technical aspects of the present invention, and are not to be construed as limiting the scope of the present invention.
As shown in fig. 2, the present invention provides a method for designing an infrared phase change material based on deep learning, comprising the steps of:
firstly, a process of manufacturing a data set is firstly to determine which materials, such as an inner layer material and an outer layer material GeTe of Ag, al and Au, are specifically included, and the number of layers and the thickness range of each layer are respectively 2 layers, 4 layers or 6 layers, and each layer is defined to be about 30-200 nm. Since GeTe is a phase change material, we mainly use two dielectric constants of GeTe at low and high temperatures. The design of optical structures may not achieve the desired spectral response due to limitations in the properties of the individual materials, and no optimal structure may be achieved anyway, since we only consider combinations of two materials during the design process, without trying a selected combination of materials. The structure of the multi-layer micro-nano small ball phase change material is shown in figure 1.
Step two, respectively preparing corresponding data sets of Ag-GeTe, al-GeTe and Au-GeTe by using a Mie scattering transfer matrix algorithm, wherein the thicknesses of each layer of each group are randomly distributed, and after the data sets are prepared, the data sets are as follows:
| 2 layers (Metal layer-GeTe layer) | 4 layers (Metal layer-GeTe layer-Metal layer-GeTe layer) | |
| Ag-GeTe | Data set 1-1 | Data sets 1-2 |
| Al-GeTe | Data set 2-1 | Data set 2-2 |
| Au-GeTe | Data set 3-1 | Data set 3-2 |
And thirdly, determining DNN network super parameters, and performing DNN network training on the data set based on the GPU. All data sets are integrated (i.e. data sets 1-1 to 3-2), a multi-layer perceptron of classified material is built, which is a deep learning model based on the fact that the physical model of the spectral response corresponding material is recorded by means of fully connected layers, the so-called deep neural network (deep nerual network) DNN. The reverse design method based on the deep learning algorithm is a multi-layer characterization learning technology, which can discover and classify the required characterization from the original data, and enables the neural network to learn the mapping from a very complex structure to response in a data driven manner through the combination of a large number of cascaded nonlinear neurons.
The network input layer is a discrete value point of a spectrum, the optimized wave band is an infrared wave band of 1000-3000nm, 401 discrete points are shared every 5nm as a gap, and the output classification layers respectively represent three materials.
And step four, carrying out classification training by taking all data sets, and judging which material is used for the next design according to response characteristics of the spectrum by the final classification result. The method comprises the following steps: the classification network of the spectral response (input) -material (output) is pre-trained, and the output layer adopts a single thermal coding (softmax) mode, namely, the metal layer in the form of 1-0-0 of the output layer is Ag, the metal layer in the form of 0-1-0 is Al and the metal layer in the form of 0-0-1 is Au, so that the accuracy of the classification network in training can be ensured. Then after the network is obtained, a target spectral response is added to the input layer, and a metal layer material which is most suitable after network screening can be obtained in the output layer, after which the data set of the material can be used for detailed design of each layer thickness of the multilayer particles.
And fifthly, training and optimizing a sub-network corresponding to the proper material, and designing structural parameters according to the sub-network. After the materials are determined, we train the sub-networks of the 2, 4 layer structure, respectively. The variables designed by the invention are the thickness of each of the Ag layer and the GeTe layer, and the number of the layers is 2 or 4 or more, and the variables are all relevant to the spectral response. For example, according to spectral characteristics, we determine that Ag-GeTe is selected for design according to the network, and then we train the multi-layer network of Ag-GeTe. The input of the network is the spectrum Y0, the output is the thickness value of each layer, the structure result predicted by the final network can be brought into the forward network to obtain the correct predicted spectrum Y1, then the mean square error formula is used for obtaining the loss values of Y1 and Y0, and the structure parameters of specific layers can be judged to obtain smaller loss, namely a better result.
And step six, obtaining specific parameters of materials and structures. Experiments prove that when the infrared phase-change scattering material is designed, the Ag-GeTe multilayer pellet with a two-layer structure has ideal effect, the loss value is 0.00532, and the loss value of four layers of materials is more than 1.
As shown in fig. 3, since the wavelength modulation capability is not large at high temperature, the scattering rate of the pellets at low temperature is mainly designed. The specific scattering peak position is defined in the near infrared band, the network can select the structural parameter of each layer (namely the thickness of each layer) according to the position, the scattering spectrum designed by two layers of Ag-GeTe particles for different target peaks is shown in fig. 3, and obvious regulation and control design effects can be seen.
