CN113920408A - Photonic crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion - Google Patents
Photonic crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion Download PDFInfo
- Publication number
- CN113920408A CN113920408A CN202111061720.5A CN202111061720A CN113920408A CN 113920408 A CN113920408 A CN 113920408A CN 202111061720 A CN202111061720 A CN 202111061720A CN 113920408 A CN113920408 A CN 113920408A
- Authority
- CN
- China
- Prior art keywords
- transmission spectrum
- photonic crystal
- rnn
- cnn
- parallel fusion
- 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.)
- Granted
Links
- 230000004927 fusion Effects 0.000 title claims abstract description 55
- 239000004038 photonic crystal Substances 0.000 title claims abstract description 45
- 238000000411 transmission spectrum Methods 0.000 title claims abstract description 39
- 238000000605 extraction Methods 0.000 title claims abstract description 25
- 238000003062 neural network model Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000000034 method Methods 0.000 claims description 28
- 238000011156 evaluation Methods 0.000 claims description 14
- 238000005457 optimization Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 230000002238 attenuated effect Effects 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000001228 spectrum Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 2
- 238000013461 design Methods 0.000 abstract description 2
- 230000000737 periodic effect Effects 0.000 description 5
- 239000013078 crystal Substances 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 102100032202 Cornulin Human genes 0.000 description 1
- 101000920981 Homo sapiens Cornulin Proteins 0.000 description 1
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000003989 dielectric material Substances 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000000053 physical method Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—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)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Optical Modulation, Optical Deflection, Nonlinear Optics, Optical Demodulation, Optical Logic Elements (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a photonic crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion, which relates to the processing and analysis of general sequence data, and comprises the following steps: collecting a transmission spectrum sequence sample of the one-dimensional photonic crystal and structural parameters to form a training data set, and performing feature extraction and parameter prediction through a constructed neural network model based on CNN-RNN parallel fusion. The transmission spectrum sequence is analyzed by utilizing the more comprehensive sequence characteristics after fusion, and the characteristic extraction and structure design performance of the transmission spectrum sequence is effectively improved.
Description
Technical Field
The technical scheme of the invention relates to a general sequence type data processing method, in particular to a photonic crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion.
Background
The concept of photonic crystals is based on the conventional analogy to electronic crystals. A typical crystal, such as a semiconductor, is a periodic structure formed by an ordered arrangement of internal atoms. It is this periodic arrangement that produces a periodic potential field in the crystal, so that electrons moving therein are bragg-scattered by the periodic potential field to form an energy band structure, with possible energy gaps between bands; similarly, photonic crystals are formed by periodically arranging dielectric materials having different dielectric constants in space, and can control light waves propagating therein to form a band structure.
Generally, the photonic crystal is analyzed according to electromagnetic theory, and mainly there are a scattering matrix method, a Finite Difference Time Domain (FDTD), and the like. The scattering matrix method occupies a large amount of memory and time in the solving process and is difficult to apply to practical research, and although the latter method greatly reduces the operation amount, the latter method is difficult to accurately solve the photonic crystal with a special shape due to neglect of the shape of the crystal lattice in the calculating process. In addition, there are also difficulties in the preparation of photonic crystals. Due to the irreversibility of manufacturing, error detection of the finished product will inevitably cause equipment damage when applying traditional physical methods.
The development of deep learning provides a new perspective for photonic crystal research, and many optical applications are increasingly combined and benefit from deep network models. The former people realized the prediction between the dispersion relation and the photonic band gap in the two-dimensional photonic crystal through the most basic multilayer perceptron, but the characteristic extraction of the basic neural network model is too simple, and the characteristic extraction and the parameter prediction cannot be well carried out on some special sequence type data.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the method can fuse the space geometric characteristics and the time sequence characteristics extracted through a one-dimensional convolution network and a circulation network, analyzes the transmission spectrum sequence by utilizing the more comprehensive sequence characteristics after fusion, and effectively improves the characteristic extraction and structure design performance of the transmission spectrum sequence.
