Yin Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Yin Zhang has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Libo Yuan
Guilin University of Electronic Technology
3 shared publications Kai Zhang
Key Laboratory of In-Fiber Integrated Optics of Ministry of Education, College of Physics, Optoelectronic Engineering, Harbin Engineering University, Harbin, China
2 shared publicationsLunhui Xu
South China University of Technology, Guangzhou, Guangdong CHINA
2 shared publicationsDonghui Wang
Post-doctoral Mobile Station of Optical Engineering, School of Optoelectronic Engineering, Guilin University of Electronic Technology, Guilin, China
1 shared publicationFang Luo
Qingyuan Polytechnic, Qingyuan, Guangdong China
1 shared publication- Hongwei Li
Guilin University of Electronic Technology
1 shared publication
Latest Publications
- 19 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 43 (20) , 9733-9741
Long Short-Term Memory Network Enhances Performance of Optical Fiber Strain Sensors Based on Optical Vernier Effect
High sensitivity and large experiment measurement range are a pair of mutually restricted physical characteristics that restrict the development and application of optical fiber interferometric sensors. In this article, we achieve high sensitivity and a large measurement range with optical fiber strain sensors through Long Short-Term Memory (LSTM) networks, thereby effectively overcoming the problem of free spectral range (FSR) limitation or interference spectrum overlap existing in traditional peak tracking methods. LSTM networks can accurately identify and map the complex relationships between full spectral features and strain, enabling a measurement range that far exceeds approximately 5 times that of a Sagnac interferometer based on the Vernier effect using the traditional peak tracking method. This approach achieves a significantly expanded strain measurement range and improves the stability and reliability of optical fiber sensing. This new method not only maintains high sensitivity and a large measurement range but also avoids complicated and cumbersome data processing in traditional peak tracking methods. More importantly, the whole waveband of the spectrometer can be used, which can theoretically achieve an ultra-large measurement range and even identify interference dips that are difficult to distinguish with the naked eye. This method can further promote the application and development of optical fiber interference sensors.
- 19 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 43 (20) , 9733-9741
Long Short-Term Memory Network Enhances Performance of Optical Fiber Strain Sensors Based on Optical Vernier Effect
High sensitivity and large experiment measurement range are a pair of mutually restricted physical characteristics that restrict the development and application of optical fiber interferometric sensors. In this article, we achieve high sensitivity and a large measurement range with optical fiber strain sensors through Long Short-Term Memory (LSTM) networks, thereby effectively overcoming the problem of free spectral range (FSR) limitation or interference spectrum overlap existing in traditional peak tracking methods. LSTM networks can accurately identify and map the complex relationships between full spectral features and strain, enabling a measurement range that far exceeds approximately 5 times that of a Sagnac interferometer based on the Vernier effect using the traditional peak tracking method. This approach achieves a significantly expanded strain measurement range and improves the stability and reliability of optical fiber sensing. This new method not only maintains high sensitivity and a large measurement range but also avoids complicated and cumbersome data processing in traditional peak tracking methods. More importantly, the whole waveband of the spectrometer can be used, which can theoretically achieve an ultra-large measurement range and even identify interference dips that are difficult to distinguish with the naked eye. This method can further promote the application and development of optical fiber interference sensors.
- 22 December 2024
- Published by Wiley in Biotechnology and Bioengineering
- Vol. 122 (3) , 570-578
Research on Intelligent Monitoring and Concentration Prediction for Penicillin Fermentation Process
In the biopharmaceutical industry, accurately predicting penicillin concentration during fermentation is key to boosting production efficiency and quality assurance. This study leverages the PenSim simulation data set and applies various machine learning and deep learning techniques to forecast penicillin fermentation concentration. Initially, through correlation analysis, nine feature variables with significant impacts on penicillin concentration were screened, and the data underwent preprocessing and standardization. Using grid search, we systematically optimize the hyperparameters of various prediction models. Results show that the ridge regression model excels, achieving a mean squared error of 0.0512 and a mean absolute error of 0.0361. This indicates a strong linear relationship between penicillin concentration and the selected features. Our study offers data‐driven insights for intelligent monitoring and optimization of penicillin fermentation processes. It also showcases the potential of artificial intelligence in enhancing control of biotechnological facilities, paving the way for future research.