The foregoing is only a preferred embodiment of the invention, it being noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present invention, and such modifications and adaptations are intended to be comprehended within the scope of the invention.
Claims (2)
1. A method for designing an infrared phase change material based on deep learning, which is characterized in that: the method comprises the following steps:
a, determining the types of inner layer materials and outer layer materials, wherein the inner layer materials are Ag, al and Au, the outer layer materials are GeTe, and determining the number of layers and the thickness range of the inner layer and the outer layer, wherein the number of layers of the inner layer and the outer layer is 2 layers, 4 layers or 6 layers, and each layer is defined at 30-200 nm;
step b, respectively manufacturing data sets corresponding to Ag-GeTe, al-GeTe and Au-GeTe by using a Mie scattering transfer matrix algorithm, wherein the thickness of each layer of each group of material combination is randomly distributed;
step c, determining DNN network super parameters, performing DNN network training on the data set based on the GPU, wherein DNN network input layers are discrete value points of a spectrum, wave bands are infrared wave bands of 1000-3000nm, each 5nm is used as an interval, 401 discrete points are arranged in total, and the output classification layers respectively represent different inner layer materials;
step d, determining a proper material for next design according to response characteristics of spectrums in classification results, firstly pre-training a classification network with the spectrum response as input and the material as output, wherein an output layer adopts a single-heat coding mode, namely a metal layer in the form of 1-0-0 of the output layer is Ag, a metal layer in the form of 0-1-0 is Al and a metal layer in the form of 0-0-1 is Au, then adding a target spectrum response into the input layer after obtaining the network, and obtaining the most proper metal layer material Ag-GeTe after network screening in the output layer;
step e, training and optimizing a sub-network corresponding to the Ag-GeTe, wherein the input of the sub-network is a spectrum Y according to the design structural parameters of the sub-network 0 The output is the thickness value of each layer, and the correct predicted spectrum Y is obtained by bringing the structure result predicted by the final network into the forward network 1 Then the mean square error formula is used for obtaining Y 1 And Y is equal to 0 The loss value of the specific layers is judged, so that smaller loss can be obtained;
and f, determining the infrared phase change material to be an Ag-GeTe multilayer pellet with a two-layer structure.
2. A method of designing an infrared phase change material based on deep learning as claimed in claim 1, wherein: in step b, two dielectric constants of the outer layer phase change material at low and high temperatures are selected as data set data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011022757.2A CN112182965B (en) | 2020-09-25 | 2020-09-25 | Method for designing infrared phase change material based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011022757.2A CN112182965B (en) | 2020-09-25 | 2020-09-25 | Method for designing infrared phase change material based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112182965A CN112182965A (en) | 2021-01-05 |
| CN112182965B true CN112182965B (en) | 2024-02-02 |
Family
ID=73944874
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011022757.2A Active CN112182965B (en) | 2020-09-25 | 2020-09-25 | Method for designing infrared phase change material based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112182965B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108872130A (en) * | 2018-06-25 | 2018-11-23 | 北京空间飞行器总体设计部 | Typical aircraft Facing material recognition methods neural network based |
| CN109271700A (en) * | 2018-09-10 | 2019-01-25 | 柯瑞林 | Battery thermal management method and system based on the modeling of deep learning multitiered network |
| JP6624533B1 (en) * | 2018-11-08 | 2019-12-25 | ジャパンモード株式会社 | Material property estimation program, Material generation mechanism estimation program |
| CN110826289A (en) * | 2019-10-29 | 2020-02-21 | 中国地质大学（武汉） | Deep learning-based nano structure design method |
| CN111177976A (en) * | 2019-12-25 | 2020-05-19 | 广东省焊接技术研究所（广东省中乌研究院） | Arc welding seam forming accurate prediction method based on deep learning |
| WO2020117348A2 (en) * | 2018-12-06 | 2020-06-11 | Western Digital Technologies, Inc. | Non-volatile memory die with deep learning neural network |