The technical scheme adopted by the invention for solving the technical problem is as follows: collecting a transmission spectrum sequence sample of the one-dimensional photonic crystal and structural parameters to form a training data set, and performing feature extraction and parameter prediction through a constructed neural network model based on CNN-RNN parallel fusion. The method comprises the following specific steps:
(1) acquiring a training data set:
(1a) collecting a transmission spectrum sequence of the one-dimensional photonic crystal, and exporting a mat format file as a data sample;
(1b) carrying out data preprocessing on the photonic crystal structure parameters corresponding to each transmission spectrum sequence, dividing the data by the maximum value of each parameter respectively, carrying out normalization, and deriving a csv format file as a sample label;
(2) constructing a neural network model based on CNN-RNN parallel fusion;
(3) training the constructed neural network model by using the acquired data set, and improving the characteristic extraction and parameter prediction performance of the model by adopting an optimization algorithm;
(4) and evaluating the performance of the network model by calculating indexes such as mean square error, root mean square error, average absolute error, average relative error, average absolute percentage error, symmetric average absolute percentage error and the like.
According to the method for extracting the characteristic of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion, the transmission spectrum sequences of various types of photonic crystals with continuous periods, discontinuous periods and the like are respectively collected as different training samples by obtaining the training data set, the relevant structure parameters after pretreatment are used as label data, and the universality of the characteristic extraction network is verified through multiple groups of experiments.
The method for extracting the characteristics of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion comprises the following steps that a neural network model structure comprises a parallel fusion module and a full connection layer, wherein the parallel fusion module is used for extracting the spatial pixel characteristics and the time sequence characteristics of a spectrum sequence and performing fusion transmission; and the full connection layer is used for mapping the fusion sequence characteristics output by the last parallel fusion module into photonic crystal structure parameters.
The parallel fusion module comprises a one-dimensional convolution layer and a circulation layer, when the characteristic sequence passes through the module, the two network layers are used for acquiring space pixel characteristics and time sequence characteristics at the same time, the two network layers are fused according to the ratio of 1:1, and the obtained new characteristic sequence is used as the input of the next layer of parallel fusion module to continue to extract the characteristics.
According to the method for extracting the photonic crystal transmission spectrum sequence characteristics based on the CNN-RNN parallel fusion, the neural network model dynamically adjusts the learning rate by using the adaptive algorithm ReduceLROnPateau in the training process, and when the verification loss does not change in a certain iteration period, the learning rate is attenuated according to a specified multiplying power.
The method for extracting the characteristics of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion comprises the following steps of evaluating the characteristics of the network model and predicting the parameters of the network model by using various evaluation indexes including mean square error, root mean square error, mean absolute error, mean relative error, mean absolute percentage error, symmetrical mean absolute percentage error and the like:
wherein n represents the number of samples,
indicates the predicted value, yiThe actual value is represented, the difference of the real data and the prediction data is reflected by the six evaluation indexes in different modes, and the higher the model performance is, the closer the evaluation index value is to 0.
The invention has the beneficial effects that: compared with the prior art, the invention has the prominent substantive characteristics and remarkable progress as follows:
the method comprises the steps of obtaining a training set containing a photonic crystal transmission spectrum sequence and related structure parameters; constructing a depth network model based on CNN-RNN parallel fusion; training the constructed network model by using the acquired data set, and improving the model feature extraction and parameter prediction performance by adopting an optimization algorithm; and performing model performance evaluation through multiple regression prediction evaluation indexes. In the process, the space geometric characteristics extracted through the one-dimensional convolution network and the time sequence characteristics extracted through the circulation layer are fused, and the transmission spectrum sequence is analyzed by utilizing the fused more comprehensive sequence characteristics, so that an excellent parameter prediction effect can be obtained when the periodic continuous photonic crystal is faced, and the same effect can be obtained in other similar serialized sample analysis.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a schematic flow chart of a photonic crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion provided by the invention.
FIG. 2 is a neural network model for CNN-RNN parallel fusion provided by the present invention.
FIG. 3 is a schematic diagram of a parallel fusion module of the neural network model provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.
As shown in FIG. 1, the invention provides a photonic crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion. Specific examples are as follows:
(1) acquiring a training data set:
(1a) collecting a transmission spectrum sequence of the DCG double-period one-dimensional photonic crystal, and exporting a mat format file as a data sample;
(1b) carrying out data preprocessing on the photonic crystal structure parameters corresponding to each transmission spectrum sequence, dividing the data by the maximum value of each parameter respectively, carrying out normalization, and deriving a csv format file as a sample label;
(2) constructing a neural network model based on CNN-RNN parallel fusion;
(3) training the constructed neural network model by using the acquired data set, and improving the characteristic extraction and parameter prediction performance of the model by adopting an optimization algorithm;
(4) and evaluating the performance of the network model by calculating indexes such as mean square error, root mean square error, average absolute error, average relative error, average absolute percentage error, symmetric average absolute percentage error and the like.