- 22 December 2024
- Published by Wiley in Biotechnology and Bioengineering
- Vol. 122 (3) , 570-578
Research on Intelligent Monitoring and Concentration Prediction for Penicillin Fermentation Process
In the biopharmaceutical industry, accurately predicting penicillin concentration during fermentation is key to boosting production efficiency and quality assurance. This study leverages the PenSim simulation data set and applies various machine learning and deep learning techniques to forecast penicillin fermentation concentration. Initially, through correlation analysis, nine feature variables with significant impacts on penicillin concentration were screened, and the data underwent preprocessing and standardization. Using grid search, we systematically optimize the hyperparameters of various prediction models. Results show that the ridge regression model excels, achieving a mean squared error of 0.0512 and a mean absolute error of 0.0361. This indicates a strong linear relationship between penicillin concentration and the selected features. Our study offers data‐driven insights for intelligent monitoring and optimization of penicillin fermentation processes. It also showcases the potential of artificial intelligence in enhancing control of biotechnological facilities, paving the way for future research.
- 18 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Technology Letters
- Vol. 36 (24) , 1397-1400
Precise Temperature Prediction for Small-Sample Fiber Optic Spectra Based on Deep Learning
This study addresses the issues of strong data dependency and limited prediction accuracy in temperature prediction using optical fiber spectroscopy technology by proposing a small-sample optical fiber spectroscopy-based temperature prediction method leveraging deep learning. This method aims to achieve high-precision temperature prediction with limited data samples through the powerful feature extraction and generalization capabilities of deep learning models. To achieve this goal, we first designed a precise experimental protocol to collect optical fiber spectroscopy data covering a temperature range from 30 ° C to 130 ° C, resulting in 84 high-quality data samples under controlled temperature variations. In the data processing stage, advanced signal processing techniques were employed to remove noise and outliers, and the data were normalized to ensure reliability and consistency. Subsequently, various models from the deep learning domain were utilized to train and learn from the processed spectroscopy data. By optimizing the model structures and parameters, we successfully established a nonlinear mapping relationship between spectroscopy and temperature. Experimental results demonstrate that compared to traditional methods, the deep learning model exhibits higher prediction accuracy and stronger robustness in spectroscopic temperature prediction, particularly under small-sample conditions. This study not only provides a novel and effective approach for spectroscopic temperature prediction with limited samples but also expands the application scope of deep learning in spectral analysis. Furthermore, this method holds broad application prospects in various fields such as industrial production, environmental monitoring, and biomedicine, promising to offer more precise and efficient solutions for temperature monitoring and control in related areas.
- 18 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Technology Letters
- Vol. 36 (24) , 1397-1400
Precise Temperature Prediction for Small-Sample Fiber Optic Spectra Based on Deep Learning
This study addresses the issues of strong data dependency and limited prediction accuracy in temperature prediction using optical fiber spectroscopy technology by proposing a small-sample optical fiber spectroscopy-based temperature prediction method leveraging deep learning. This method aims to achieve high-precision temperature prediction with limited data samples through the powerful feature extraction and generalization capabilities of deep learning models. To achieve this goal, we first designed a precise experimental protocol to collect optical fiber spectroscopy data covering a temperature range from 30 ° C to 130 ° C, resulting in 84 high-quality data samples under controlled temperature variations. In the data processing stage, advanced signal processing techniques were employed to remove noise and outliers, and the data were normalized to ensure reliability and consistency. Subsequently, various models from the deep learning domain were utilized to train and learn from the processed spectroscopy data. By optimizing the model structures and parameters, we successfully established a nonlinear mapping relationship between spectroscopy and temperature. Experimental results demonstrate that compared to traditional methods, the deep learning model exhibits higher prediction accuracy and stronger robustness in spectroscopic temperature prediction, particularly under small-sample conditions. This study not only provides a novel and effective approach for spectroscopic temperature prediction with limited samples but also expands the application scope of deep learning in spectral analysis. Furthermore, this method holds broad application prospects in various fields such as industrial production, environmental monitoring, and biomedicine, promising to offer more precise and efficient solutions for temperature monitoring and control in related areas.
