Wenjuan Sheng
Wenjuan Sheng has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
G.D. Peng
3 shared publicationsGangDing Peng
University of New South Wales
2 shared publicationsJianxiang Wen
2 shared publications- Dirk Söffker
University of Duisburg-Essen
1 shared publication Haiqi Dang
Shanghai University of Electric Power
1 shared publicationHaitao Lou
Shanghai Univ Elect Power, Shanghai 200090, Peoples R China
1 shared publication
Latest Publications
- 19 November 2025
- Published by Elsevier in Measurement
- Vol. 259
Enhanced state of charge estimation and surface strain evolution of lithium-ion batteries via quasi-distributed optical sensing
- 19 November 2025
- Published by Elsevier in Measurement
- Vol. 259
Enhanced state of charge estimation and surface strain evolution of lithium-ion batteries via quasi-distributed optical sensing
- 30 December 2023
- Published by Elsevier in Optics & Laser Technology
- Vol. 172
Accurate and fast calibration for FBG demodulation based on deep learning and ensemble learning
- 30 December 2023
- Published by Elsevier in Optics & Laser Technology
- Vol. 172
Accurate and fast calibration for FBG demodulation based on deep learning and ensemble learning
- 30 October 2023
- Published by Elsevier in Sensors and Actuators A: Physical
- Vol. 363
Online temperature drift compensation of Fabry-Perot filter based on machine learning and linear fitting
- 30 October 2023
- Published by Elsevier in Sensors and Actuators A: Physical
- Vol. 363
Online temperature drift compensation of Fabry-Perot filter based on machine learning and linear fitting
- 1 March 2023
- Published by Optica Publishing Group in Optics Express
- Vol. 31 (6) , 9657-9668
Drift calibration method of Fabry-Perot filters using two-stage decomposition and hybrid modeling
Although tunable Fabry-Perot (F-P) filters are widely acknowledged as fiber Bragg grating (FBG) demodulators, F-P filters exhibit drift error when subjected to ambient temperature and piezo-electrical transducer (PZT) hysteresis. To address the drift issue, the majority of the existing literature makes use of additional devices like the F-P etalon and gas chamber. In this study, a novel drift calibration method based on two-stage decomposition and hybrid modeling is proposed. The initial drift error sequences are broken down into three frequency components using the variational mode decomposition (VMD), and the medium-frequency components are further broken down using the secondary VMD. The initial drift error sequences are significantly simplified by the two-stage VMD. On this foundation, the long short-term memory (LSTM) network and polynomial fitting (PF) are used to forecast the low-frequency and high-frequency drift errors, respectively. The LSTM enables the prediction of intricate nonlinear local behaviors, while the PF method predicts the overall trend. The benefits of LSTM and PF can be effectively utilized in this manner. Compared to the single-stage decomposition, two-stage decomposition achieves superior results. The suggested method is an affordable and effective alternative to the current drift calibration techniques.
- 1 March 2023
- Published by Optica Publishing Group in Optics Express
- Vol. 31 (6) , 9657-9668
Drift calibration method of Fabry-Perot filters using two-stage decomposition and hybrid modeling
Although tunable Fabry-Perot (F-P) filters are widely acknowledged as fiber Bragg grating (FBG) demodulators, F-P filters exhibit drift error when subjected to ambient temperature and piezo-electrical transducer (PZT) hysteresis. To address the drift issue, the majority of the existing literature makes use of additional devices like the F-P etalon and gas chamber. In this study, a novel drift calibration method based on two-stage decomposition and hybrid modeling is proposed. The initial drift error sequences are broken down into three frequency components using the variational mode decomposition (VMD), and the medium-frequency components are further broken down using the secondary VMD. The initial drift error sequences are significantly simplified by the two-stage VMD. On this foundation, the long short-term memory (LSTM) network and polynomial fitting (PF) are used to forecast the low-frequency and high-frequency drift errors, respectively. The LSTM enables the prediction of intricate nonlinear local behaviors, while the PF method predicts the overall trend. The benefits of LSTM and PF can be effectively utilized in this manner. Compared to the single-stage decomposition, two-stage decomposition achieves superior results. The suggested method is an affordable and effective alternative to the current drift calibration techniques.
- 27 May 2022
- Published by SAGE Publications in Structural Health Monitoring
- Vol. 22 (2) , 1273-1289
A novel adaptive boosting algorithm with distance-based weighted least square support vector machine and filter factor for carbon fiber reinforced polymer multi-damage classification
Adaptive boosting (AdaBoost) algorithms fuse multiple weak classifiers to generate a strong classifier by adaptively determining the fusion weights of the weak classifiers. According to incorrect or correct classification results, sample weights become larger or smaller. However, this weight update scheme neglects valuable information in the results. Moreover, an important requirement for weak classifiers is an accuracy higher than random guessing. This requirement is likely to lead to an unexpected result. This means that several generated weak classifiers with similar classification results cannot learn from each other. Consequently, the advantage of fusing multiple weak classifiers disappears. The classification and therefore distinction of different failure modes in materials is a typical task for classical nondestructive testing approaches as well as for new approaches based on machine learning methods. In the case different approaches can be applied, the main question is, which and how tuned approaches provide the best results in terms of accuracy or similar metrics. In the multi-damage classification task distinguishing different physical failure mechanisms in Carbon Fiber Reinforced Polymer (CFRP), the above two aspects complicate the application of AdaBoost algorithms. To improve the results, a novel AdaBoost with distance-based weighted least square support vector machine (WLSSVM) and filter factor is proposed. The distance-based WLSSVM is employed to increase the diversity of weak classifiers, the distances of the classification plane and samples are used to measure the classification task. The filter factor is proposed to filter out unnecessary classifiers contributing less to the final decision. The experimental results demonstrate that the improved AdaBoost schemes with distance-based WLSSVM and filter factor outperform state-of-the-art algorithms. The effects of the scheme using the new weighted update and the filter factor on the algorithm are discussed, respectively. The experimental results show that the combination of the two schemes perform better than other schemes.