In a specific implementation manner of the method for extracting the characteristic of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion, further, the acquisition of the training data set respectively collects transmission spectrum sequences of multiple types of photonic crystals with continuous periods, discontinuous periods and the like as different training samples, and verifies the universality of the characteristic extraction network through multiple groups of experiments by using relevant structure parameters after preprocessing as label data.
In the foregoing specific embodiment of the method for extracting the characteristic of the photonic crystal transmission spectrum sequence based on CNN-RNN parallel fusion, as shown in fig. 2 and 3, the neural network model structure includes a parallel fusion module and a full connection layer, and the parallel fusion module is configured to extract spatial pixel characteristics and time sequence characteristics of the spectrum sequence and perform fusion transmission; and the full connection layer is used for mapping the fusion sequence characteristics output by the last parallel fusion module into photonic crystal structure parameters.
In a specific implementation manner of the method for extracting the characteristic of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion, further, the parallel fusion module includes a one-dimensional convolution layer and a circulation layer, when the characteristic sequence passes through the module, the two network layers are used to obtain the spatial pixel characteristic and the time sequence characteristic at the same time, and fusion is performed according to a ratio of 1:1, so that the obtained new characteristic sequence is used as the input of the next parallel fusion module to continue to perform the characteristic extraction.
In a specific implementation manner of the method for extracting the transmission spectrum sequence features of the photonic crystal based on the CNN-RNN parallel fusion, further, the neural network model dynamically adjusts the learning rate by using an adaptive algorithm reduce lron plateau during training, and when the verification loss does not change within a certain iteration period, the learning rate is attenuated according to a specified multiplying factor.
In a specific implementation manner of the method for extracting the characteristic of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion, further, the multiple network model performance evaluation indexes, including multiple evaluation indexes such as mean square error, root mean square error, average absolute error, average relative error, average absolute percentage error, symmetric average absolute percentage error, etc., evaluate the characteristic extraction and parameter prediction performance of the network model:
wherein n represents the number of samples,
indicates the predicted value, yiThe actual value is represented, the difference of the real data and the prediction data is reflected by the six evaluation indexes in different modes, and the higher the model performance is, the closer the evaluation index value is to 0.
The embodiment compares the prediction performances of different neural network models through different evaluation indexes. In particular as
Shown in table 1.
| MSE | RMSE | MAE | MRE | MAPE | SMAPE | |
| MLP | 3.51e-05 | 5.92e-03 | 4.58e-03 | 9.19e-03 | 0.9194％ | 0.9195％ |
| 1D-CNN | 2.66e-05 | 5.15e-03 | 3.98e-03 | 8.67e-03 | 0.8669％ | 0.8667％ |
| RNN | 5.96e-05 | 7.72e-03 | 6.19e-03 | 1.08e-02 | 1.0811％ | 1.0785％ |
| CRNN | 1.59e-05 | 3.99e-03 | 3.04e-03 | 6.65e-03 | 0.6647％ | 0.6671％ |
| Ours | 4.73e-06 | 2.18e-03 | 1.69e-04 | 3.73e-04 | 0.3727％ | 0.3728％ |
As can be seen from Table 1, the neural network model based on CNN-RNN parallel fusion has the highest prediction performance among all the neural network models, and the prediction performance is respectively 4.73e-06, 2.18e-03, 1.69e-04, 3.73e-04, 0.3727% and 0.3728% for different evaluation index comparisons. The effectiveness of the method is proved, and the effect that different characteristic sequences complement each other is also shown.
It should be noted that the above mentioned embodiments are only examples of the present invention, and are not intended to limit the scope of the claims, and that the present invention is not limited thereto.