- Open Access
- 27 May 2024
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 57 (6) , 1-32
Reliable multiplex semi-local random walk based on influential nodes to improve link prediction in complex networks
In recent years, the exponential growth of online social networks as complex networks has presented challenges in expanding networks and forging new connections. Link prediction emerges as a crucial technique to anticipate future relationships among users, leveraging the current network state to address this challenge effectively. While link prediction models on monoplex networks have a well-established history, the exploration of similar tasks on multilayer networks has garnered considerable attention. Extracting topological and multimodal features for weighting links can improve link prediction in weighted complex networks. Meanwhile, establishing reliable and trustworthy paths between users is a useful way to create metrics that convert unweighted to weighted similarity. The local random walk is a widely used technique for predicting links in weighted monoplex networks. The aim of this paper is to develop a semi-local random walk over reliable paths to improve link prediction on a multilayer social network as a complex network, which is denoted as Reliable Multiplex semi-Local Random Walk (RMLRW). RMLRW leverages the semi-local random walk technique over reliable paths, integrating intra-layer and inter-layer information from multiplex features to conduct a trustworthy biased random walk for predicting new links within a target layer of multilayer networks. In order to make RMLRW scalable, we develop a semi-local random walk-based network embedding to represent the network in a lower-dimensional space while preserving its original characteristics. Extensive experimental studies on several real-world multilayer networks demonstrate the performance assurance of RMLRW compared to equivalent methods. Specifically, RMLRW improves the average f-measure of the link prediction by 3.2% and 2.5% compared to SEM-Path and MLRW, respectively.
- 27 May 2024
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 57 (6) , 1-32
Open AccessReliable multiplex semi-local random walk based on influential nodes to improve link prediction in complex networks
In recent years, the exponential growth of online social networks as complex networks has presented challenges in expanding networks and forging new connections. Link prediction emerges as a crucial technique to anticipate future relationships among users, leveraging the current network state to address this challenge effectively. While link prediction models on monoplex networks have a well-established history, the exploration of similar tasks on multilayer networks has garnered considerable attention. Extracting topological and multimodal features for weighting links can improve link prediction in weighted complex networks. Meanwhile, establishing reliable and trustworthy paths between users is a useful way to create metrics that convert unweighted to weighted similarity. The local random walk is a widely used technique for predicting links in weighted monoplex networks. The aim of this paper is to develop a semi-local random walk over reliable paths to improve link prediction on a multilayer social network as a complex network, which is denoted as Reliable Multiplex semi-Local Random Walk (RMLRW). RMLRW leverages the semi-local random walk technique over reliable paths, integrating intra-layer and inter-layer information from multiplex features to conduct a trustworthy biased random walk for predicting new links within a target layer of multilayer networks. In order to make RMLRW scalable, we develop a semi-local random walk-based network embedding to represent the network in a lower-dimensional space while preserving its original characteristics. Extensive experimental studies on several real-world multilayer networks demonstrate the performance assurance of RMLRW compared to equivalent methods. Specifically, RMLRW improves the average f-measure of the link prediction by 3.2% and 2.5% compared to SEM-Path and MLRW, respectively.
- 24 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (15) , 5428-5434
A Frequency-Scanned Φ-OTDR Implemented by a Frequency Shifting Loop Used for Distributed Strain Sensing
In this paper, a frequency-scanned Φ-OTDR based on a frequency shifting loop (FSL) is proposed. The FSL is a fiber optic loop containing an acousto-optic frequency shifter (AOFS) and a fiber optic amplifier that converts a single-frequency pulse into a train of pulses with a fixed time and frequency spacing. By introducing bias light and precise switching control the AOFS, the FSL outputs low ASE noise pulses with a frequency spacing of 20 MHz and a scanning range of 460 MHz, which are used as the probe light for the frequency-scanned Φ-OTDR. In addition, a real-time update of the reference Rayleigh scattering pattern (RSP) is proposed to extend the strain measurement range. Experiments on a 3.96-km fiber shows that the strain ASD noise is better than 0.4 nϵ/Hz 1/2 , and approximate white noise at frequencies higher than 2 Hz with a spatial resolution better than 2.4 m. This frequency-scanned Φ-OTDR is suitable for low-frequency and quasi-static distributed strain measurements, and is expected to be applied to microseismic, teleseismic, and structural health monitoring.
- 24 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 42 (15) , 5428-5434
A Frequency-Scanned Φ-OTDR Implemented by a Frequency Shifting Loop Used for Distributed Strain Sensing
In this paper, a frequency-scanned Φ-OTDR based on a frequency shifting loop (FSL) is proposed. The FSL is a fiber optic loop containing an acousto-optic frequency shifter (AOFS) and a fiber optic amplifier that converts a single-frequency pulse into a train of pulses with a fixed time and frequency spacing. By introducing bias light and precise switching control the AOFS, the FSL outputs low ASE noise pulses with a frequency spacing of 20 MHz and a scanning range of 460 MHz, which are used as the probe light for the frequency-scanned Φ-OTDR. In addition, a real-time update of the reference Rayleigh scattering pattern (RSP) is proposed to extend the strain measurement range. Experiments on a 3.96-km fiber shows that the strain ASD noise is better than 0.4 nϵ/Hz 1/2 , and approximate white noise at frequencies higher than 2 Hz with a spatial resolution better than 2.4 m. This frequency-scanned Φ-OTDR is suitable for low-frequency and quasi-static distributed strain measurements, and is expected to be applied to microseismic, teleseismic, and structural health monitoring.
- Open Access
- 17 January 2024
- Published by SAGE Publications in Measurement and Control
- Vol. 57 (6) , 751-762
Mask wearing detection algorithm based on improved YOLOv7
The ongoing COVID-19 pandemic remains a significant threat, emphasizing the critical importance of mask-wearing to reduce infection risks. However, existing methods for mask detection encounter challenges such as identifying small targets and achieving high accuracy. In this paper, we present an enhanced YOLOv7 model tailored for mask-wearing detection. we employing a Generative Adversarial Network (GAN) to augment the original dataset, introducing the Convolutional Block Attention Module (CBAM) mechanism into the YOLOv7 model to enhance its small target detection capabilities, and replacing the model’s activation function with Parametric Rectified Linear Unit (FReLU) to improve overall performance. Experimental validation on a dataset showcases an average precision of 97.8% and a real-time inference speed of 64 frames per second (fps), meeting the real-time mask-wearing detection requirements effectively.
- 17 January 2024
- Published by SAGE Publications in Measurement and Control
- Vol. 57 (6) , 751-762
Open AccessMask wearing detection algorithm based on improved YOLOv7
The ongoing COVID-19 pandemic remains a significant threat, emphasizing the critical importance of mask-wearing to reduce infection risks. However, existing methods for mask detection encounter challenges such as identifying small targets and achieving high accuracy. In this paper, we present an enhanced YOLOv7 model tailored for mask-wearing detection. we employing a Generative Adversarial Network (GAN) to augment the original dataset, introducing the Convolutional Block Attention Module (CBAM) mechanism into the YOLOv7 model to enhance its small target detection capabilities, and replacing the model’s activation function with Parametric Rectified Linear Unit (FReLU) to improve overall performance. Experimental validation on a dataset showcases an average precision of 97.8% and a real-time inference speed of 64 frames per second (fps), meeting the real-time mask-wearing detection requirements effectively.
- Open Access
- 20 August 2022
- Published by SAGE Publications in Measurement and Control
- Vol. 55 (9-10) , 1088-1096
Research on hierarchical pedestrian detection based on SVM classifier with improved kernel function
The research of pedestrian target detection in complex scenes is still of great significance. Aiming at the problem of high missed detection rate and poor timeliness of pedestrian target detection in complex scenes. This paper proposes an improved classification method. First, Haar features were extracted from the images to be detected, and the candidate areas of pedestrians were determined by Adaboost classifier. Then, the traditional SVM classifier was improved by using the combined kernel function instead of the single kernel function, and the optimal proportion of each function in the combined kernel function was found by using the adaptive particle swarm optimization algorithm. Finally, the improved SVM classifier was combined with the fusion feature to further detect the candidate area to accurately locate the pedestrian’s position. Experimental results show that compared with the traditional detection framework, the proposed method can effectively improve the detection speed and the detection accuracy. This method has certain practical significance for pedestrian target detection in complex scenes.
- 20 August 2022
- Published by SAGE Publications in Measurement and Control
- Vol. 55 (9-10) , 1088-1096
Open AccessResearch on hierarchical pedestrian detection based on SVM classifier with improved kernel function
The research of pedestrian target detection in complex scenes is still of great significance. Aiming at the problem of high missed detection rate and poor timeliness of pedestrian target detection in complex scenes. This paper proposes an improved classification method. First, Haar features were extracted from the images to be detected, and the candidate areas of pedestrians were determined by Adaboost classifier. Then, the traditional SVM classifier was improved by using the combined kernel function instead of the single kernel function, and the optimal proportion of each function in the combined kernel function was found by using the adaptive particle swarm optimization algorithm. Finally, the improved SVM classifier was combined with the fusion feature to further detect the candidate area to accurately locate the pedestrian’s position. Experimental results show that compared with the traditional detection framework, the proposed method can effectively improve the detection speed and the detection accuracy. This method has certain practical significance for pedestrian target detection in complex scenes.