- 27 May 2022
- Published by SAGE Publications in Structural Health Monitoring
- Vol. 22 (2) , 1273-1289
A novel adaptive boosting algorithm with distance-based weighted least square support vector machine and filter factor for carbon fiber reinforced polymer multi-damage classification
Adaptive boosting (AdaBoost) algorithms fuse multiple weak classifiers to generate a strong classifier by adaptively determining the fusion weights of the weak classifiers. According to incorrect or correct classification results, sample weights become larger or smaller. However, this weight update scheme neglects valuable information in the results. Moreover, an important requirement for weak classifiers is an accuracy higher than random guessing. This requirement is likely to lead to an unexpected result. This means that several generated weak classifiers with similar classification results cannot learn from each other. Consequently, the advantage of fusing multiple weak classifiers disappears. The classification and therefore distinction of different failure modes in materials is a typical task for classical nondestructive testing approaches as well as for new approaches based on machine learning methods. In the case different approaches can be applied, the main question is, which and how tuned approaches provide the best results in terms of accuracy or similar metrics. In the multi-damage classification task distinguishing different physical failure mechanisms in Carbon Fiber Reinforced Polymer (CFRP), the above two aspects complicate the application of AdaBoost algorithms. To improve the results, a novel AdaBoost with distance-based weighted least square support vector machine (WLSSVM) and filter factor is proposed. The distance-based WLSSVM is employed to increase the diversity of weak classifiers, the distances of the classification plane and samples are used to measure the classification task. The filter factor is proposed to filter out unnecessary classifiers contributing less to the final decision. The experimental results demonstrate that the improved AdaBoost schemes with distance-based WLSSVM and filter factor outperform state-of-the-art algorithms. The effects of the scheme using the new weighted update and the filter factor on the algorithm are discussed, respectively. The experimental results show that the combination of the two schemes perform better than other schemes.
- 19 November 2021
- Published by Optica Publishing Group in Optics Express
- Vol. 29 (24) , 40547-40558
Hysteresis and temperature drift compensation for FBG demodulation by utilizing adaptive weight least square support vector regression
Hysteresis and temperature drift deteriorate the demodulation performance of tunable Fabry-Perot (F-P) filters. This study addresses a novel adaptive weight least square support vector regression (AWLSSVR) to compensate for the hysteresis and temperature drift of F-P filters. The temperature drift of a referent fiber Bragg grating(FBG) is used to estimate the temperature drifts of other three sensing FBGs, and a novel adaptive weighting strategy with an asymmetric noise interval is proposed, to eliminate the effects of noise in the training dataset. The experimental results show that when the temperature-changing modes of the training and testing datasets were close to each other, the error of the proposed method is reduced to 8.7 pm, while the errors of the other three conventional methods based on LSSVR are more than 10.8 pm. Further, when the temperature-changing modes of the training and testing datasets were partly different, the error of the proposed method was reduced to 5.4 pm, while the errors of other methods were more than 11.9 pm. It was verified experimentally that the proposed AWLSSVR method is more accurate and robust than other versions of WLSSVR for training samples with noise, requires no additional hardware, and covers the entire C band.
- 19 November 2021
- Published by Optica Publishing Group in Optics Express
- Vol. 29 (24) , 40547-40558
Hysteresis and temperature drift compensation for FBG demodulation by utilizing adaptive weight least square support vector regression
Hysteresis and temperature drift deteriorate the demodulation performance of tunable Fabry-Perot (F-P) filters. This study addresses a novel adaptive weight least square support vector regression (AWLSSVR) to compensate for the hysteresis and temperature drift of F-P filters. The temperature drift of a referent fiber Bragg grating(FBG) is used to estimate the temperature drifts of other three sensing FBGs, and a novel adaptive weighting strategy with an asymmetric noise interval is proposed, to eliminate the effects of noise in the training dataset. The experimental results show that when the temperature-changing modes of the training and testing datasets were close to each other, the error of the proposed method is reduced to 8.7 pm, while the errors of the other three conventional methods based on LSSVR are more than 10.8 pm. Further, when the temperature-changing modes of the training and testing datasets were partly different, the error of the proposed method was reduced to 5.4 pm, while the errors of other methods were more than 11.9 pm. It was verified experimentally that the proposed AWLSSVR method is more accurate and robust than other versions of WLSSVR for training samples with noise, requires no additional hardware, and covers the entire C band.