Measurement最新文献_第10页

Comprehensive study of machining mechanism, experimental measurement and numerical analysis in ultrasonic-assisted micro-drilling of printed circuit board 超声辅助印刷电路板微孔加工机理、实验测量和数值分析的综合研究

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-26 DOI: 10.1016/j.measurement.2026.120954

Moran Xu , Yixiao Yang , Xunchen Chu , Longchao Cao , Junxiang Li , Jia Ge , Shuo Chen , David Adzah , Saood Ali , Wang Cai , Changping Li , Tae Jo Ko

As the core carrier for achieving high-density interconnections in electronic devices, the quality of micro hole fabrication in printed circuit board (PCB) plays a crucial role in determining electrical performance and overall system reliability. Conventional micro drilling (CMD) often leads to defects such as burr formation, copper foil tearing, substrate delamination, and thermal damage, which severely restrict its applicability in high-precision fields. By integrating experimental measurements with finite element simulations, this study systematically verifies the pronounced advantages of ultrasonic-assisted micro-drilling (UAMD) in regulating the cutting process and enhancing machining performance in PCB micro-hole fabrication. Utilizing a self-developed high-frequency ultrasonic vibration spindle system, both CMD and UAMD processes were conducted to analyze the effects of spindle speed and feed rate on drilling thrust, hole entrance/exit morphology, inner wall integrity, chip formation, and tool wear. The experimental measurements are validated through comparison with simulation models. The results indicate that the high-frequency vibrations induced by UAMD markedly enhance cutting force stability and facilitate efficient chip removal. In comparison to CMD, UAMD achieved a maximum reduction of 19.1% in cutting force when drilling copper and 21.5% when drilling glass fiber reinforced polymer (GFRP), accompanied by a 14.8% decrease in the damage factor. Furthermore, tool wear was significantly reduced due to the reduction in heat accumulation. This study shows that UAMD optimizes cutting dynamics, reduces cutting forces, improves chip evacuation and hole-wall integrity, and enhances the consistency and overall quality of PCB micro-holes, providing theoretical and technical support for high-efficiency fabrication in advanced PCB manufacturing.

作为实现电子器件高密度互连的核心载体，印刷电路板（PCB）微孔的制造质量对电气性能和整个系统的可靠性起着至关重要的作用。传统的微钻技术存在毛刺形成、铜箔撕裂、基片剥离、热损伤等缺陷，严重制约了其在高精度领域的应用。通过实验测量和有限元模拟相结合，系统验证了超声辅助微钻在PCB微孔加工中调节加工过程和提高加工性能方面的显著优势。利用自主开发的高频超声振动主轴系统，进行了CMD和UAMD工艺，分析了主轴转速和进给速率对钻进推力、孔进出口形貌、内壁完整性、切屑形成和刀具磨损的影响。通过与仿真模型的比较，验证了实验结果。结果表明，UAMD引起的高频振动显著提高了切削力的稳定性，有利于高效的切屑去除。与CMD相比，UAMD在钻铜时的切削力最大降低了19.1%，在钻玻璃纤维增强聚合物（GFRP）时最大降低了21.5%，同时损伤系数降低了14.8%。此外，由于热积累的减少，刀具磨损显著减少。研究表明，UAMD优化了切削动力学，减小了切削力，提高了切屑排出率和孔壁完整性，提高了PCB微孔的一致性和整体质量，为先进PCB制造中的高效制造提供了理论和技术支持。

{"title":"Comprehensive study of machining mechanism, experimental measurement and numerical analysis in ultrasonic-assisted micro-drilling of printed circuit board","authors":"Moran Xu , Yixiao Yang , Xunchen Chu , Longchao Cao , Junxiang Li , Jia Ge , Shuo Chen , David Adzah , Saood Ali , Wang Cai , Changping Li , Tae Jo Ko","doi":"10.1016/j.measurement.2026.120954","DOIUrl":"10.1016/j.measurement.2026.120954","url":null,"abstract":"<div><div>As the core carrier for achieving high-density interconnections in electronic devices, the quality of micro hole fabrication in printed circuit board (PCB) plays a crucial role in determining electrical performance and overall system reliability. Conventional micro drilling (CMD) often leads to defects such as burr formation, copper foil tearing, substrate delamination, and thermal damage, which severely restrict its applicability in high-precision fields. By integrating experimental measurements with finite element simulations, this study systematically verifies the pronounced advantages of ultrasonic-assisted micro-drilling (UAMD) in regulating the cutting process and enhancing machining performance in PCB micro-hole fabrication. Utilizing a self-developed high-frequency ultrasonic vibration spindle system, both CMD and UAMD processes were conducted to analyze the effects of spindle speed and feed rate on drilling thrust, hole entrance/exit morphology, inner wall integrity, chip formation, and tool wear. The experimental measurements are validated through comparison with simulation models. The results indicate that the high-frequency vibrations induced by UAMD markedly enhance cutting force stability and facilitate efficient chip removal. In comparison to CMD, UAMD achieved a maximum reduction of 19.1% in cutting force when drilling copper and 21.5% when drilling glass fiber reinforced polymer (GFRP), accompanied by a 14.8% decrease in the damage factor. Furthermore, tool wear was significantly reduced due to the reduction in heat accumulation. This study shows that UAMD optimizes cutting dynamics, reduces cutting forces, improves chip evacuation and hole-wall integrity, and enhances the consistency and overall quality of PCB micro-holes, providing theoretical and technical support for high-efficiency fabrication in advanced PCB manufacturing.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120954"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387886","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}

引用次数: 0

Incipient bearing fault diagnosis based on optical fiber sensor and feature-informed geometric partition entropy guided informative frequency band extraction 基于光纤传感器和特征信息几何分割熵的信息频带提取轴承早期故障诊断

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-28 DOI: 10.1016/j.measurement.2026.121003

Kai Zheng , Zihao Long , Pengyuan Zhao , Jiaquan Tang , Bo Wei , Maolin Luo

Fiber Bragg Grating (FBG) sensors, with advantages of compactness and immunity to electromagnetic interference, offers a transformative solution for overcoming the challenges of bearing fault diagnosis under harsh operating conditions. The FBG dynamic strain signal can be used for detecting early weak bearing fault signal. However, when the FBG dynamic strain signals are processed through second-order differentiation to derive equivalent acceleration signals, the noise is significantly amplified, which poses significant challenges to extract the weak fault feature. To address this issue, this paper proposes a feature-informed geometric partition entropy (FIGPE) guided informative frequency bands (IFBs) extraction strategy for bearing fault diagnosis based on FBG dynamic strain signal. Initially, the cyclostationarity and impulsiveness of the equivalent acceleration signal of FBG strain caused by bearing faults is revealed. After that, a new indicator named as FIGPE with no prior fault information is developed. This indicator comprehensively considers both the cyclostationarity and impulsiveness of the FBG equivalent acceleration signal. Finally, a bandpass filter based on Linear Piecewise Windowing (LPW) is constructed to isolate fault-related frequency components from the signal, where the FIGPE is guided to adaptively determine optimal center frequencies and bandwidth parameters of the filter based on Bayesian optimization. The results show that the proposed method can effectively identify bearing fault features in FBG equivalent acceleration signals without prior knowledge while suppressing noise. Its effectiveness has also been further verified through simulations, experiments, and comparisons with existing methods.

光纤布拉格光栅（FBG）传感器具有紧凑和抗电磁干扰的优点，为克服恶劣运行条件下轴承故障诊断的挑战提供了一种变革性的解决方案。光纤光栅动态应变信号可用于检测轴承早期微弱故障信号。然而，当对FBG动态应变信号进行二阶微分处理得到等效加速度信号时，噪声被显著放大，这给微弱故障特征的提取带来了很大的挑战。针对这一问题，提出了一种基于FBG动态应变信号的基于特征的几何分割熵（FIGPE）引导信息频带（IFBs）提取策略。首先揭示了由轴承故障引起的光纤光栅应变等效加速度信号的循环平稳性和冲动性。在此基础上，提出了一种不含先验故障信息的新指标FIGPE。该指标综合考虑了光纤光栅等效加速度信号的循环平稳性和脉冲性。最后，构建了基于线性分段加窗（LPW）的带通滤波器，隔离信号中与故障相关的频率分量，并引导FIGPE基于贝叶斯优化自适应确定滤波器的最优中心频率和带宽参数。结果表明，该方法可以在不需要先验知识的情况下，有效地识别FBG等效加速度信号中的轴承故障特征，同时抑制噪声。通过仿真、实验和与现有方法的比较，进一步验证了该方法的有效性。

{"title":"Incipient bearing fault diagnosis based on optical fiber sensor and feature-informed geometric partition entropy guided informative frequency band extraction","authors":"Kai Zheng , Zihao Long , Pengyuan Zhao , Jiaquan Tang , Bo Wei , Maolin Luo","doi":"10.1016/j.measurement.2026.121003","DOIUrl":"10.1016/j.measurement.2026.121003","url":null,"abstract":"<div><div>Fiber Bragg Grating (FBG) sensors, with advantages of compactness and immunity to electromagnetic interference, offers a transformative solution for overcoming the challenges of bearing fault diagnosis under harsh operating conditions. The FBG dynamic strain signal can be used for detecting early weak bearing fault signal. However, when the FBG dynamic strain signals are processed through second-order differentiation to derive equivalent acceleration signals, the noise is significantly amplified, which poses significant challenges to extract the weak fault feature. To address this issue, this paper proposes a feature-informed geometric partition entropy (FIGPE) guided informative frequency bands (IFBs) extraction strategy for bearing fault diagnosis based on FBG dynamic strain signal. Initially, the cyclostationarity and impulsiveness of the equivalent acceleration signal of FBG strain caused by bearing faults is revealed. After that, a new indicator named as FIGPE with no prior fault information is developed. This indicator comprehensively considers both the cyclostationarity and impulsiveness of the FBG equivalent acceleration signal. Finally, a bandpass filter based on Linear Piecewise Windowing (LPW) is constructed to isolate fault-related frequency components from the signal, where the FIGPE is guided to adaptively determine optimal center frequencies and bandwidth parameters of the filter based on Bayesian optimization. The results show that the proposed method can effectively identify bearing fault features in FBG equivalent acceleration signals without prior knowledge while suppressing noise. Its effectiveness has also been further verified through simulations, experiments, and comparisons with existing methods.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 121003"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387294","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}

引用次数: 0

A point cloud registration method based on RGB-D information and voting mechanism 基于RGB-D信息和投票机制的点云配准方法

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-26 DOI: 10.1016/j.measurement.2026.120989

Yuandong Huang, Jinbo Chen, Mei Liu, Zhiming Wang

Accurate point cloud registration under RGB-D sensing remains challenging when the data suffer from high noise, low quality, and missing geometric features. This study proposes a RGB-D fusion-based registration method to enhance robustness and accuracy in such adverse conditions. A graph structure is first constructed from initial correspondences, and nodal clustering coefficients are utilized for pruning to improve computational efficiency. An image difference weight is then introduced to evaluate node reliability, while a dynamic visual coefficient refines the node scoring process. Subsequently, a voting mechanism combining geometric and image features is designed to select high-confidence correspondences through node–edge interactions. High-scoring nodes are extracted as initial inliers, followed by RANSAC-based optimization to further refine inlier selection. The proposed method was validated on a custom RGB-D dataset and public datasets. Experimental results demonstrate that the method achieves superior registration precision and stability, especially when dealing with low-quality, weakly geometric, high-noise, or unknown-overlap point clouds, confirming its effectiveness for practical RGB-D measurement applications.

RGB-D遥感下点云的配准存在高噪声、低质量和几何特征缺失等问题。本研究提出了一种基于RGB-D融合的配准方法，以提高在这种不利条件下的鲁棒性和准确性。首先从初始对应关系构造图结构，利用节点聚类系数进行剪枝，提高计算效率。然后引入图像差值权重来评估节点可靠性，同时引入动态视觉系数来细化节点评分过程。随后，设计了一种结合几何特征和图像特征的投票机制，通过节点-边缘交互选择高置信度对应。提取高分节点作为初始拟合点，然后基于ransac优化进一步细化拟合点选择。在自定义RGB-D数据集和公共数据集上对该方法进行了验证。实验结果表明，该方法具有较高的配准精度和稳定性，特别是在低质量、弱几何、高噪声或未知重叠的点云中，验证了该方法在实际RGB-D测量应用中的有效性。

{"title":"A point cloud registration method based on RGB-D information and voting mechanism","authors":"Yuandong Huang, Jinbo Chen, Mei Liu, Zhiming Wang","doi":"10.1016/j.measurement.2026.120989","DOIUrl":"10.1016/j.measurement.2026.120989","url":null,"abstract":"<div><div>Accurate point cloud registration under RGB-D sensing remains challenging when the data suffer from high noise, low quality, and missing geometric features. This study proposes a RGB-D fusion-based registration method to enhance robustness and accuracy in such adverse conditions. A graph structure is first constructed from initial correspondences, and nodal clustering coefficients are utilized for pruning to improve computational efficiency. An image difference weight is then introduced to evaluate node reliability, while a dynamic visual coefficient refines the node scoring process. Subsequently, a voting mechanism combining geometric and image features is designed to select high-confidence correspondences through node–edge interactions. High-scoring nodes are extracted as initial inliers, followed by RANSAC-based optimization to further refine inlier selection. The proposed method was validated on a custom RGB-D dataset and public datasets. Experimental results demonstrate that the method achieves superior registration precision and stability, especially when dealing with low-quality, weakly geometric, high-noise, or unknown-overlap point clouds, confirming its effectiveness for practical RGB-D measurement applications.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120989"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387335","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}

引用次数: 0

A lightweight CNN-ISO collaborative optimization framework addressing high-dimensional redundancy and class imbalance in rolling bearing fault diagnosis 针对滚动轴承故障诊断中高维冗余和类不平衡的轻量级CNN-ISO协同优化框架

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-20 DOI: 10.1016/j.measurement.2026.120912

Siyu Chen , Bangcheng Zhang , Xiao He , Xiaopeng Chang , Yubo Shao , Zhi Gao

A lightweight CNN-ISO collaborative framework with dual feature selection is proposed to address the Accuracy-efficiency trade-off arising from high-dimensional redundancy and class imbalance in rolling bearing fault diagnosis. In the first stage, a one-dimensional CNN automatically extracts 256-dimensional deep features from the raw vibration signals, constituting the initial feature selection phase. In the second stage, an improved snake swarm optimization (ISO) algorithm, guided by a joint fitness function, conducts a synchronous search in both the feature and parameter spaces, thereby achieving simultaneous optimization of feature compression and classification Accuracy. Experimental evaluations on the CWRU, Paderborn, and XJTU-SY datasets show that the proposed framework achieves an average feature compression rate of approximately 0.75 while maintaining an Accuracy above 0.90. Compared with CNN, LSTM, CNN_SVM, and CNN_SO_SVM, it yields a Macro-F1 improvement of approximately 0.10–0.50, particularly on datasets exhibiting inter-class imbalance, with test times ranging from 0.0008 to 0.02 s. These results confirm that the proposed dual feature selection and collaborative optimization strategy not only effectively mitigates high-dimensional redundancy and class imbalance but also offers an efficient and practical solution for real-time fault diagnosis in industrial applications.

针对滚动轴承故障诊断中高维冗余和类不平衡导致的精度与效率权衡问题，提出了一种轻量级的CNN-ISO双特征选择协同框架。第一阶段，一维CNN自动从原始振动信号中提取256维深度特征，构成初始特征选择阶段。第二阶段，采用改进的蛇群优化（ISO）算法，在联合适应度函数的指导下，对特征空间和参数空间进行同步搜索，实现特征压缩和分类精度的同步优化。在CWRU、Paderborn和XJTU-SY数据集上的实验评估表明，所提出的框架实现了大约0.75的平均特征压缩率，同时保持了0.90以上的精度。与CNN、LSTM、CNN_SVM和CNN_SO_SVM相比，该方法的Macro-F1改进约为0.10-0.50，特别是在类间不平衡的数据集上，测试时间范围为0.0008 ~ 0.02 s。结果表明，本文提出的双特征选择和协同优化策略不仅有效地缓解了高维冗余和类不平衡，而且为工业应用中的实时故障诊断提供了高效实用的解决方案。

{"title":"A lightweight CNN-ISO collaborative optimization framework addressing high-dimensional redundancy and class imbalance in rolling bearing fault diagnosis","authors":"Siyu Chen , Bangcheng Zhang , Xiao He , Xiaopeng Chang , Yubo Shao , Zhi Gao","doi":"10.1016/j.measurement.2026.120912","DOIUrl":"10.1016/j.measurement.2026.120912","url":null,"abstract":"<div><div>A lightweight CNN-ISO collaborative framework with dual feature selection is proposed to address the Accuracy-efficiency trade-off arising from high-dimensional redundancy and class imbalance in rolling bearing fault diagnosis. In the first stage, a one-dimensional CNN automatically extracts 256-dimensional deep features from the raw vibration signals, constituting the initial feature selection phase. In the second stage, an improved snake swarm optimization (ISO) algorithm, guided by a joint fitness function, conducts a synchronous search in both the feature and parameter spaces, thereby achieving simultaneous optimization of feature compression and classification Accuracy. Experimental evaluations on the CWRU, Paderborn, and XJTU-SY datasets show that the proposed framework achieves an average feature compression rate of approximately 0.75 while maintaining an Accuracy above 0.90. Compared with CNN, LSTM, CNN_SVM, and CNN_SO_SVM, it yields a Macro-F1 improvement of approximately 0.10–0.50, particularly on datasets exhibiting inter-class imbalance, with test times ranging from 0.0008 to 0.02 s. These results confirm that the proposed dual feature selection and collaborative optimization strategy not only effectively mitigates high-dimensional redundancy and class imbalance but also offers an efficient and practical solution for real-time fault diagnosis in industrial applications.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120912"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387377","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}

引用次数: 0

A VIS-NIR spectral -temporal- spatial resolved measurement system and its application in bio-samples sensing VIS-NIR光谱-时空分辨测量系统及其在生物样品传感中的应用

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-28 DOI: 10.1016/j.measurement.2026.121000

Yunfei Li , Guancheng Li , Yizhen Wang , Xiuling Liu , Fuhong Cai

Traditional diffuse reflectance spectroscopy (DRS) primarily targets the near‑infrared spectral range and seldom integrates temporal and spatial domain information for a comprehensive investigation of biological tissues. In this work, we present a compact visible–near-infrared system that acquires continuous diffuse reflectance spectra from 400 to 800 nm at multiple source–detector separations ranging from 6.75 to 15.75 mm for transcutaneous measurements. A slit-based detection window improves compactness and collection efficiency, yielding enhanced short-wavelength performance with the signal-to-noise ratio increasing from 1.34 to 6.76 in the 400 to 500 nm range compared with a conventional spectrometer. Comprehensive spectral information enables more accurate extraction of hemoglobin and melanin dynamics, thereby enhancing the analysis in transcutaneous detection applications. The integration of temporal information enables the extraction of pulse waveform signals across the full spectral range of VIS–NIR. Temporal analysis further enables photoplethysmography (PPG) waveform retrieval across the full 400 to 800 nm range, showing strong cross-wavelength synchrony, with a peak-matching score of 0.805. Moreover, the incorporation of spatial information facilitates the extraction of depth-resolved optical properties, enabling the quantification of effective attenuation coefficients to differentiate tissue types. In the 600–700 nm band, the effective attenuation coefficient in the palm is 1.90-fold higher than that in the forearm, consistent with differences in tissue optical properties. This provides a key tool for future clinical physiological monitoring, particularly in overcoming signal variations caused by differences in scattering coefficients among individuals. In the future, the spectral-temporal-spatial resolved measurement system can provide more comprehensive optical information for DRS and holds the potential to enable more accurate monitoring of physiological parameters in clinical settings.

传统的漫反射光谱（DRS）主要针对近红外光谱范围，很少结合时空域信息来全面研究生物组织。在这项工作中，我们提出了一种紧凑的可见-近红外系统，该系统在多源探测器距离为6.75至15.75 mm的情况下获得400至800 nm的连续漫反射光谱，用于经皮测量。基于狭缝的检测窗口提高了紧凑性和收集效率，在400 ~ 500 nm范围内，与传统光谱仪相比，信噪比从1.34提高到6.76，增强了短波性能。全面的光谱信息能够更准确地提取血红蛋白和黑色素动态，从而增强经皮检测应用中的分析。时间信息的整合使得在VIS-NIR的全光谱范围内提取脉冲波形信号成为可能。时间分析进一步使光体积脉搏波（PPG）波形在整个400至800 nm范围内检索，显示出强的跨波长同步性，峰值匹配得分为0.805。此外，空间信息的结合有助于提取深度分辨光学特性，从而量化有效衰减系数以区分组织类型。在600-700 nm波段，手掌的有效衰减系数比前臂高1.90倍，与组织光学特性的差异一致。这为未来临床生理监测提供了关键工具，特别是在克服个体间散射系数差异引起的信号变化方面。在未来，光谱-时空分辨测量系统可以为DRS提供更全面的光学信息，并有可能在临床环境中实现更准确的生理参数监测。

{"title":"A VIS-NIR spectral -temporal- spatial resolved measurement system and its application in bio-samples sensing","authors":"Yunfei Li , Guancheng Li , Yizhen Wang , Xiuling Liu , Fuhong Cai","doi":"10.1016/j.measurement.2026.121000","DOIUrl":"10.1016/j.measurement.2026.121000","url":null,"abstract":"<div><div>Traditional diffuse reflectance spectroscopy (DRS) primarily targets the near‑infrared spectral range and seldom integrates temporal and spatial domain information for a comprehensive investigation of biological tissues. In this work, we present a compact visible–near-infrared system that acquires continuous diffuse reflectance spectra from 400 to 800 nm at multiple source–detector separations ranging from 6.75 to 15.75 mm for transcutaneous measurements. A slit-based detection window improves compactness and collection efficiency, yielding enhanced short-wavelength performance with the signal-to-noise ratio increasing from 1.34 to 6.76 in the 400 to 500 nm range compared with a conventional spectrometer. Comprehensive spectral information enables more accurate extraction of hemoglobin and melanin dynamics, thereby enhancing the analysis in transcutaneous detection applications. The integration of temporal information enables the extraction of pulse waveform signals across the full spectral range of VIS–NIR. Temporal analysis further enables photoplethysmography (PPG) waveform retrieval across the full 400 to 800 nm range, showing strong cross-wavelength synchrony, with a peak-matching score of 0.805. Moreover, the incorporation of spatial information facilitates the extraction of depth-resolved optical properties, enabling the quantification of effective attenuation coefficients to differentiate tissue types. In the 600–700 nm band, the effective attenuation coefficient in the palm is 1.90-fold higher than that in the forearm, consistent with differences in tissue optical properties. This provides a key tool for future clinical physiological monitoring, particularly in overcoming signal variations caused by differences in scattering coefficients among individuals. In the future, the spectral-temporal-spatial resolved measurement system can provide more comprehensive optical information for DRS and holds the potential to enable more accurate monitoring of physiological parameters in clinical settings.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 121000"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387431","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}

引用次数: 0

A multi-stage bearing remaining useful life prediction framework integrating dual-layer data fusion and feature correction 结合双层数据融合和特征校正的多级轴承剩余使用寿命预测框架

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-27 DOI: 10.1016/j.measurement.2026.120991

Zhen Zhang , Zhijian Wang , Junyuan Wang , Zhongxin Chen , Weibo Ren , Yanfeng Li , Wenhua Du , Lei Donga , Xin Fan

In complex operating conditions, multi-sensor data fusion and degradation modeling often face dynamic uncertainties that affect the reliability of Remaining Useful Life (RUL) predictions. To address this issue, this paper proposes a confidence-guided multi-stage RUL prediction framework. First, the Double-layer Time-weighted Multi-sensor Fusion Strategy (DTMFS) algorithm is proposed, which constructs a dual-layer fusion mechanism by dynamically evaluating sensor-level and system-level confidence, guiding adaptive sensor weighting. Second, feature subset selection and primary-auxiliary feature fusion strategies are adopted in Health Indicator (HI) construction to optimize information utilization. An Adaptive Degradation Trajectory Correction (ADTC) method is also proposed to ensure the monotonicity and physical consistency of the HI. Finally, a confidence score(CS) is proposed to dynamically assess model output credibility, enabling adaptive selection in multi-stage modeling. Validation through accelerated bearing degradation experiments, the public XJTU-SY dataset, and the PHM 2012 Challenge dataset shows that the proposed framework significantly enhances prediction stability and reliability.

在复杂的工作条件下，多传感器数据融合和退化建模经常面临动态不确定性，影响剩余使用寿命（RUL）预测的可靠性。为了解决这一问题，本文提出了一个基于置信度的多阶段RUL预测框架。首先，提出了双层时间加权多传感器融合策略（DTMFS）算法，该算法通过动态评估传感器级和系统级置信度，引导自适应传感器加权，构建了双层融合机制；其次，在健康指标构建中采用特征子集选择和主辅特征融合策略，优化信息利用；为了保证HI的单调性和物理一致性，提出了自适应退化轨迹校正（ADTC）方法。最后，提出了一个置信度分数（CS）来动态评估模型输出的可信度，实现了多阶段建模的自适应选择。通过加速轴承退化实验、XJTU-SY公共数据集和PHM 2012 Challenge数据集的验证表明，该框架显著提高了预测的稳定性和可靠性。

{"title":"A multi-stage bearing remaining useful life prediction framework integrating dual-layer data fusion and feature correction","authors":"Zhen Zhang , Zhijian Wang , Junyuan Wang , Zhongxin Chen , Weibo Ren , Yanfeng Li , Wenhua Du , Lei Donga , Xin Fan","doi":"10.1016/j.measurement.2026.120991","DOIUrl":"10.1016/j.measurement.2026.120991","url":null,"abstract":"<div><div>In complex operating conditions, multi-sensor data fusion and degradation modeling often face dynamic uncertainties that affect the reliability of Remaining Useful Life (RUL) predictions. To address this issue, this paper proposes a confidence-guided multi-stage RUL prediction framework. First, the Double-layer Time-weighted Multi-sensor Fusion Strategy (DTMFS) algorithm is proposed, which constructs a dual-layer fusion mechanism by dynamically evaluating sensor-level and system-level confidence, guiding adaptive sensor weighting. Second, feature subset selection and primary-auxiliary feature fusion strategies are adopted in Health Indicator (HI) construction to optimize information utilization. An Adaptive Degradation Trajectory Correction (ADTC) method is also proposed to ensure the monotonicity and physical consistency of the HI. Finally, a confidence score(CS) is proposed to dynamically assess model output credibility, enabling adaptive selection in multi-stage modeling. Validation through accelerated bearing degradation experiments, the public XJTU-SY dataset, and the PHM 2012 Challenge dataset shows that the proposed framework significantly enhances prediction stability and reliability.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120991"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387604","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}

引用次数: 0

Dual-tube electrical sensing zone fusion and dual-Weibull inversion for wide-range particle size measurement of Bayer-process slurry 双管电感测区融合和双weibull反演用于拜耳法料浆大范围粒度测量

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-26 DOI: 10.1016/j.measurement.2026.120986

Qi Feng , Hongjian Cao , Jia Zhao , Youjian Su , Jingjing Qiu , Xiaodong Wang

Accurate online monitoring of Bayer-process bauxite slurry particle-size distribution underpins process control and optimisation in alumina refineries. However, conventional instruments—including the standard Bayer-process sieving method—provide only offline, narrow-range information and therefore cannot support continuous operation. To address this limitation, we develop a wide-range particle-size distribution reconstruction method that integrates a dual-tube Coulter-Principle sensor with a stitching—fusion inversion strategy. A dual-channel electrical sensing zone system with two partially overlapping apertures provides direct particle-size measurements over approximately 50–640 µm, while a fusion-based inversion model reconstructs the PSD over an inferred range of 1–1000 µm; results below 50 µm are not directly validated by ESZ in this study. The two resulting spectra are combined by an adaptive stitching algorithm that determines a stable overlap interval from local slope characteristics, yielding a smooth and physically consistent composite distribution. Coarse-end reconstruction is further enhanced using Boltzmann weighting and a dual-Weibull inversion model. Across three industrial Bayer slurries, the fused ESZ indices F₁₂₅, F₂₀₀ and F₈₀₀ agree with the Bayer-process sieving method within ±3.3 percentage points. Within the validated operating window (≥50 µm), the method provides stable and reliable particle-size distribution information suitable for online monitoring, blending control and process optimisation in alumina production.

准确的在线监测拜耳过程铝土矿浆料粒度分布支撑过程控制和优化在氧化铝精炼厂。然而，传统的仪器——包括标准的拜耳工艺筛分方法——只能提供离线、窄范围的信息，因此不能支持连续操作。为了解决这一限制，我们开发了一种大范围粒径分布重建方法，该方法将双管库尔特原理传感器与缝合融合反演策略集成在一起。具有两个部分重叠孔径的双通道电传感区系统提供了大约50-640 μ m的直接粒径测量，而基于融合的反演模型在1-1000 μ m的推断范围内重建PSD；在本研究中，小于50µm的结果未被ESZ直接验证。两个结果光谱通过自适应拼接算法组合，该算法根据局部坡度特征确定稳定的重叠区间，从而产生光滑且物理一致的复合分布。利用玻尔兹曼加权和双威布尔反演模型进一步增强粗端重构。在三种拜耳工业浆料中，融合的ESZ指数F125、F200和F800与拜耳工艺筛分方法的一致性在±3.3个百分点以内。在经过验证的操作窗口（≥50µm）内，该方法可提供稳定可靠的粒度分布信息，适用于氧化铝生产中的在线监测、混合控制和工艺优化。

{"title":"Dual-tube electrical sensing zone fusion and dual-Weibull inversion for wide-range particle size measurement of Bayer-process slurry","authors":"Qi Feng , Hongjian Cao , Jia Zhao , Youjian Su , Jingjing Qiu , Xiaodong Wang","doi":"10.1016/j.measurement.2026.120986","DOIUrl":"10.1016/j.measurement.2026.120986","url":null,"abstract":"<div><div>Accurate online monitoring of Bayer-process bauxite slurry particle-size distribution underpins process control and optimisation in alumina refineries. However, conventional instruments—including the standard Bayer-process sieving method—provide only offline, narrow-range information and therefore cannot support continuous operation. To address this limitation, we develop a wide-range particle-size distribution reconstruction method that integrates a dual-tube Coulter-Principle sensor with a stitching—fusion inversion strategy. A dual-channel electrical sensing zone system with two partially overlapping apertures provides direct particle-size measurements over approximately 50–640 µm, while a fusion-based inversion model reconstructs the PSD over an inferred range of 1–1000 µm; results below 50 µm are not directly validated by ESZ in this study. The two resulting spectra are combined by an adaptive stitching algorithm that determines a stable overlap interval from local slope characteristics, yielding a smooth and physically consistent composite distribution. Coarse-end reconstruction is further enhanced using Boltzmann weighting and a dual-Weibull inversion model. Across three industrial Bayer slurries, the fused ESZ indices F<sub>125</sub>, F<sub>200</sub> and F<sub>800</sub> agree with the Bayer-process sieving method within ±3.3 percentage points. Within the validated operating window (≥50 µm), the method provides stable and reliable particle-size distribution information suitable for online monitoring, blending control and process optimisation in alumina production.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120986"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387613","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}

引用次数: 0

A Data-Driven framework for precise geometric measurement of tunnel structures using 3D point clouds and Bayesian optimization 基于三维点云和贝叶斯优化的隧道结构精确几何测量数据驱动框架

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-23 DOI: 10.1016/j.measurement.2026.120931

Fei Meng, Shifan Qiao, Daolong Chen

The geometric state assessment of complex engineering structures from 3D point clouds is challenged by the insufficient accuracy and poor reproducibility of conventional techniques, which are highly sensitive to processing parameters. To overcome these limitations, a novel data-driven measurement framework is proposed. The framework enhances a conventional dual-plane projection technique by integrating a sliding window mechanism for refined local analysis. A Bayesian optimization algorithm is subsequently introduced to automate the selection of the sliding window’s critical parameters, thereby eliminating operator subjectivity and ensuring measurement optimality. The entire automated workflow is implemented within a high-performance computational architecture that synergizes MATLAB and C++. Validation through an engineering case study of a tunnel structure demonstrated that the Bayesian optimization converged to an optimal parameter set, minimizing the objective function to 1.4201. The geometric parameters extracted using this optimized method, such as an average cross-sectional ovality of

9.65 \times 10^{- 5}

, exhibited significantly higher sensitivity to local geometric deviations compared to conventional global methods (

1.93 \times 10^{- 5}

). The proposed framework provides a robust and automated solution for 3D geometric measurement, fundamentally improving the accuracy, reliability, and reproducibility of the measurement process for complex structures by systematically addressing parameter uncertainty.

传统的基于三维点云的复杂工程结构几何状态评估方法对加工参数高度敏感，精度不足，重现性差。为了克服这些限制，提出了一种新的数据驱动测量框架。该框架通过集成滑动窗口机制来改进传统的双平面投影技术，以进行精细的局部分析。随后引入贝叶斯优化算法自动选择滑动窗口的关键参数，从而消除了操作员的主观性，保证了测量的最优性。整个自动化工作流是在MATLAB和c++协同的高性能计算架构中实现的。通过隧道结构的工程实例研究验证，贝叶斯优化收敛于最优参数集，使目标函数最小为1.4201。与传统的全局方法（1.93×10-5）相比，使用该优化方法提取的几何参数（如平均横截面椭圆度9.65×10-5）对局部几何偏差的敏感性显著提高。该框架为三维几何测量提供了一个鲁棒的自动化解决方案，通过系统地解决参数不确定性，从根本上提高了复杂结构测量过程的准确性、可靠性和可重复性。

{"title":"A Data-Driven framework for precise geometric measurement of tunnel structures using 3D point clouds and Bayesian optimization","authors":"Fei Meng, Shifan Qiao, Daolong Chen","doi":"10.1016/j.measurement.2026.120931","DOIUrl":"10.1016/j.measurement.2026.120931","url":null,"abstract":"<div><div>The geometric state assessment of complex engineering structures from 3D point clouds is challenged by the insufficient accuracy and poor reproducibility of conventional techniques, which are highly sensitive to processing parameters. To overcome these limitations, a novel data-driven measurement framework is proposed. The framework enhances a conventional dual-plane projection technique by integrating a sliding window mechanism for refined local analysis. A Bayesian optimization algorithm is subsequently introduced to automate the selection of the sliding window’s critical parameters, thereby eliminating operator subjectivity and ensuring measurement optimality. The entire automated workflow is implemented within a high-performance computational architecture that synergizes MATLAB and C++. Validation through an engineering case study of a tunnel structure demonstrated that the Bayesian optimization converged to an optimal parameter set, minimizing the objective function to 1.4201. The geometric parameters extracted using this optimized method, such as an average cross-sectional ovality of <span><math><mrow><mn>9.65</mn><mo>×</mo><msup><mn>10</mn><mrow><mo>-</mo><mn>5</mn></mrow></msup></mrow></math></span>, exhibited significantly higher sensitivity to local geometric deviations compared to conventional global methods (<span><math><mrow><mn>1.93</mn><mo>×</mo><msup><mn>10</mn><mrow><mo>-</mo><mn>5</mn></mrow></msup></mrow></math></span>). The proposed framework provides a robust and automated solution for 3D geometric measurement, fundamentally improving the accuracy, reliability, and reproducibility of the measurement process for complex structures by systematically addressing parameter uncertainty.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120931"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387617","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}

引用次数: 0

Measurement of particle parameters of condensed phase products in solid rocket motor plume under the rarefied atmospheric environment 稀薄大气环境下固体火箭发动机羽流凝聚相产物颗粒参数的测量

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-23 DOI: 10.1016/j.measurement.2026.120926

Feng Yang , Xueting Ran , Yu Zhang , Li Ping , Chaoqi Xu , Bin Yang

To realize particle parameters measurements of condensed phase products in a solid rocket motor (SRM) plume under the rarefied atmospheric environment, a laser diffraction-based particle parameter measurement system was established based on the principle of laser diffraction particle sizing. Static particle size measurements were validated by using standard particles with diameters of 0.8 μm, 1.5 μm, 20 μm, 100 μm and 240 μm. The results indicate that the maximum relative deviation of particle size measurement is 6.7%. On this basis, the measurement system was applied to measure particle parameters of condensed phase products in the SRM plume under the normal and rarefied atmospheric environments. The results show that the volume-weighted mean particle size (D₄₃) of condensed phase products in SRM plume under the rarefied atmospheric environment is approximately 10 μm, which is much smaller than the normal atmospheric environment. The light transmittance of SRM plume under the rarefied atmospheric environment during stable working stage ranges from 96.0% to 98.5%, while that under the normal atmospheric environment is below 30.0%. Furthermore, when distance from the nozzle exit increases, the particle size distribution broadens firstly and then narrows. The above measurement results of particle parameters of condensed phase products in the SRM plume can provide direct data support for the evaluation of plume characteristic signals, plume simulation modeling, and optimization design of SRM.

为了实现稀薄大气环境下固体火箭发动机（SRM）羽流凝聚相产物颗粒参数的测量，基于激光衍射粒度测量原理，建立了基于激光衍射的颗粒参数测量系统。采用直径为0.8 μm、1.5 μm、20 μm、100 μm和240 μm的标准颗粒对静态粒径测量结果进行了验证。结果表明，粒径测量的最大相对偏差为6.7%。在此基础上，应用该测量系统对正常大气环境和稀薄大气环境下SRM羽流中凝聚相产物的颗粒参数进行了测量。结果表明：稀薄大气环境下SRM羽流中凝聚相产物的体积加权平均粒径（D43）约为10 μm，远小于正常大气环境；稀薄大气环境下SRM羽流稳定工作阶段的透光率在96.0% ~ 98.5%之间，而正常大气环境下的透光率在30.0%以下。随着与喷嘴出口距离的增加，颗粒粒径分布先变宽后变窄。以上SRM羽流中凝聚相产物颗粒参数的测量结果可以为羽流特征信号评价、羽流模拟建模和SRM优化设计提供直接的数据支持。

{"title":"Measurement of particle parameters of condensed phase products in solid rocket motor plume under the rarefied atmospheric environment","authors":"Feng Yang , Xueting Ran , Yu Zhang , Li Ping , Chaoqi Xu , Bin Yang","doi":"10.1016/j.measurement.2026.120926","DOIUrl":"10.1016/j.measurement.2026.120926","url":null,"abstract":"<div><div>To realize particle parameters measurements of condensed phase products in a solid rocket motor (SRM) plume under the rarefied atmospheric environment, a laser diffraction-based particle parameter measurement system was established based on the principle of laser diffraction particle sizing. Static particle size measurements were validated by using standard particles with diameters of 0.8 μm, 1.5 μm, 20 μm, 100 μm and 240 μm. The results indicate that the maximum relative deviation of particle size measurement is 6.7%. On this basis, the measurement system was applied to measure particle parameters of condensed phase products in the SRM plume under the normal and rarefied atmospheric environments. The results show that the volume-weighted mean particle size (D<sub>43</sub>) of condensed phase products in SRM plume under the rarefied atmospheric environment is approximately 10 μm, which is much smaller than the normal atmospheric environment. The light transmittance of SRM plume under the rarefied atmospheric environment during stable working stage ranges from 96.0% to 98.5%, while that under the normal atmospheric environment is below 30.0%. Furthermore, when distance from the nozzle exit increases, the particle size distribution broadens firstly and then narrows. The above measurement results of particle parameters of condensed phase products in the SRM plume can provide direct data support for the evaluation of plume characteristic signals, plume simulation modeling, and optimization design of SRM.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120926"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387655","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}

引用次数: 0

MTKS-Mamba: A multi-task knowledge sharing network with Mambaformer for joint battery state estimation MTKS-Mamba：基于Mambaformer的多任务知识共享网络，用于联合电池状态估计

IF 5.6 2区工程技术 Q1 ENGINEERING, MULTIDISCIPLINARY

Measurement

Pub Date : 2026-04-28 Epub Date: 2026-02-22 DOI: 10.1016/j.measurement.2026.120922

Ruopeng Wu, Ying Shi, Changjun Xie

Accurate estimation of state of charge (SOC) and state of health (SOH) for lithium-ion batteries is crucial for electric vehicles. Existing studies typically focus on estimating individual states, with little consideration given to their inherent relationships. In this paper, we propose a joint battery state estimation framework, called MTKS-Mamba, based on a multi-task knowledge sharing network (MTKSN) and a hybrid deep learning architecture Mambaformer. First, through the cycle charging and discharging aging experiment of lithium-ion batteries, the key features are extracted for estimating SOC and SOH. Next, the MTKSN explores inherent relationships between SOC and SOH estimation tasks, sharing knowledge across layers through similar convolutional kernels. Finally, Mambaformer is introduced to capture long-short range dependencies in time series for learning battery aging and charging behaviors. The experimental results show that the proposed method can precisely estimate the SOC and SOH of lithium-ion batteries, with the RMSE and MAE of SOC estimation not exceeding 1.18% and 1.02%, respectively, and the RMSE and MAE of SOH estimation not exceeding 2.44% and 1.78%, respectively, outperforming existing methods in terms of accuracy and robustness.

准确估计锂离子电池的荷电状态（SOC）和健康状态（SOH）对电动汽车至关重要。现有的研究通常侧重于估计单个状态，很少考虑它们之间的内在关系。在本文中，我们提出了一个基于多任务知识共享网络（MTKSN）和混合深度学习架构Mambaformer的联合电池状态估计框架MTKS-Mamba。首先，通过锂离子电池循环充放电老化实验，提取关键特征，用于估算SOC和SOH；接下来，MTKSN探索SOC和SOH估计任务之间的内在关系，通过类似的卷积核跨层共享知识。最后，引入Mambaformer捕获时间序列中的长短距离依赖关系，用于学习电池老化和充电行为。实验结果表明，该方法能够准确估计锂离子电池的荷电状态和SOH，荷电状态估计的RMSE和MAE分别不超过1.18%和1.02%，SOH估计的RMSE和MAE分别不超过2.44%和1.78%，在准确性和鲁棒性方面均优于现有方法。

{"title":"MTKS-Mamba: A multi-task knowledge sharing network with Mambaformer for joint battery state estimation","authors":"Ruopeng Wu, Ying Shi, Changjun Xie","doi":"10.1016/j.measurement.2026.120922","DOIUrl":"10.1016/j.measurement.2026.120922","url":null,"abstract":"<div><div>Accurate estimation of state of charge (SOC) and state of health (SOH) for lithium-ion batteries is crucial for electric vehicles. Existing studies typically focus on estimating individual states, with little consideration given to their inherent relationships. In this paper, we propose a joint battery state estimation framework, called MTKS-Mamba, based on a multi-task knowledge sharing network (MTKSN) and a hybrid deep learning architecture Mambaformer. First, through the cycle charging and discharging aging experiment of lithium-ion batteries, the key features are extracted for estimating SOC and SOH. Next, the MTKSN explores inherent relationships between SOC and SOH estimation tasks, sharing knowledge across layers through similar convolutional kernels. Finally, Mambaformer is introduced to capture long-short range dependencies in time series for learning battery aging and charging behaviors. The experimental results show that the proposed method can precisely estimate the SOC and SOH of lithium-ion batteries, with the RMSE and MAE of SOC estimation not exceeding 1.18% and 1.02%, respectively, and the RMSE and MAE of SOH estimation not exceeding 2.44% and 1.78%, respectively, outperforming existing methods in terms of accuracy and robustness.</div></div>","PeriodicalId":18349,"journal":{"name":"Measurement","volume":"271 ","pages":"Article 120922"},"PeriodicalIF":5.6,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147387656","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}

引用次数: 0