Pub Date : 2025-02-20DOI: 10.1109/TIM.2024.3502719
Lin Jiao;Haiyun Liu;Zheng Liang;Peng Chen;Rujing Wang;Kang Liu
The number of wheat spikes is a crucial index for evaluating the yield, and the precise detection of wheat spikes in an image plays an important role. Among various methods, deep learning-based approaches show impressive results in the task of wheat spike detection. However, the precise detection and recognition of wheat spike encounters large challenges due to complicated backgrounds, arbitrary orientations, and dense distribution in wheat spike images. To alleviate these issues, we have developed an anchor-free refining feature pyramid network (AFRFPN) that gets rid of horizontal bounding boxes (HBBs) from the network. First, the refining feature pyramid network (RFPN) has been introduced into extract richer features of wheat spike with highly variant appearances and multiple scales. Then, learning from the idea of coarse-to-fine, the two-stage anchor-free oriented detection (AFOD) module has been designed. The AFOD module first generates a set of coarse detection (CoDet) results in the way of anchor-free and then further fines them to achieve high-quality predicting bounding boxes (BBs). The number of wheat spike images is insufficient, resulting in poor performance of wheat spike detection modules. To mitigate the lack of the data in the task of oriented wheat spike detection, based on the global wheat head detection (GWHD) dataset, we released a new large-scale wheat spike dataset by relabeling the samples, termed it as rotated GWHD (R-GWHD) dataset. Massive experiments show that the proposed method can achieve 90.6% mAP and 96.7% recall, outperforming other state-of-the-art methods. Additionally, the experiments related to the counting of wheat spikes have been conducted, showing that the developed module can achieve the MAE of 4.95 and RMSE of 7.68, which demonstrates the excellent performance of the proposed method.
{"title":"An Anchor-Free Refining Feature Pyramid Network for Dense and Multioriented Wheat Spikes Detection Under UAV","authors":"Lin Jiao;Haiyun Liu;Zheng Liang;Peng Chen;Rujing Wang;Kang Liu","doi":"10.1109/TIM.2024.3502719","DOIUrl":"https://doi.org/10.1109/TIM.2024.3502719","url":null,"abstract":"The number of wheat spikes is a crucial index for evaluating the yield, and the precise detection of wheat spikes in an image plays an important role. Among various methods, deep learning-based approaches show impressive results in the task of wheat spike detection. However, the precise detection and recognition of wheat spike encounters large challenges due to complicated backgrounds, arbitrary orientations, and dense distribution in wheat spike images. To alleviate these issues, we have developed an anchor-free refining feature pyramid network (AFRFPN) that gets rid of horizontal bounding boxes (HBBs) from the network. First, the refining feature pyramid network (RFPN) has been introduced into extract richer features of wheat spike with highly variant appearances and multiple scales. Then, learning from the idea of coarse-to-fine, the two-stage anchor-free oriented detection (AFOD) module has been designed. The AFOD module first generates a set of coarse detection (CoDet) results in the way of anchor-free and then further fines them to achieve high-quality predicting bounding boxes (BBs). The number of wheat spike images is insufficient, resulting in poor performance of wheat spike detection modules. To mitigate the lack of the data in the task of oriented wheat spike detection, based on the global wheat head detection (GWHD) dataset, we released a new large-scale wheat spike dataset by relabeling the samples, termed it as rotated GWHD (R-GWHD) dataset. Massive experiments show that the proposed method can achieve 90.6% mAP and 96.7% recall, outperforming other state-of-the-art methods. Additionally, the experiments related to the counting of wheat spikes have been conducted, showing that the developed module can achieve the MAE of 4.95 and RMSE of 7.68, which demonstrates the excellent performance of the proposed method.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-14"},"PeriodicalIF":5.6,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143471109","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1109/TIM.2025.3541694
Zhen Jia;Helei Dong;Yong Ruan;Yu Wang;Yongqing Zhang;Jie Ma;Yuxin Miao;Qiulin Tan
Laser heating has become a common means of dynamic response testing, but accurate measurement of the dynamic response time poses a challenge, mainly due to the interference from the thermophysical parameters of K-type thin-film thermocouple, temperature measurement environments, and stray signals. To address this issue, this article designs a dynamic response testing platform for thermocouples based on lasers and constructs a WOA-BP algorithm model. This model aims to accurately predict the dynamic response time and output peak voltage of the thermocouple, providing guidance for parameter optimization during the experimental process and ensuring efficient capture of the thermocouple’s dynamic response signals. Meanwhile, this article compared the WOA-BP algorithmic model with back propagation (BP) and other optimized BP models, evaluated by RMSE, MAE, and R2. The results demonstrate that with the guidance of parameter optimization by the WOA-BP algorithm model, the dynamic testing system is capable of accurately performing dynamic performance tests on thermocouples. Besides, the dynamic response time is inversely proportional to laser power and directly proportional to laser pulsewidth, but independent of repetition frequency. The output peak voltage increases with the increase of laser power and pulsewidth, but is also independent of laser repetition frequency. And the WOA-BP algorithm model can predict dynamic response time and output peak voltage accurately, whose R2 values of dynamic response time and output peak voltage are 0.9954 and 0.9982, the RMSE values are 0.5766 and 0.2981, and the MAE values are 0.4152 and 0.2700, respectively, being the best compared with BP, PSO-BP, GA-BP, MFO-BP, and GWO-BP prediction models.
{"title":"Dynamic Response Testing Based on Pulsed Laser and WOA-BP Neural Network","authors":"Zhen Jia;Helei Dong;Yong Ruan;Yu Wang;Yongqing Zhang;Jie Ma;Yuxin Miao;Qiulin Tan","doi":"10.1109/TIM.2025.3541694","DOIUrl":"https://doi.org/10.1109/TIM.2025.3541694","url":null,"abstract":"Laser heating has become a common means of dynamic response testing, but accurate measurement of the dynamic response time poses a challenge, mainly due to the interference from the thermophysical parameters of K-type thin-film thermocouple, temperature measurement environments, and stray signals. To address this issue, this article designs a dynamic response testing platform for thermocouples based on lasers and constructs a WOA-BP algorithm model. This model aims to accurately predict the dynamic response time and output peak voltage of the thermocouple, providing guidance for parameter optimization during the experimental process and ensuring efficient capture of the thermocouple’s dynamic response signals. Meanwhile, this article compared the WOA-BP algorithmic model with back propagation (BP) and other optimized BP models, evaluated by RMSE, MAE, and R2. The results demonstrate that with the guidance of parameter optimization by the WOA-BP algorithm model, the dynamic testing system is capable of accurately performing dynamic performance tests on thermocouples. Besides, the dynamic response time is inversely proportional to laser power and directly proportional to laser pulsewidth, but independent of repetition frequency. The output peak voltage increases with the increase of laser power and pulsewidth, but is also independent of laser repetition frequency. And the WOA-BP algorithm model can predict dynamic response time and output peak voltage accurately, whose R2 values of dynamic response time and output peak voltage are 0.9954 and 0.9982, the RMSE values are 0.5766 and 0.2981, and the MAE values are 0.4152 and 0.2700, respectively, being the best compared with BP, PSO-BP, GA-BP, MFO-BP, and GWO-BP prediction models.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-12"},"PeriodicalIF":5.6,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143489111","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1109/TIM.2025.3529055
Jianqing Peng;Weihao Li;Lin Liu;Wanquan Liu;Yu Han
Adaptive grasping (AG) of objects with unknown shapes and contact force sensing during grasping are two critical issues in object grasping. However, most of the current grippers that rely on passive grasping mechanisms do not consider both of these functions at the same time. In this article, we design a multibranch compliant flexible gripper (MBCFG), propose a method for intrinsic force perception and extrinsic deformation measurement without tactile sensor measurement, and finally achieve stable grasping of irregular objects and “force–deformation” sensing in real time. Furthermore, based on the above perceptual model, the envelope degree of grasping and the grasping force constraint model are established, which in turn indicates the evaluation method of grasping quality (GQ). Experimental results show that the flexible gripper (FG) has high sensing accuracy (i.e., its average error is less than 0.78 mm) for the external force applied at the two end nodes, but the error occurs much more when the external force is applied at its middle node (i.e., its average error is about 2.76 mm). In addition, the FG assembled at the end-effector of the robot successfully grasped a variety of objects with different shapes. Besides, the experiments of pick-and-place operation showed that the grasping performance based on “force–deformation” perception was good. Moreover, the fitted envelope curve and the shape curve of the FG also matched very well, which further proved the AG and contact force perception capability of the FG for objects with unknown shapes.
{"title":"Design, Perceptual Modeling, and Grasping Performance Evaluation of Multibranch Flexible Grippers","authors":"Jianqing Peng;Weihao Li;Lin Liu;Wanquan Liu;Yu Han","doi":"10.1109/TIM.2025.3529055","DOIUrl":"https://doi.org/10.1109/TIM.2025.3529055","url":null,"abstract":"Adaptive grasping (AG) of objects with unknown shapes and contact force sensing during grasping are two critical issues in object grasping. However, most of the current grippers that rely on passive grasping mechanisms do not consider both of these functions at the same time. In this article, we design a multibranch compliant flexible gripper (MBCFG), propose a method for intrinsic force perception and extrinsic deformation measurement without tactile sensor measurement, and finally achieve stable grasping of irregular objects and “force–deformation” sensing in real time. Furthermore, based on the above perceptual model, the envelope degree of grasping and the grasping force constraint model are established, which in turn indicates the evaluation method of grasping quality (GQ). Experimental results show that the flexible gripper (FG) has high sensing accuracy (i.e., its average error is less than 0.78 mm) for the external force applied at the two end nodes, but the error occurs much more when the external force is applied at its middle node (i.e., its average error is about 2.76 mm). In addition, the FG assembled at the end-effector of the robot successfully grasped a variety of objects with different shapes. Besides, the experiments of pick-and-place operation showed that the grasping performance based on “force–deformation” perception was good. Moreover, the fitted envelope curve and the shape curve of the FG also matched very well, which further proved the AG and contact force perception capability of the FG for objects with unknown shapes.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-18"},"PeriodicalIF":5.6,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143455347","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-20DOI: 10.1109/TIM.2025.3539442
Kristen M. Donnell;Paweł Niewczas;Mohamed Abou-Khousa;Teddy Surya Gunawan
{"title":"Guest Editorial Special Section on 2023 IEEE International Instrumentation and Measurement Technology Conference","authors":"Kristen M. Donnell;Paweł Niewczas;Mohamed Abou-Khousa;Teddy Surya Gunawan","doi":"10.1109/TIM.2025.3539442","DOIUrl":"https://doi.org/10.1109/TIM.2025.3539442","url":null,"abstract":"","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"73 ","pages":"1-3"},"PeriodicalIF":5.6,"publicationDate":"2025-02-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10896957","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143455284","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Cell/particle manipulation has made significant strides in fields such as biomedical research, chemical analysis, and materials science, emerging as a crucial discipline. This article introduces an innovative microchip featuring a wavy floating electrode structure, capable of multimode cell/particle manipulation. The floating-electrode-based microchip enabling different modal manipulation of cells/particles can be realized by simply adjusting the voltage and frequency parameters of the driving signal. Vortices can be generated at the edges of floating electrode in the low-frequency range of the driving signal, orbital revolution of popular cells and out-of-plane rotation of single cells can be achieved by adjusting the voltage amplitude. The orbital revolution can be used for the enrichment of cells/particles, while the out-of-plane rotation provides the possibility to reconstruct the 3-D model of single cells and measure the physical parameters (ellipticity, surface area, and volume). As the frequency increases, the dielectrophoretic forces gradually become dominant, enabling the effective separation of cells/particles with different electrical properties. The separation feasibility of the microchip was verified by the effective separation of particles and yeast cells. Through numerical simulation and experiments, the microchip is proved to have the advantages of simple structure, convenient operation, and multimodal manipulation of cells/particles.
{"title":"Floating-Electrode-Based Microchip Enabling Multimodal Manipulation and Physical Parameters Measurement of Cells/Particles","authors":"Liang Huang;Wenru Dai;Jingui Qian;Yongqing Wei;Jin Zhang;Haojie Xia","doi":"10.1109/TIM.2025.3541705","DOIUrl":"https://doi.org/10.1109/TIM.2025.3541705","url":null,"abstract":"Cell/particle manipulation has made significant strides in fields such as biomedical research, chemical analysis, and materials science, emerging as a crucial discipline. This article introduces an innovative microchip featuring a wavy floating electrode structure, capable of multimode cell/particle manipulation. The floating-electrode-based microchip enabling different modal manipulation of cells/particles can be realized by simply adjusting the voltage and frequency parameters of the driving signal. Vortices can be generated at the edges of floating electrode in the low-frequency range of the driving signal, orbital revolution of popular cells and out-of-plane rotation of single cells can be achieved by adjusting the voltage amplitude. The orbital revolution can be used for the enrichment of cells/particles, while the out-of-plane rotation provides the possibility to reconstruct the 3-D model of single cells and measure the physical parameters (ellipticity, surface area, and volume). As the frequency increases, the dielectrophoretic forces gradually become dominant, enabling the effective separation of cells/particles with different electrical properties. The separation feasibility of the microchip was verified by the effective separation of particles and yeast cells. Through numerical simulation and experiments, the microchip is proved to have the advantages of simple structure, convenient operation, and multimodal manipulation of cells/particles.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-9"},"PeriodicalIF":5.6,"publicationDate":"2025-02-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143489161","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Fast-neutron-energy measurements are essential for nuclear particle research and dosimetry. However, such measurements are challenging due to the low interaction probability of fast neutrons with the matter, given their small cross section for scattering and absorption compared to thermal neutrons, as well as their zero charge. Traditional neutron-energy measurement methods have limitations related to the distance and detector size. Therefore, this study proposes a novel kinematic neutron-energy compensation method that measures neutron scattering using the first detector and directly captures fast neutrons using the second detector. The scattering and post-scattering energies of neutrons are measured and used to reconstruct the neutron energy, enabling more accurate measurements of high-energy monoenergetic neutrons. This method is less sensitive to the energy and angle distribution of neutrons as they interact with the first detector and scatter toward the second detector. The energy deposited in the first detector is measured, while the scattered fast-neutron energy is determined through nuclear reactions within the second detector, enabling event-by-event compensation in energy reconstruction. Therefore, the system is not sensitive to the distance between detectors or the solid angle determined by the detector size. The performance of the system is verified using EJ-309 liquid and 7Li-enriched Cs2LiYCl6:Ce3+ scintillators. In addition, at Korea Research Institute of Standards and Science (KRISS), 14.8-MeV monoenergetic neutrons were used to characterize the proposed method, achieving an energy resolution of 2.8% full-width at half-maximum (FWHM) for measurements and energy reconstruction.
{"title":"A Novel Method for Kinematic Neutron-Energy Compensation in Fast-Neutron Spectroscopy","authors":"HyeoungWoo Park;Sinchul Kang;Young Soo Yoon;Hyeonseo Park;Jungho Kim;Joong Hyun Kim;Hong Joo Kim;DongWoo Jeong;Jin Jegal","doi":"10.1109/TIM.2025.3541752","DOIUrl":"https://doi.org/10.1109/TIM.2025.3541752","url":null,"abstract":"Fast-neutron-energy measurements are essential for nuclear particle research and dosimetry. However, such measurements are challenging due to the low interaction probability of fast neutrons with the matter, given their small cross section for scattering and absorption compared to thermal neutrons, as well as their zero charge. Traditional neutron-energy measurement methods have limitations related to the distance and detector size. Therefore, this study proposes a novel kinematic neutron-energy compensation method that measures neutron scattering using the first detector and directly captures fast neutrons using the second detector. The scattering and post-scattering energies of neutrons are measured and used to reconstruct the neutron energy, enabling more accurate measurements of high-energy monoenergetic neutrons. This method is less sensitive to the energy and angle distribution of neutrons as they interact with the first detector and scatter toward the second detector. The energy deposited in the first detector is measured, while the scattered fast-neutron energy is determined through nuclear reactions within the second detector, enabling event-by-event compensation in energy reconstruction. Therefore, the system is not sensitive to the distance between detectors or the solid angle determined by the detector size. The performance of the system is verified using EJ-309 liquid and 7Li-enriched Cs2LiYCl6:Ce3+ scintillators. In addition, at Korea Research Institute of Standards and Science (KRISS), 14.8-MeV monoenergetic neutrons were used to characterize the proposed method, achieving an energy resolution of 2.8% full-width at half-maximum (FWHM) for measurements and energy reconstruction.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-8"},"PeriodicalIF":5.6,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143465731","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-17DOI: 10.1109/TIM.2025.3540121
Yifan Si;Shuo Li;Xiaodong Wang;Sailing He
Computed tomography imaging spectrometry (CTIS) is a snapshot hyperspectral imaging (HSI) technique capable of capturing projections of the target scene from multiple wavelengths in one single exposure. The CTIS inversion problem is usually very challenging, and solving it from a single snapshot measurement often requires time-consuming iterative algorithms. And most deep learning-based algorithms in computational imaging need the priori of many samples, which brings a heavy data collection burden. In this article, to reconstruct hyperspectral cubes from CTIS measurements in an efficient way, we introduce a new CITS framework named ASP-Model based on the angular spectrum propagation theory to model the forward CITS process and efficiently reconstruct hyperspectral. Specifically, our method acquires simulation data using angular spectrum propagation for training and reconstructs real data captured by our custom-built CTIS system during inference. This framework allows us to eliminate the need to acquire extensive real data for network training. Moreover, the proposed network can reconstruct 26 spectral channels from one single measurement and demonstrates state-of-the-art results over existing reconstruction algorithms both in simulation and experimental results. We also release a new dataset containing simulated and real CTIS data for public comparison. The code and dataset are publicly available at https://github.com/YifanSi/ASP_Model.
{"title":"ASP-Model: An Advanced Deep Learning Framework to Reconstruct Hyperspectral Cubes for Computed Tomography Imaging System","authors":"Yifan Si;Shuo Li;Xiaodong Wang;Sailing He","doi":"10.1109/TIM.2025.3540121","DOIUrl":"https://doi.org/10.1109/TIM.2025.3540121","url":null,"abstract":"Computed tomography imaging spectrometry (CTIS) is a snapshot hyperspectral imaging (HSI) technique capable of capturing projections of the target scene from multiple wavelengths in one single exposure. The CTIS inversion problem is usually very challenging, and solving it from a single snapshot measurement often requires time-consuming iterative algorithms. And most deep learning-based algorithms in computational imaging need the priori of many samples, which brings a heavy data collection burden. In this article, to reconstruct hyperspectral cubes from CTIS measurements in an efficient way, we introduce a new CITS framework named ASP-Model based on the angular spectrum propagation theory to model the forward CITS process and efficiently reconstruct hyperspectral. Specifically, our method acquires simulation data using angular spectrum propagation for training and reconstructs real data captured by our custom-built CTIS system during inference. This framework allows us to eliminate the need to acquire extensive real data for network training. Moreover, the proposed network can reconstruct 26 spectral channels from one single measurement and demonstrates state-of-the-art results over existing reconstruction algorithms both in simulation and experimental results. We also release a new dataset containing simulated and real CTIS data for public comparison. The code and dataset are publicly available at <uri>https://github.com/YifanSi/ASP_Model</uri>.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-10"},"PeriodicalIF":5.6,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143471121","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Transfer learning (TL) has garnered significant interest in mechanical fault diagnosis. Many current TL approaches typically assume that ample data are available and that both the source and target domains possess identical label spaces. However, these TL methods often fail to address real-world issues, particularly when the number of samples in different conditions is unequal (i.e., imbalance) and the target label space is a subset of the source label space [i.e., partial transfer learning (PTL)]. To address these issues, this study proposes the imbalanced partial transfer network (IPTN). The IPTN introduces a weighted maximum density divergence (MDD) loss and a discriminative sample generator (DSG). The DSG identifies distinctive samples in the target domain and expands the dataset by augmenting these distinctive samples to solve the sample imbalance problem. Meanwhile, the new loss function termed weighted MDD promotes the ability of PTL by increasing interclass distance and intraclass density. Experiments on two datasets demonstrate the superior diagnostic performance of the IPTN compared to several comparison methods, highlighting its powerful transfer capability in situations involving sample imbalance and PTL.
{"title":"Imbalanced Partial Transfer Network for Intelligent Machine Fault Diagnosis","authors":"Chuancang Ding;Yanlin Zhou;Xuyan Liu;Baoxiang Wang;Weiguo Huang;Zhongkui Zhu","doi":"10.1109/TIM.2025.3542138","DOIUrl":"https://doi.org/10.1109/TIM.2025.3542138","url":null,"abstract":"Transfer learning (TL) has garnered significant interest in mechanical fault diagnosis. Many current TL approaches typically assume that ample data are available and that both the source and target domains possess identical label spaces. However, these TL methods often fail to address real-world issues, particularly when the number of samples in different conditions is unequal (i.e., imbalance) and the target label space is a subset of the source label space [i.e., partial transfer learning (PTL)]. To address these issues, this study proposes the imbalanced partial transfer network (IPTN). The IPTN introduces a weighted maximum density divergence (MDD) loss and a discriminative sample generator (DSG). The DSG identifies distinctive samples in the target domain and expands the dataset by augmenting these distinctive samples to solve the sample imbalance problem. Meanwhile, the new loss function termed weighted MDD promotes the ability of PTL by increasing interclass distance and intraclass density. Experiments on two datasets demonstrate the superior diagnostic performance of the IPTN compared to several comparison methods, highlighting its powerful transfer capability in situations involving sample imbalance and PTL.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-11"},"PeriodicalIF":5.6,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143489239","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1109/TIM.2025.3542111
Chuanxu Chen;Quansheng Guan;Quanxue Guan;Xin Jin;Zhan Shi
Locations and parameters monitoring for soft faults in a cable network is of importance to prevent hard faults at an early stage and maintain the stability of the power system. The existing fault detection methods often identify faults using the fixed parameters in a reference cable model to locate the faults. However, the changes of the reference model by faults bring model mismatch, for example, the signal propagation speed is different in the line-like soft faults. The model mismatch will lead to inaccurate fault location and not to mention parameter imaging for faults. To this end, this article proposes a residual voltage inversion (RVI) method to learn the model of the cable network with unknown faults. RVI uses the residual voltages, that is, the difference between the scattering voltages measured at the ports and those generated by the current model, as the gradient to update the multiple parameter distributions of the cable network iteratively. The learned model can then be used to calculate the precise location, imaging of the length, capacitance, and resistance for line-like soft faults. The simulation results show that RVI locates the range of line-like soft faults with an accuracy over 90%, and achieves insulation layer and core conductor imaging with accuracies larger than 95% and 80%, respectively. In addition, the experimental tests are carried out to verify the feasibility and performance of RVI.
{"title":"Soft Fault Location and Imaging Using Residual Voltage Inversion in Cable Networks","authors":"Chuanxu Chen;Quansheng Guan;Quanxue Guan;Xin Jin;Zhan Shi","doi":"10.1109/TIM.2025.3542111","DOIUrl":"https://doi.org/10.1109/TIM.2025.3542111","url":null,"abstract":"Locations and parameters monitoring for soft faults in a cable network is of importance to prevent hard faults at an early stage and maintain the stability of the power system. The existing fault detection methods often identify faults using the fixed parameters in a reference cable model to locate the faults. However, the changes of the reference model by faults bring model mismatch, for example, the signal propagation speed is different in the line-like soft faults. The model mismatch will lead to inaccurate fault location and not to mention parameter imaging for faults. To this end, this article proposes a residual voltage inversion (RVI) method to learn the model of the cable network with unknown faults. RVI uses the residual voltages, that is, the difference between the scattering voltages measured at the ports and those generated by the current model, as the gradient to update the multiple parameter distributions of the cable network iteratively. The learned model can then be used to calculate the precise location, imaging of the length, capacitance, and resistance for line-like soft faults. The simulation results show that RVI locates the range of line-like soft faults with an accuracy over 90%, and achieves insulation layer and core conductor imaging with accuracies larger than 95% and 80%, respectively. In addition, the experimental tests are carried out to verify the feasibility and performance of RVI.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-16"},"PeriodicalIF":5.6,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143489236","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1109/TIM.2025.3537644
Emiliano Schena;Lorenzo Scalise
{"title":"Guest Editorial Special section on IEEE MeMeA 2023","authors":"Emiliano Schena;Lorenzo Scalise","doi":"10.1109/TIM.2025.3537644","DOIUrl":"https://doi.org/10.1109/TIM.2025.3537644","url":null,"abstract":"","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"73 ","pages":"1-2"},"PeriodicalIF":5.6,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10887366","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143422794","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：