Claims (6)
1. A photonic crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion is characterized by comprising the following steps: collecting a transmission spectrum sequence sample of a one-dimensional photonic crystal and structural parameters to form a training data set, and performing feature extraction and parameter prediction through a constructed neural network model based on CNN-RNN parallel fusion, wherein the method comprises the following specific steps:
(1) acquiring a training data set:
(1a) collecting a transmission spectrum sequence of the one-dimensional photonic crystal, and exporting a mat format file as a data sample;
(1b) carrying out data preprocessing on the photonic crystal structure parameters corresponding to each transmission spectrum sequence, dividing the data by the maximum value of each parameter respectively, carrying out normalization, and deriving a csv format file as a sample label;
(2) constructing a neural network model based on CNN-RNN parallel fusion;
(3) training the constructed neural network model by using the acquired data set, and improving the characteristic extraction and parameter prediction performance of the model by adopting an optimization algorithm;
(4) and evaluating the performance of the network model by calculating indexes such as mean square error, root mean square error, average absolute error, average relative error, average absolute percentage error, symmetric average absolute percentage error and the like.
2. The method for extracting the transmission spectrum sequence features of the photonic crystal based on the CNN-RNN parallel fusion as claimed in claim 1, wherein: in the step (1), transmission spectrum sequences of multiple types of photonic crystals with continuous periods, discontinuous periods and the like are respectively collected as different training samples, relevant structure parameters after pretreatment are used as label data, and universality of the feature extraction network is verified through multiple groups of experiments.
3. The method for extracting the characteristic of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion as claimed in claim 1, wherein in the step (2), the neural network model structure comprises a parallel fusion module and a full connection layer, and the parallel fusion module is used for extracting the spatial pixel characteristic and the time sequence characteristic of the spectrum sequence and performing fusion transmission; and the full connection layer is used for mapping the fusion sequence characteristics output by the last parallel fusion module into photonic crystal structure parameters.
4. The method for extracting the characteristic of the photonic crystal transmission spectrum sequence based on the CNN-RNN parallel fusion as claimed in claim 3, wherein the parallel fusion module comprises a one-dimensional convolution layer and a circulation layer, when the characteristic sequence passes through the module, the two network layers are simultaneously used for obtaining the spatial pixel characteristic and the time sequence characteristic, the characteristic sequence is fused according to the ratio of 1:1, and the obtained new characteristic sequence is used as the input of the next layer of parallel fusion module to continue the characteristic extraction.
5. The method for extracting the transmission spectrum sequence feature of the photonic crystal based on the CNN-RNN parallel fusion as claimed in claim 1, wherein in the step (3), during the training process, the neural network model dynamically adjusts the learning rate by using an adaptive algorithm ReduceLROnPlateau, and when the verification loss does not change within a certain iteration period, the learning rate is attenuated according to a specified rate.
6. The method for extracting the characteristics of the transmission spectrum sequence of the photonic crystal based on the CNN-RNN parallel fusion as claimed in claim 1, wherein in the step (4), the characteristics extraction and parameter prediction performance of the network model are evaluated by adopting a plurality of evaluation indexes such as mean square error, root mean square error, mean absolute error, mean relative error, mean absolute percentage error and symmetric mean absolute percentage error:
wherein n represents the number of samples,
indicates the predicted value, yiThe actual value is represented, the difference of the real data and the prediction data is reflected by the six evaluation indexes in different modes, and the higher the model performance is, the closer the evaluation index value is to 0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111061720.5A CN113920408B (en) | 2021-09-10 | 2021-09-10 | Photon crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111061720.5A CN113920408B (en) | 2021-09-10 | 2021-09-10 | Photon crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113920408A true CN113920408A (en) | 2022-01-11 |
| CN113920408B CN113920408B (en) | 2024-07-09 |
Family
ID=79234561
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111061720.5A Active CN113920408B (en) | 2021-09-10 | 2021-09-10 | Photon crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113920408B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111738512A (en) * | 2020-06-22 | 2020-10-02 | 昆明理工大学 | Short-term power load prediction method based on CNN-IPSO-GRU hybrid model |
| CN112001486A (en) * | 2020-08-28 | 2020-11-27 | 河北工业大学 | Load decomposition method based on deep learning |
| WO2021082809A1 (en) * | 2019-10-29 | 2021-05-06 | 山东科技大学 | Training optimization method for foreign exchange time series prediction |
| WO2021082811A1 (en) * | 2019-10-29 | 2021-05-06 | 山东科技大学 | Foreign exchange time series prediction method |
| CN113111958A (en) * | 2021-04-23 | 2021-07-13 | 中南大学 | Spectrum model transfer method based on CNN-SVR model and transfer learning |
| CN113158563A (en) * | 2021-04-16 | 2021-07-23 | 华南理工大学 | One-dimensional photonic crystal reverse design method based on neural network and Monte Carlo |
-
2021
- 2021-09-10 CN CN202111061720.5A patent/CN113920408B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021082809A1 (en) * | 2019-10-29 | 2021-05-06 | 山东科技大学 | Training optimization method for foreign exchange time series prediction |
| WO2021082811A1 (en) * | 2019-10-29 | 2021-05-06 | 山东科技大学 | Foreign exchange time series prediction method |
| CN111738512A (en) * | 2020-06-22 | 2020-10-02 | 昆明理工大学 | Short-term power load prediction method based on CNN-IPSO-GRU hybrid model |
| CN112001486A (en) * | 2020-08-28 | 2020-11-27 | 河北工业大学 | Load decomposition method based on deep learning |
| CN113158563A (en) * | 2021-04-16 | 2021-07-23 | 华南理工大学 | One-dimensional photonic crystal reverse design method based on neural network and Monte Carlo |
| CN113111958A (en) * | 2021-04-23 | 2021-07-13 | 中南大学 | Spectrum model transfer method based on CNN-SVR model and transfer learning |
Non-Patent Citations (2)
| Title |
|---|
| 卢霞;王春艳;谢宏全;彭红春;李海英: "浒苔水体光谱特征参量时间序列数据拟合研究", 海洋科学, vol. 39, no. 4, 31 December 2015 (2015-12-31) * |
| 郭淑涛;赵德新;: "一种基于深度学习的中文图像描述模型", 天津理工大学学报, no. 03, 15 June 2020 (2020-06-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113920408B (en) | 2024-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113673775B (en) | Space-time combined prediction method based on CNN-LSTM and deep learning | |
| CN112350876A (en) | Network flow prediction method based on graph neural network | |
| CN111193256A (en) | Power load prediction method based on variational modal decomposition and gated cyclic unit | |
| CN115203865B (en) | Product assembly process mechanical performance online prediction method based on digital twinning | |
| CN113887328A (en) | Method for extracting space-time characteristics of photonic crystal space transmission spectrum in parallel by ECA-CNN fusion dual-channel RNN | |
| CN108182316B (en) | Electromagnetic simulation method based on artificial intelligence and electromagnetic brain thereof | |
| CN113761777B (en) | HP-OVMD-based ultra-short-term photovoltaic power prediction method | |
| CN109904878B (en) | Multi-wind-field power generation time sequence simulation scene construction method | |
| CN112613536A (en) | Near infrared spectrum diesel grade identification method based on SMOTE and deep learning | |
| CN114693064B (en) | Building group scheme generation performance evaluation method | |
| CN116258914B (en) | Remote Sensing Image Classification Method Based on Machine Learning and Local and Global Feature Fusion | |
| CN105825227A (en) | Hyperspectral image sparseness demixing method based on MFOCUSS and low-rank expression | |
| Wang et al. | Global aligned structured sparsity learning for efficient image super-resolution | |
| CN116128141A (en) | Storm surge prediction method and device, storage medium and electronic equipment | |
| CN110516727B (en) | Hyperspectral image classification method based on FPGA (field programmable Gate array) depth edge filter | |
| CN110516792A (en) | Non-stable time series forecasting method based on wavelet decomposition and shallow-layer neural network | |
| CN117875481A (en) | Carbon emission prediction method, electronic device, and computer-readable medium | |
| CN113920408B (en) | Photon crystal transmission spectrum sequence feature extraction method based on CNN-RNN parallel fusion | |
| CN113705394A (en) | Behavior identification method combining long and short time domain features | |
| CN111680846A (en) | Simplified width learning system based on L1 and L2 norms | |
| CN114974476A (en) | Method for selecting perovskite battery modification material | |
| CN112836936B (en) | Risk factor screening method and system for power sensing network perception layer planning | |
| CN112231933B (en) | Feature selection method for radar electromagnetic interference effect analysis | |
| Xiaosheng et al. | Short-Term wind power prediction based on wavelet transform and convolutional neural networks | |
| CN113361417B (en) | Human behavior recognition method based on variable time sequence |
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 | ||
| CP03 | Change of name, title or address |
Address after: 300384 No. 391 Binshui West Road, Xiqing District, Tianjin Patentee after: TIANJIN University OF TECHNOLOGY Country or region after: China Address before: 300384 No.391, Binshui West Road, Xiqing District, Tianjin Patentee before: TIANJIN University OF TECHNOLOGY Country or region before: China |