Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural Network
To figure out the energy attenuation of micro/nanofibers (MNFs) more flexibly and conveniently, a backpropagation neural network (BPNN) is proposed to forecast the output intensity of rhodamine B (RhB) doped polymer microfibers (PMFs). According to the diameter, doping concentration, and...
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Photonics Journal |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9706252/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849722411790893056 |
|---|---|
| author | Hang Yu Juan Liu Jinjin Han Minghui Chen Mingjun Ke Zhili Lin Zhijun Wu Jixiong Pu Xining Zhang Hao Dai |
| author_facet | Hang Yu Juan Liu Jinjin Han Minghui Chen Mingjun Ke Zhili Lin Zhijun Wu Jixiong Pu Xining Zhang Hao Dai |
| author_sort | Hang Yu |
| collection | DOAJ |
| description | To figure out the energy attenuation of micro/nanofibers (MNFs) more flexibly and conveniently, a backpropagation neural network (BPNN) is proposed to forecast the output intensity of rhodamine B (RhB) doped polymer microfibers (PMFs). According to the diameter, doping concentration, and propagation distance (<italic>L</italic>), we realize the <italic>L</italic>-dependence of output energy predictions for the excitation light (<italic>I<sub>E</sub></italic>) and fluorescence (<italic>I<sub>F</sub></italic>) of the doped PMFs. Hundreds of propagation distance-intensity data pairs acquired from dozens of RhB doped PMFs are used for the BPNN training. The prediction ability of the model is evaluated by the root-mean-square error (RMSE), the mean absolute percentage error (MAPE), and R<sup>2</sup>. The output intensity prediction performance of BPNN is compared with the traditional exponential-fitting (EF) method. The prediction results indicate that the two-hidden-layer network with one and seventeen neurons respectively provides the best performance. After training, BPNN gives a good intensity prediction for both the <italic>I</italic><sub>E</sub> (RMSE = 3.16×10<sup>−2</sup>, MAPE = 7.3%, and R<sup>2</sup> = 0.9802) and the <italic>I</italic><sub>F</sub> (RMSE = 0.91×10<sup>−2</sup>, MAPE = 0.89%, and R<sup>2</sup> = 0.9696) from the output end of the PMF with different diameters and doping concentrations. The energy losses of the two kinds of light from different doped PMFs are also calculated based on the predicted values, which are similar to the ones obtained from the EF method. The approach based on the BPNN prediction for the energy attenuation of the PMFs shows superiority in flexibility and applicability toward the traditional methods, which could promote the optimal design of the MNF devices and the practical application. |
| format | Article |
| id | doaj-art-d8eafcf9be284fe89ea5187a9080c8ab |
| institution | DOAJ |
| issn | 1943-0655 |
| language | English |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Photonics Journal |
| spelling | doaj-art-d8eafcf9be284fe89ea5187a9080c8ab2025-08-20T03:11:21ZengIEEEIEEE Photonics Journal1943-06552022-01-011421810.1109/JPHOT.2022.31487379706252Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural NetworkHang Yu0Juan Liu1https://orcid.org/0000-0002-5526-6573Jinjin Han2Minghui Chen3Mingjun Ke4Zhili Lin5https://orcid.org/0000-0002-1336-9347Zhijun Wu6https://orcid.org/0000-0002-1057-0205Jixiong Pu7https://orcid.org/0000-0001-8781-6683Xining Zhang8https://orcid.org/0000-0003-1845-1400Hao Dai9https://orcid.org/0000-0002-3843-8185Fujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaFujian Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, ChinaInstitute of Ocean Exploration Technology, College of Ocean and Earth Sciences, Xiamen University, Xiamen, ChinaTo figure out the energy attenuation of micro/nanofibers (MNFs) more flexibly and conveniently, a backpropagation neural network (BPNN) is proposed to forecast the output intensity of rhodamine B (RhB) doped polymer microfibers (PMFs). According to the diameter, doping concentration, and propagation distance (<italic>L</italic>), we realize the <italic>L</italic>-dependence of output energy predictions for the excitation light (<italic>I<sub>E</sub></italic>) and fluorescence (<italic>I<sub>F</sub></italic>) of the doped PMFs. Hundreds of propagation distance-intensity data pairs acquired from dozens of RhB doped PMFs are used for the BPNN training. The prediction ability of the model is evaluated by the root-mean-square error (RMSE), the mean absolute percentage error (MAPE), and R<sup>2</sup>. The output intensity prediction performance of BPNN is compared with the traditional exponential-fitting (EF) method. The prediction results indicate that the two-hidden-layer network with one and seventeen neurons respectively provides the best performance. After training, BPNN gives a good intensity prediction for both the <italic>I</italic><sub>E</sub> (RMSE = 3.16×10<sup>−2</sup>, MAPE = 7.3%, and R<sup>2</sup> = 0.9802) and the <italic>I</italic><sub>F</sub> (RMSE = 0.91×10<sup>−2</sup>, MAPE = 0.89%, and R<sup>2</sup> = 0.9696) from the output end of the PMF with different diameters and doping concentrations. The energy losses of the two kinds of light from different doped PMFs are also calculated based on the predicted values, which are similar to the ones obtained from the EF method. The approach based on the BPNN prediction for the energy attenuation of the PMFs shows superiority in flexibility and applicability toward the traditional methods, which could promote the optimal design of the MNF devices and the practical application.https://ieeexplore.ieee.org/document/9706252/Backpropagation neural networkdye-doped polymer microfiberenergy attenuation prediction |
| spellingShingle | Hang Yu Juan Liu Jinjin Han Minghui Chen Mingjun Ke Zhili Lin Zhijun Wu Jixiong Pu Xining Zhang Hao Dai Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural Network IEEE Photonics Journal Backpropagation neural network dye-doped polymer microfiber energy attenuation prediction |
| title | Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural Network |
| title_full | Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural Network |
| title_fullStr | Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural Network |
| title_full_unstemmed | Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural Network |
| title_short | Energy Attenuation Prediction of Dye-Doped PMMA Microfibers by Backpropagation Neural Network |
| title_sort | energy attenuation prediction of dye doped pmma microfibers by backpropagation neural network |
| topic | Backpropagation neural network dye-doped polymer microfiber energy attenuation prediction |
| url | https://ieeexplore.ieee.org/document/9706252/ |
| work_keys_str_mv | AT hangyu energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT juanliu energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT jinjinhan energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT minghuichen energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT mingjunke energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT zhililin energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT zhijunwu energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT jixiongpu energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT xiningzhang energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork AT haodai energyattenuationpredictionofdyedopedpmmamicrofibersbybackpropagationneuralnetwork |