Pub Date : 2026-02-09DOI: 10.1016/j.measurement.2026.120771
Zhen Yang , Menggang Kang , Zekun Hao , Lei Chen , Yun Xu , Hua Yang
Volumetric flow measurement techniques based on particle image velocimetry (PIV) have demonstrated significant potential for quantitatively capturing unsteady flow characteristics in various experimental applications of fluid mechanics. Despite the advancements achieved in tomographic PIV, the development of particle field reconstruction techniques remains challenging due to issues such as ghost particles, elongated particles, and high computational costs. This work introduces a method that employs Gaussian splatting (GS) to accurately determine the three-dimensional (3D) spatial distribution of particles. The method models particles using 3D Gaussians and applies differentiable Gaussian rasterization to render synthetic particle images for loss computation. By optimizing the Gaussian parameters via gradient backpropagation, the approach yields an accurate 3D particle field. Experimental evaluations demonstrate that the proposed method effectively reconstructs the 3D particle distribution. We believe that the improved accuracy and generalizability of this approach will significantly advance PIV technology.
{"title":"A novel Gaussian splatting-based particle field reconstruction method for tomographic particle image velocimetry","authors":"Zhen Yang , Menggang Kang , Zekun Hao , Lei Chen , Yun Xu , Hua Yang","doi":"10.1016/j.measurement.2026.120771","DOIUrl":"10.1016/j.measurement.2026.120771","url":null,"abstract":"<div><div>Volumetric flow measurement techniques based on particle image velocimetry (PIV) have demonstrated significant potential for quantitatively capturing unsteady flow characteristics in various experimental applications of fluid mechanics. Despite the advancements achieved in tomographic PIV, the development of particle field reconstruction techniques remains challenging due to issues such as ghost particles, elongated particles, and high computational costs. This work introduces a method that employs Gaussian splatting (GS) to accurately determine the three-dimensional (3D) spatial distribution of particles. The method models particles using 3D Gaussians and applies differentiable Gaussian rasterization to render synthetic particle images for loss computation. By optimizing the Gaussian parameters via gradient backpropagation, the approach yields an accurate 3D particle field. Experimental evaluations demonstrate that the proposed method effectively reconstructs the 3D particle distribution. We believe that the improved accuracy and generalizability of this approach will significantly advance PIV technology.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120771"},"PeriodicalIF":5.6,"publicationDate":"2026-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147479","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 : 2026-02-09DOI: 10.1016/j.measurement.2026.120766
Xuehua Zhu , Ziruo Ren , Xinyu Liu , Shuanglong Yang , Keneng Xu , Yawen Zhu
Optical single sideband (OSSB) signals are widely used in optical communication systems owing to their effective anti-dispersion capability and high spectral efficiency. To address the challenge of high efficiency generation of OSSB signals, we propose a method based on multi-harmonic phase modulation (MHPM) with the mountain gazelle optimizer (MGO). By constructing a multi-objective optimization model, synergistic optimization of modulation depth, phase, and harmonic order was achieved, followed by simulation and experimental investigation. Simulation results indicate that at the 10th harmonic order, the energy efficiency (EF) of the OSSB signal, defined as the ratio of target sideband optical power to total output optical power, including the carrier, reached 88.04 %. This corresponds to an improvement of more than 38 % over the traditional optical double sideband (ODSB) scheme and yielding a high carrier suppression ratio (CSR) of 23.69 dB and sideband suppression ratio (SSR) of 18.53 dB. In the experimental stage, a narrow-linewidth fiber laser and an electro-optic phase modulator were employed to perform 10th-harmonic phase modulation, yielding an OSSB signal with an EF of approximately 80 %. The experimental results showed good agreement with the simulations, confirming that the intelligent algorithm can effectively address the multi-objective optimization challenges introduced by MHPM. This work provides a novel approach for complex multi-parameter coupling optimization in phase modulation systems and offers new insights for efficiently handling related problems in the fields of optical fiber communications and microwave photonics.
{"title":"Generation and optimization of optical single-sideband signal using multi-harmonic phase modulation","authors":"Xuehua Zhu , Ziruo Ren , Xinyu Liu , Shuanglong Yang , Keneng Xu , Yawen Zhu","doi":"10.1016/j.measurement.2026.120766","DOIUrl":"10.1016/j.measurement.2026.120766","url":null,"abstract":"<div><div>Optical single sideband (OSSB) signals are widely used in optical communication systems owing to their effective anti-dispersion capability and high spectral efficiency. To address the challenge of high efficiency generation of OSSB signals, we propose a method based on multi-harmonic phase modulation (MHPM) with the mountain gazelle optimizer (MGO). By constructing a multi-objective optimization model, synergistic optimization of modulation depth, phase, and harmonic order was achieved, followed by simulation and experimental investigation. Simulation results indicate that at the 10th harmonic order, the energy efficiency (EF) of the OSSB signal, defined as the ratio of target sideband optical power to total output optical power, including the carrier, reached 88.04 %. This corresponds to an improvement of more than 38 % over the traditional optical double sideband (ODSB) scheme and yielding a high carrier suppression ratio (CSR) of 23.69 dB and sideband suppression ratio (SSR) of 18.53 dB. In the experimental stage, a narrow-linewidth fiber laser and an electro-optic phase modulator were employed to perform 10th-harmonic phase modulation, yielding an OSSB signal with an EF of approximately 80 %. The experimental results showed good agreement with the simulations, confirming that the intelligent algorithm can effectively address the multi-objective optimization challenges introduced by MHPM. This work provides a novel approach for complex multi-parameter coupling optimization in phase modulation systems and offers new insights for efficiently handling related problems in the fields of optical fiber communications and microwave photonics.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120766"},"PeriodicalIF":5.6,"publicationDate":"2026-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147505","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 : 2026-02-09DOI: 10.1016/j.measurement.2026.120784
Yongxin Cui , Zheng Dou , Yabin Zhang
Accurate measurement of Extremely Low Frequency (ELF) electromagnetic (EM) fields in seawater is essential for oceanographic sensing, geophysical monitoring, and the calibration of undersea EM sensors. However, the lack of real-time, traceable measurement models limits the precision of in-situ sensor calibration, as exact Sommerfeld integrals (SIs) evaluations are computationally prohibitive for embedded systems. In this paper, we derive a compact analytical measurement model for the ELF field components in seawater generated by a vertical electric dipole (VED) reference source. The model is constructed via asymptotic decomposition and trigonometric-series expansion of SIs, explicitly linking instrument readings to the EM field intensity without restriction on observation depth. A comprehensive uncertainty framework is established to quantify the model’s systematic bias and sensitivity to environmental parameters. Comparisons with numerical integration standards confirm that the proposed measurement model achieves high accuracy with negligible computational cost. The resulting formulas provide a theoretical basis for metrological characterization in underwater sensing networks, enabling real-time uncertainty evaluation and dynamic sensor correction.
{"title":"Metrological characterization and measurement model for ELF air-to-undersea fields via compact analytical VED formulas","authors":"Yongxin Cui , Zheng Dou , Yabin Zhang","doi":"10.1016/j.measurement.2026.120784","DOIUrl":"10.1016/j.measurement.2026.120784","url":null,"abstract":"<div><div>Accurate measurement of Extremely Low Frequency (ELF) electromagnetic (EM) fields in seawater is essential for oceanographic sensing, geophysical monitoring, and the calibration of undersea EM sensors. However, the lack of real-time, traceable measurement models limits the precision of in-situ sensor calibration, as exact Sommerfeld integrals (SIs) evaluations are computationally prohibitive for embedded systems. In this paper, we derive a compact analytical measurement model for the ELF field components in seawater generated by a vertical electric dipole (VED) reference source. The model is constructed via asymptotic decomposition and trigonometric-series expansion of SIs, explicitly linking instrument readings to the EM field intensity without restriction on observation depth. A comprehensive uncertainty framework is established to quantify the model’s systematic bias and sensitivity to environmental parameters. Comparisons with numerical integration standards confirm that the proposed measurement model achieves high accuracy with negligible computational cost. The resulting formulas provide a theoretical basis for metrological characterization in underwater sensing networks, enabling real-time uncertainty evaluation and dynamic sensor correction.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120784"},"PeriodicalIF":5.6,"publicationDate":"2026-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147504","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 : 2026-02-09DOI: 10.1016/j.measurement.2026.120783
Hiu-Shan Rachel Tsui
Drawing-force estimation and die-geometry optimization for cold rod drawing are recast as a measurement-model-based estimation problem, emphasizing traceability to a reference model and reproducibility. Finite element simulations (FEM) provide the reference measurement of drawing force. Artificial Neural Network (ANN) interpolators are used strictly for interpolation-only densification of the force–parameter manifold across die geometry (semi-die angle α, bearing length L) and process variables (reduction ratio r, friction coefficient μ, drawing speed v), while preserving smooth, consistent trends. The densified dataset trains an eXtreme Gradient Boosting (XGBoost) regressor as the estimator. Performance is reported using mean absolute error (MAE), mean squared error (MSE), and the coefficient of determination (R2). Five-fold cross-validation (CV) with fixed seeds ensures repeatability. Interpretability is assessed using SHapley Additive exPlanations (SHAP), with Local Interpretable Model-agnostic Explanations (LIME) used as a check. Under five-fold CV, the estimator achieves R2 = 0.9999, MAE = 0.0525 kN, and MSE = 0.0075 kN2. On five FEM-only unseen test configurations, relative errors remain within ± 2.73% across interior and boundary-like conditions within the declared domain. SHAP ranks the reduction ratio, friction coefficient, and semi-die angle as dominant contributors, while drawing speed and bearing length show minimal influence, consistent with forming mechanics. The pipeline treats FEM as a traceable reference, enriches measurement data via ANN interpolation without brute-force sweeps, and deploys a calibrated estimator for rapid, reproducible force estimation and inverse die design. The approach aligns with Guide to the Expression of Uncertainty in Measurement (GUM) inspired reporting and is readily replicable from the provided details.
{"title":"Metrology-guided estimation and inverse design of drawing force in cold rod drawing using a FEM–ANN–XGBoost hybrid","authors":"Hiu-Shan Rachel Tsui","doi":"10.1016/j.measurement.2026.120783","DOIUrl":"10.1016/j.measurement.2026.120783","url":null,"abstract":"<div><div>Drawing-force estimation and die-geometry optimization for cold rod drawing are recast as a measurement-model-based estimation problem, emphasizing traceability to a reference model and reproducibility. Finite element simulations (FEM) provide the reference measurement of drawing force. Artificial Neural Network (ANN) interpolators are used strictly for interpolation-only densification of the force–parameter manifold across die geometry (semi-die angle α, bearing length L) and process variables (reduction ratio r, friction coefficient μ, drawing speed v), while preserving smooth, consistent trends. The densified dataset trains an eXtreme Gradient Boosting (XGBoost) regressor as the estimator. Performance is reported using mean absolute error (MAE), mean squared error (MSE), and the coefficient of determination (R<sup>2</sup>). Five-fold cross-validation (CV) with fixed seeds ensures repeatability. Interpretability is assessed using SHapley Additive exPlanations (SHAP), with Local Interpretable Model-agnostic Explanations (LIME) used as a check. Under five-fold CV, the estimator achieves R<sup>2</sup> = 0.9999, MAE = 0.0525 kN, and MSE = 0.0075 kN<sup>2</sup>. On five FEM-only unseen test configurations, relative errors remain within ± 2.73% across interior and boundary-like conditions within the declared domain. SHAP ranks the reduction ratio, friction coefficient, and semi-die angle as dominant contributors, while drawing speed and bearing length show minimal influence, consistent with forming mechanics. The pipeline treats FEM as a traceable reference, enriches measurement data via ANN interpolation without brute-force sweeps, and deploys a calibrated estimator for rapid, reproducible force estimation and inverse die design. The approach aligns with Guide to the Expression of Uncertainty in Measurement (GUM) inspired reporting and is readily replicable from the provided details.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120783"},"PeriodicalIF":5.6,"publicationDate":"2026-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147470","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 : 2026-02-09DOI: 10.1016/j.measurement.2026.120595
Wei Jia , Zexiao Xie , Yuanyuan Huang , Ruiyao Wang , Xiaomin Wang
Accurate and real-time measurement of underwater robot kinematics is essential for the development and verification of advanced motion-control strategies. This paper presents UW-MBSM, a framework for accurate kinematic state measurement of underwater robots, which is implemented using a multi-binocular vision system with non-overlapping fields of view. The system integrates an imaging model for non-overlapping binocular groups, a refraction compensation model for multi-interface underwater imaging, and a fast circular coded marker detection algorithm optimized for degraded underwater visual conditions. A rigid-body-aware Kalman filtering strategy is further introduced to fuse trajectories of multiple markers and recover temporally consistent position, velocity, and acceleration. Experiments conducted in realistic underwater settings demonstrate that UW-MBSM achieves accurate, robust, and real-time kinematic measurement over an extended workspace.
{"title":"UW-MBSM: Multi-binocular vision system with non-overlapping fields of view-based underwater kinematic state measurement framework","authors":"Wei Jia , Zexiao Xie , Yuanyuan Huang , Ruiyao Wang , Xiaomin Wang","doi":"10.1016/j.measurement.2026.120595","DOIUrl":"10.1016/j.measurement.2026.120595","url":null,"abstract":"<div><div>Accurate and real-time measurement of underwater robot kinematics is essential for the development and verification of advanced motion-control strategies. This paper presents UW-MBSM, a framework for accurate kinematic state measurement of underwater robots, which is implemented using a multi-binocular vision system with non-overlapping fields of view. The system integrates an imaging model for non-overlapping binocular groups, a refraction compensation model for multi-interface underwater imaging, and a fast circular coded marker detection algorithm optimized for degraded underwater visual conditions. A rigid-body-aware Kalman filtering strategy is further introduced to fuse trajectories of multiple markers and recover temporally consistent position, velocity, and acceleration. Experiments conducted in realistic underwater settings demonstrate that UW-MBSM achieves accurate, robust, and real-time kinematic measurement over an extended workspace.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120595"},"PeriodicalIF":5.6,"publicationDate":"2026-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147477","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 : 2026-02-08DOI: 10.1016/j.measurement.2026.120747
Hui Ding , Youyou Guo , Jibo Wang , Yonghong Liu , Li Li
Addressing the limited characterization of heterogeneous electric vehicle (EV) travel patterns on real-world road networks, this study introduced the concepts of periodic intensity and dynamic spatial autocorrelation within a spatiotemporal framework. Based on real Radio Frequency Identification (RFID) monitoring data across diverse EV types, the dynamic travel patterns of electric buses (E.Buses), taxis (E.Taxis), light passenger cars (E.LPCs), and light-duty trucks (E.LDTs) were quantitatively analyzed. A spatiotemporal attention graph convolutional network (SAGCN) incorporating periodic and spatiotemporal feature extraction was constructed, achieving precise simulation for varied EV travel. Under the strategy of using immediate neighboring windows for periodic feature extraction across hourly, daily, and weekly scales, the model achieves optimal performance. Results indicate that dynamic periodicity and correlation in heterogeneous EV travel lead to fluctuating performance in the SAGCN model across different EV types and spatiotemporal conditions. E.Buses exhibited the best periodicity with intensity reaching 0.8, but poorer spatial correlation. E.Taxis demonstrated the best spatial correlation but the worst periodicity. Finally, E.Buses obtained the best simulation accuracy in strong spatial correlation with MAPE about 0.19, followed by E.LPCs, while E.Taxis had the worst simulation accuracy. This study not only assisted managers in understanding the spatiotemporal travel patterns of public and private electric vehicles but also supported the rational spatial layout of public charging piles and reasonable scheduling.
{"title":"Exposing and simulating spatiotemporal patterns of varied electric vehicles travel on a large-scale network based on real-time RFID data","authors":"Hui Ding , Youyou Guo , Jibo Wang , Yonghong Liu , Li Li","doi":"10.1016/j.measurement.2026.120747","DOIUrl":"10.1016/j.measurement.2026.120747","url":null,"abstract":"<div><div>Addressing the limited characterization of heterogeneous electric vehicle (EV) travel patterns on real-world road networks, this study introduced the concepts of periodic intensity and dynamic spatial autocorrelation within a spatiotemporal framework. Based on real Radio Frequency Identification (RFID) monitoring data across diverse EV types, the dynamic travel patterns of electric buses (E.Buses), taxis (E.Taxis), light passenger cars (E.LPCs), and light-duty trucks (E.LDTs) were quantitatively analyzed. A spatiotemporal attention graph convolutional network (SAGCN) incorporating periodic and spatiotemporal feature extraction was constructed, achieving precise simulation for varied EV travel. Under the strategy of using immediate neighboring windows for periodic feature extraction across hourly, daily, and weekly scales, the model achieves optimal performance. Results indicate that dynamic periodicity and correlation in heterogeneous EV travel lead to fluctuating performance in the SAGCN model across different EV types and spatiotemporal conditions. E.Buses exhibited the best periodicity with intensity reaching 0.8, but poorer spatial correlation. E.Taxis demonstrated the best spatial correlation but the worst periodicity. Finally, E.Buses obtained the best simulation accuracy in strong spatial correlation with MAPE about 0.19, followed by E.LPCs, while E.Taxis had the worst simulation accuracy. This study not only assisted managers in understanding the spatiotemporal travel patterns of public and private electric vehicles but also supported the rational spatial layout of public charging piles and reasonable scheduling.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120747"},"PeriodicalIF":5.6,"publicationDate":"2026-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147476","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 : 2026-02-08DOI: 10.1016/j.measurement.2026.120745
Lu Qian, Hao Chen, Yaqiong Lv, Yifan Li
Effective real-time state monitoring is crucial for the robotic grinding process and quality reliability. However, changes in process parameters and material types result in significant differences in the distribution of signals during the practical robotic grinding process. Meanwhile most related works have inadequately extracted grinding signal features, posing challenges in capturing subtle characteristics sensitive to grinding state. To address the issues, a multidimensional feature fusion-driven domain alignment (MFFDA) transfer learning method is proposed in this paper to achieve high-precision state monitoring under variable conditions by combining distribution-aligned transfer learning with multidimensional feature fusion. To this end, a multidimensional adaptive feature extraction network is constructed first to simultaneously extract time-domain, frequency-domain, and domain adaptation network features based on vibration signals. Then, a dynamic multidimensional fusion network is developed, which integrates a Transformer encoder with multi-head cross-attention mechanism to achieve effective fusion of multidimensional information. In addition, a multi-task joint optimization strategy is designed to optimize domain adaptation network, where an auxiliary supervision loss is introduced to enhance the feature discriminability. Finally, experiments on a collaborative robot grinding platform are carried out to validate the proposed method. The experiment results demonstrate that the MFFDA achieves diagnostic accuracies above 98% across diverse transfer tasks, outperforming other state-of-the-art domain adaptation methods.
{"title":"Domain alignment transfer learning based on multidimensional feature fusion for cross-domain state monitoring in robotic grinding","authors":"Lu Qian, Hao Chen, Yaqiong Lv, Yifan Li","doi":"10.1016/j.measurement.2026.120745","DOIUrl":"10.1016/j.measurement.2026.120745","url":null,"abstract":"<div><div>Effective real-time state monitoring is crucial for the robotic grinding process and quality reliability. However, changes in process parameters and material types result in significant differences in the distribution of signals during the practical robotic grinding process. Meanwhile most related works have inadequately extracted grinding signal features, posing challenges in capturing subtle characteristics sensitive to grinding state. To address the issues, a multidimensional feature fusion-driven domain alignment (MFFDA) transfer learning method is proposed in this paper to achieve high-precision state monitoring under variable conditions by combining distribution-aligned transfer learning with multidimensional feature fusion. To this end, a multidimensional adaptive feature extraction network is constructed first to simultaneously extract time-domain, frequency-domain, and domain adaptation network features based on vibration signals. Then, a dynamic multidimensional fusion network is developed, which integrates a Transformer encoder with multi-head cross-attention mechanism to achieve effective fusion of multidimensional information. In addition, a multi-task joint optimization strategy is designed to optimize domain adaptation network, where an auxiliary supervision loss is introduced to enhance the feature discriminability. Finally, experiments on a collaborative robot grinding platform are carried out to validate the proposed method. The experiment results demonstrate that the MFFDA achieves diagnostic accuracies above 98% across diverse transfer tasks, outperforming other state-of-the-art domain adaptation methods.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120745"},"PeriodicalIF":5.6,"publicationDate":"2026-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147469","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 : 2026-02-08DOI: 10.1016/j.measurement.2026.120732
Xuming Li , Huafeng Liang , Wenjun Luo , Wenjie Guo , Chao Zou
The impact of structure-borne noise on the indoor environment is receiving increasing attention. This study investigates the relationship between building structure-borne noise and out-of-plane vibrations caused by railway train operations. Measurements were conducted to explore reverberation time, indoor structure-borne noise at various locations, and out-of-plane vibrations in floors and walls to identify the generation, propagation, and attenuation of structure-borne noise. Additionally, the out-of-plane vibration distribution of the wall, the uncertainty of the vibro-acoustic relationship, and the applicability of existing empirical formulas were discussed. The indoor structure-borne noise originates from the out-of-plane vibration excitations of the floor and wall, respectively. These two types of out-of-plane vibrations exhibit distinct characteristic frequencies, thereby giving different contribution to two characteristic frequencies of the structural noise at 31.5 Hz and 80 Hz. As an essential component, the wall not only defines the boundary of the room but also significantly contributes to the excitation of structure-borne noise. Additionally, it determines several standing wave resonance frequencies in the low-frequency range. In the evaluation of building structure-borne noise, it is not sufficient to consider only the influence of the floor; rather, both the excitation effect and boundary conditions of the wall must be considered to achieve higher accuracy. The findings of this research contribute to a clearer and more comprehensive understanding of structure-borne noise.
{"title":"Vibro-acoustic coupling characteristic of railway-adjacent building: From out-of-plane vibrations of wall and floor to structure-borne noise","authors":"Xuming Li , Huafeng Liang , Wenjun Luo , Wenjie Guo , Chao Zou","doi":"10.1016/j.measurement.2026.120732","DOIUrl":"10.1016/j.measurement.2026.120732","url":null,"abstract":"<div><div>The impact of structure-borne noise on the indoor environment is receiving increasing attention. This study investigates the relationship between building structure-borne noise and out-of-plane vibrations caused by railway train operations. Measurements were conducted to explore reverberation time, indoor structure-borne noise at various locations, and out-of-plane vibrations in floors and walls to identify the generation, propagation, and attenuation of structure-borne noise. Additionally, the out-of-plane vibration distribution of the wall, the uncertainty of the vibro-acoustic relationship, and the applicability of existing empirical formulas were discussed. The indoor structure-borne noise originates from the out-of-plane vibration excitations of the floor and wall, respectively. These two types of out-of-plane vibrations exhibit distinct characteristic frequencies, thereby giving different contribution to two characteristic frequencies of the structural noise at 31.5 Hz and 80 Hz. As an essential component, the wall not only defines the boundary of the room but also significantly contributes to the excitation of structure-borne noise. Additionally, it determines several standing wave resonance frequencies in the low-frequency range. In the evaluation of building structure-borne noise, it is not sufficient to consider only the influence of the floor; rather, both the excitation effect and boundary conditions of the wall must be considered to achieve higher accuracy. The findings of this research contribute to a clearer and more comprehensive understanding of structure-borne noise.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120732"},"PeriodicalIF":5.6,"publicationDate":"2026-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147310","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 : 2026-02-08DOI: 10.1016/j.measurement.2026.120763
Yuheng Kuang , Dong Hu , Penghui Sun , Xiaoyan Xiong , Dongguang Zhang , Yali Wu
Flexible capacitive strain sensors find extensive applications in human–computer interaction and health monitoring owing to their superior repeatability and minimal power consumption. However, most existing research endeavors rely on passive optimization of materials or structures, lacking active modulation of stress line distributions, thus struggling to achieve a balance between conductive stability and strain amplification efficiency. Drawing inspiration from the directional stress transmission mechanism of earthworm segments, this study proposes and fabricates a sensor incorporating a stress guidance structure. The structure induces the stress line extrusion effect to amplify local strains, while 10 wt% nickel (Ni) particles-modified liquid metal (LM) electrodes ensure conductive stability. Combined with COMSOL simulations, the “structure-stress-capacitance” correlation is revealed, where the stress guidance structure achieves a local stress concentration factor of ∼ 4.36, effectively converting global small strains into local large deformations. Results show that the sensor exhibits a stable gauge factor (GF) of ∼ 1.38 within a broad strain range of 0 ∼ 70% with high linearity (R2 ≈ 0.997). Furthermore, the sensor demonstrates exceptional dynamic stability (rate-dependent deviation < 0.3%) and durability, with a capacitance response attenuation of only 1.5% after 5000 cycles. It can accurately capture both large joint strains and tiny physiological signals, providing a reliable solution for wearable sensing.
{"title":"High-Sensitivity flexible strain sensor with the stress line extrusion effect","authors":"Yuheng Kuang , Dong Hu , Penghui Sun , Xiaoyan Xiong , Dongguang Zhang , Yali Wu","doi":"10.1016/j.measurement.2026.120763","DOIUrl":"10.1016/j.measurement.2026.120763","url":null,"abstract":"<div><div>Flexible capacitive strain sensors find extensive applications in human–computer interaction and health monitoring owing to their superior repeatability and minimal power consumption. However, most existing research endeavors rely on passive optimization of materials or structures, lacking active modulation of stress line distributions, thus struggling to achieve a balance between conductive stability and strain amplification efficiency. Drawing inspiration from the directional stress transmission mechanism of earthworm segments, this study proposes and fabricates a sensor incorporating a stress guidance structure. The structure induces the stress line extrusion effect to amplify local strains, while 10 wt% nickel (Ni) particles-modified liquid metal (LM) electrodes ensure conductive stability. Combined with COMSOL simulations, the “structure-stress-capacitance” correlation is revealed, where the stress guidance structure achieves a local stress concentration factor of ∼ 4.36, effectively converting global small strains into local large deformations. Results show that the sensor exhibits a stable gauge factor (GF) of ∼ 1.38 within a broad strain range of 0 ∼ 70% with high linearity (<em>R</em><sup>2</sup> ≈ 0.997). Furthermore, the sensor demonstrates exceptional dynamic stability (rate-dependent deviation < 0.3%) and durability, with a capacitance response attenuation of only 1.5% after 5000 cycles. It can accurately capture both large joint strains and tiny physiological signals, providing a reliable solution for wearable sensing.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120763"},"PeriodicalIF":5.6,"publicationDate":"2026-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147475","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 : 2026-02-07DOI: 10.1016/j.measurement.2026.120718
Jiaqi Zheng , Xuhua Shi , Feifan Shen , Lingjian Ye
Soft sensors serve as virtual sensing techniques for predicting difficult-to-measure key variables in complex industrial processes. As data-driven approaches, their performance critically depends on the availability of sufficient training data. While various virtual sample generation methods have been developed to generate synthetic samples following original data distributions, most existing approaches produce individual samples and fail to preserve temporal correlations, which is a crucial aspect for time-varying industrial processes. This paper proposes a novel virtual sample generation framework based on a Time Conditional Denoising Diffusion Probabilistic Model to enhance modeling datasets for industrial processes with limited data. Unlike conventional methods, this approach generates complete virtual trajectories that maintain both static features and dynamic temporal patterns. A dual-metric filtering strategy considering both sample diversity and uncertainty, is also developed to select the most representative and valuable generated samples for data augmentation. The effectiveness of the proposed method is validated through comprehensive experiments on an two industrial applications. Specifically, the results show that the root mean square error of quality prediction by the proposed method after data augmentation is improved by 11.4% and 31.9% respectively in the two cases compared with the latest baseline, demonstrating significant improvements in prediction accuracy of soft sensors.
{"title":"A filtered time conditional denoising diffusion probabilistic model for virtual sample generation of industrial soft sensors with limited data","authors":"Jiaqi Zheng , Xuhua Shi , Feifan Shen , Lingjian Ye","doi":"10.1016/j.measurement.2026.120718","DOIUrl":"10.1016/j.measurement.2026.120718","url":null,"abstract":"<div><div>Soft sensors serve as virtual sensing techniques for predicting difficult-to-measure key variables in complex industrial processes. As data-driven approaches, their performance critically depends on the availability of sufficient training data. While various virtual sample generation methods have been developed to generate synthetic samples following original data distributions, most existing approaches produce individual samples and fail to preserve temporal correlations, which is a crucial aspect for time-varying industrial processes. This paper proposes a novel virtual sample generation framework based on a Time Conditional Denoising Diffusion Probabilistic Model to enhance modeling datasets for industrial processes with limited data. Unlike conventional methods, this approach generates complete virtual trajectories that maintain both static features and dynamic temporal patterns. A dual-metric filtering strategy considering both sample diversity and uncertainty, is also developed to select the most representative and valuable generated samples for data augmentation. The effectiveness of the proposed method is validated through comprehensive experiments on an two industrial applications. Specifically, the results show that the root mean square error of quality prediction by the proposed method after data augmentation is improved by 11.4% and 31.9% respectively in the two cases compared with the latest baseline, demonstrating significant improvements in prediction accuracy of soft sensors.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"269 ","pages":"Article 120718"},"PeriodicalIF":5.6,"publicationDate":"2026-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146147480","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号