| WO2020130513A1 (en) * | 2018-12-18 | 2020-06-25 | 주식회사 포스코 | System and method for predicting material properties using metal microstructure images based on deep learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9697556B2 (en) * | 2007-09-06 | 2017-07-04 | Mohammad A. Mazed | System and method of machine learning based user applications |
| US20190278880A1 (en) * | 2018-03-12 | 2019-09-12 | Exxonmobil Research And Engineering Company | Hybrid computational materials fabrication |
-
2020
- 2020-09-25 CN CN202011022757.2A patent/CN112182965B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108872130A (en) * | 2018-06-25 | 2018-11-23 | 北京空间飞行器总体设计部 | Typical aircraft Facing material recognition methods neural network based |
| CN109271700A (en) * | 2018-09-10 | 2019-01-25 | 柯瑞林 | Battery thermal management method and system based on the modeling of deep learning multitiered network |
| JP6624533B1 (en) * | 2018-11-08 | 2019-12-25 | ジャパンモード株式会社 | Material property estimation program, Material generation mechanism estimation program |
| WO2020117348A2 (en) * | 2018-12-06 | 2020-06-11 | Western Digital Technologies, Inc. | Non-volatile memory die with deep learning neural network |
| WO2020130513A1 (en) * | 2018-12-18 | 2020-06-25 | 주식회사 포스코 | System and method for predicting material properties using metal microstructure images based on deep learning |
| CN110826289A (en) * | 2019-10-29 | 2020-02-21 | 中国地质大学（武汉） | Deep learning-based nano structure design method |
| CN111177976A (en) * | 2019-12-25 | 2020-05-19 | 广东省焊接技术研究所（广东省中乌研究院） | Arc welding seam forming accurate prediction method based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| Deep-learning-based inverse design model for intelligent discovery of organic molecules;Kyungdoc Kim等;《computational materials》;全文 * |
| 一种基于深度学习的复合材料结构损伤导波监测方法;杨宇等;《航空科学技术》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112182965A (en) | 2021-01-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Data-driven modeling using improved multi-objective optimization based neural network for coke furnace system | |
| Son et al. | A study on genetic algorithm to select architecture of a optimal neural network in the hot rolling process | |
| CN106355330A (en) | Multi-response parameter optimization method based on radial basis function neural network prediction model | |
| US20220108185A1 (en) | Inverse and forward modeling machine learning-based generative design | |
| CN108665070A (en) | Limit TS fuzzy reasoning methods based on extreme learning machine and system | |
| CN113129585B (en) | Road traffic flow prediction method based on graph aggregation mechanism of reconstructed traffic network | |
| CN113128124B (en) | Multi-grade C-Mn steel mechanical property prediction method based on improved neural network | |
| Jaddi et al. | Taguchi-based parameter designing of genetic algorithm for artificial neural network training | |
| CN107145935A (en) | A kind of Smith Prediction Control methods based on modified neutral net | |
| CN112182965B (en) | Method for designing infrared phase change material based on deep learning | |
| Oh et al. | Hybrid fuzzy set-based polynomial neural networks and their development with the aid of genetic optimization and information granulation | |
| Omizegba et al. | Optimizing fuzzy membership functions using particle swarm algorithm | |
| CN115438842A (en) | Load prediction method based on adaptive improved dayflies and BP neural network | |
| CN114202063A (en) | Fuzzy neural network greenhouse temperature prediction method based on genetic algorithm optimization | |
| Anastassiou et al. | A recurrent neural fuzzy network | |
| Su et al. | Modular Neural Networks for Time Series Forecasting: Interpretability and Feature Selection using Attention | |
| CN114265954A (en) | Graph representation learning method based on position and structure information | |
| Virdi et al. | Feature selection with LASSO and VSURF to model mechanical properties for investment casting | |
| Chen et al. | Pre-shaped fuzzy c-means algorithm (pfcm) for transparent membership function generation | |
| Takahama et al. | Structural learning of neural networks by coevolutionary genetic algorithm with degeneration | |
| Pham et al. | Evolutionary learning of fuzzy models | |
| CN115099519B (en) | Oil well yield prediction method based on multi-machine learning model fusion | |
| Tsimenidis | Evolutionary algorithms for fuzzy cognitive maps | |
| CN117336187B (en) | Unmanned aerial vehicle communication network inference method based on inter-edge association | |
| Vergini et al. | A comperative study using the methods of simulated annealing and Non-linear Hebbian Learning for Fuzzy Cognitive Maps performances |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |