Mechanical Systems and Signal Processing最新文献_第6页

Bias drift compensation of butterfly gyroscope under temperature change with two-stage hybrid deep learning model 基于两阶段混合深度学习模型的温度变化下蝶形陀螺仪偏置漂移补偿

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-31 DOI: 10.1016/j.ymssp.2026.113937

Gao Liu , Zhanqiang Hou , Xuan Wang , Mi Zeng , Xi Chen , Qingsong Li , Dingbang Xiao , Xuezhong Wu

This paper proposes a two-stage hybrid deep learning compensation strategy that integrates Bayesian optimization (BO), Transformer, and temporal convolutional network (TCN) to suppress bias drift in butterfly gyroscope under temperature change. First, a theoretical model for the butterfly gyroscope output was established, and the composition of thermally induced bias was analyzed. Subsequently, the original bias output signal from the gyroscope control system was decomposed using variational mode decomposition (VMD). Permutation entropy and Pearson correlation coefficient were employed as screening methods to extract the intrinsic mode functions (IMFs) caused by temperature change and strip noise. Then other temperature-related parameters that can be output serve as physically interpretable feature vector, which, together with the bias signal from the previous time step, constructs an enhanced input sequence that integrates historical memory. Subsequently, a two-stage compensation strategy was established: In the first stage, the Transformer model, which is equipped with self-attention mechanisms that capture long-range dependency, was then employed to model and compensate for deterministic drift within the reconstructed bias signal; In the second stage, the preliminarily compensated non-deterministic residual signal was fed into the TCN network, which has strong local modeling capabilities to further fit residual error, achieving precise compensation for local fluctuations. Finally, BO was employed for adaptive joint hyperparameter tuning across preprocessing and submodels. Experimental results demonstrate that the proposed two-stage hybrid compensation strategy achieves superior compensation accuracy, robustness, and generalization capability compared to traditional polynomial fitting methods and various single or combined deep learning models. Specifically, it improves the standard deviation of the compensated bias signal by over 38.3% and 42.8%, respectively, across different temperature change rates and gyroscope samples.

提出了一种结合贝叶斯优化（BO）、变压器（Transformer）和时间卷积网络（TCN）的两阶段混合深度学习补偿策略，以抑制温度变化下蝶形陀螺仪的偏置漂移。首先，建立了蝶形陀螺仪输出的理论模型，分析了其热致偏置的组成。然后，利用变分模态分解（VMD）对陀螺仪控制系统的原始偏置输出信号进行分解。采用排列熵和Pearson相关系数作为筛选方法，提取温度变化和条带噪声引起的本征模态函数（IMFs）。然后，可以输出的其他与温度相关的参数作为物理可解释的特征向量，与前一个时间步长的偏置信号一起构建一个集成历史记忆的增强输入序列。随后，建立了两阶段补偿策略：第一阶段，利用Transformer模型对重构偏置信号中的确定性漂移进行建模和补偿，该模型具有捕获远程依赖的自关注机制；第二阶段，将初步补偿的不确定性残差信号送入具有较强局部建模能力的TCN网络，进一步拟合残差误差，实现对局部波动的精确补偿。最后，利用模糊神经网络进行跨预处理和子模型的自适应联合超参数整定。实验结果表明，与传统的多项式拟合方法和各种单一或组合的深度学习模型相比，所提出的两阶段混合补偿策略具有更好的补偿精度、鲁棒性和泛化能力。具体来说，在不同温度变化率和陀螺仪样本下，补偿偏置信号的标准差分别提高了38.3%和42.8%以上。

{"title":"Bias drift compensation of butterfly gyroscope under temperature change with two-stage hybrid deep learning model","authors":"Gao Liu , Zhanqiang Hou , Xuan Wang , Mi Zeng , Xi Chen , Qingsong Li , Dingbang Xiao , Xuezhong Wu","doi":"10.1016/j.ymssp.2026.113937","DOIUrl":"10.1016/j.ymssp.2026.113937","url":null,"abstract":"<div><div>This paper proposes a two-stage hybrid deep learning compensation strategy that integrates Bayesian optimization (BO), Transformer, and temporal convolutional network (TCN) to suppress bias drift in butterfly gyroscope under temperature change. First, a theoretical model for the butterfly gyroscope output was established, and the composition of thermally induced bias was analyzed. Subsequently, the original bias output signal from the gyroscope control system was decomposed using variational mode decomposition (VMD). Permutation entropy and Pearson correlation coefficient were employed as screening methods to extract the intrinsic mode functions (IMFs) caused by temperature change and strip noise. Then other temperature-related parameters that can be output serve as physically interpretable feature vector, which, together with the bias signal from the previous time step, constructs an enhanced input sequence that integrates historical memory. Subsequently, a two-stage compensation strategy was established: In the first stage, the Transformer model, which is equipped with self-attention mechanisms that capture long-range dependency, was then employed to model and compensate for deterministic drift within the reconstructed bias signal; In the second stage, the preliminarily compensated non-deterministic residual signal was fed into the TCN network, which has strong local modeling capabilities to further fit residual error, achieving precise compensation for local fluctuations. Finally, BO was employed for adaptive joint hyperparameter tuning across preprocessing and submodels. Experimental results demonstrate that the proposed two-stage hybrid compensation strategy achieves superior compensation accuracy, robustness, and generalization capability compared to traditional polynomial fitting methods and various single or combined deep learning models. Specifically, it improves the standard deviation of the compensated bias signal by over 38.3% and 42.8%, respectively, across different temperature change rates and gyroscope samples.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113937"},"PeriodicalIF":8.9,"publicationDate":"2026-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081466","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

FRF-based crack localization in AMB-Supported rotors using neural networks 基于频响函数的amb转子裂纹定位神经网络

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113939

Giovanni Donati , Chiara Camerota , Marco Mugnaini , Michele Basso , Jerzy T. Sawicki

Well-established procedures exist for monitoring and diagnosing faults in rotating machinery, and many techniques for detecting rotor cracks have been explored in the literature. However, limited progress has been made in developing non-invasive methods capable of accurately localizing rotor cracks and assessing their severity without requiring rotor disassembly or direct physical inspection.

This paper presents a novel, non-invasive approach for crack localization in flexible rotors supported by Active Magnetic Bearings (AMBs), based exclusively on frequency responses acquired through AMB excitation. The methodology involves constructing a physics-informed fault dictionary using frequency responses simulated on a high-fidelity digital twin of the rotor system, obtained through established modeling procedures, under various crack locations and severities. These responses exhibit characteristic shifts in resonance and antiresonance frequencies, which are used to define distinct fault classes.

Neural network classifiers were trained on the simulated dataset, with a 1D Convolutional Neural Network (1D-CNN) used as the primary model and an Autoencoder + Multilayer Perceptron (AE + MLP) used as a comparative baseline, to evaluate their ability to automatically identify the fault zone. The entire framework was validated experimentally on a dedicated AMB-supported test rig, confirming the ability of the proposed method to detect and localize cracks without requiring additional sensors or plant disassembly. The 1D-CNN achieved a classification accuracy of 99.4% on simulated test data, while the AE + MLP baseline reached 98.3%. Experimental validation on a dedicated AMB-supported test rig showed correct localization for all tested crack cases.

在旋转机械中存在着完善的监测和诊断故障的程序，并且在文献中探索了许多检测转子裂纹的技术。然而，在开发能够准确定位转子裂纹并评估其严重程度而无需拆卸转子或直接物理检查的非侵入性方法方面取得了有限的进展。本文提出了一种新颖的、非侵入式的基于主动磁轴承（AMBs）激励获得的频率响应的柔性转子裂纹定位方法。该方法包括通过建立的建模程序，在不同裂纹位置和严重程度下，通过在转子系统的高保真数字孪生上模拟频率响应，构建一个物理信息故障字典。这些响应表现出共振和反共振频率的特征移位，用于定义不同的故障类别。在模拟数据集上训练神经网络分类器，以1D卷积神经网络（1D- cnn）作为主要模型，以Autoencoder + Multilayer Perceptron （AE + MLP）作为比较基线，评估其自动识别断裂带的能力。整个框架在专用的amb测试平台上进行了实验验证，证实了所提出的方法能够在不需要额外传感器或拆卸设备的情况下检测和定位裂缝。1D-CNN在模拟测试数据上的分类准确率达到99.4%，AE + MLP基线达到98.3%。在amb支持的专用测试台上进行的实验验证表明，所有测试的裂纹情况都是正确的定位。

{"title":"FRF-based crack localization in AMB-Supported rotors using neural networks","authors":"Giovanni Donati , Chiara Camerota , Marco Mugnaini , Michele Basso , Jerzy T. Sawicki","doi":"10.1016/j.ymssp.2026.113939","DOIUrl":"10.1016/j.ymssp.2026.113939","url":null,"abstract":"<div><div>Well-established procedures exist for monitoring and diagnosing faults in rotating machinery, and many techniques for detecting rotor cracks have been explored in the literature. However, limited progress has been made in developing non-invasive methods capable of accurately localizing rotor cracks and assessing their severity without requiring rotor disassembly or direct physical inspection.</div><div>This paper presents a novel, non-invasive approach for crack localization in flexible rotors supported by Active Magnetic Bearings (AMBs), based exclusively on frequency responses acquired through AMB excitation. The methodology involves constructing a physics-informed fault dictionary using frequency responses simulated on a high-fidelity digital twin of the rotor system, obtained through established modeling procedures, under various crack locations and severities. These responses exhibit characteristic shifts in resonance and antiresonance frequencies, which are used to define distinct fault classes.</div><div>Neural network classifiers were trained on the simulated dataset, with a 1D Convolutional Neural Network (1D-CNN) used as the primary model and an Autoencoder + Multilayer Perceptron (AE + MLP) used as a comparative baseline, to evaluate their ability to automatically identify the fault zone. The entire framework was validated experimentally on a dedicated AMB-supported test rig, confirming the ability of the proposed method to detect and localize cracks without requiring additional sensors or plant disassembly. The 1D-CNN achieved a classification accuracy of 99.4% on simulated test data, while the AE + MLP baseline reached 98.3%. Experimental validation on a dedicated AMB-supported test rig showed correct localization for all tested crack cases.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113939"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071615","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Aligned sparse non-negative matrix factorization for vehicle-track features decoupling 车辆-轨道特征解耦的对齐稀疏非负矩阵分解

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113907

Jiyuan Huo , Jianwei Yang , Dechen Yao , Zhongshuo Hu , Yuanting Dai , Bin Zhu

Vibration signals collected from in-service urban rail vehicles exhibit strong coupling between vehicle dynamics and track geometry excitations, often compounded by environmental noise. This poses a significant challenge for the accurate decoupling of sources and the estimation of track geometric parameters, particularly curve superelevation, from vehicle acceleration data. To address this, we propose an Aligned Sparse Non-negative Matrix Factorization (ASNMF) framework to decouple of vehicle-track features: A Kurtosis-Spectral Peak (KSP) criterion is first applied to construct a Hankel matrix that enhances the representation of non-stationary features; A multi-objective optimization is then formulated by integrating a Gini-based sparsity constraint and a Maximum Mean Discrepancy (MMD) alignment term to ensure consistent component extraction; The resulting multiplicative updating algorithm yields physically interpretable decompositions. Validation using both simulated and real-world vibration data demonstrates that ASNMF effectively separates vehicle and track-induced responses under strong coupling and noise. Compared with existing matrix factorization and blind source separation methods, ASNMF achieves higher signal fidelity and more accurate track-related feature estimation, offering a robust and novel solution for decoupling and interpreting coupled vehicle–track dynamic responses under non-stationary operating conditions.

从在役城市轨道车辆收集的振动信号显示车辆动力学和轨道几何激励之间的强耦合，通常与环境噪声混合。这对从车辆加速度数据中准确解耦和估计轨道几何参数（特别是曲线超高程）提出了重大挑战。为了解决这个问题，我们提出了一个对齐稀疏非负矩阵分解（ASNMF）框架来解耦车辆-轨道特征：首先应用峭度-谱峰（KSP）准则来构建一个增强非平稳特征表示的Hankel矩阵；然后，通过整合基于gini的稀疏性约束和最大平均差异（MMD）对齐项来制定多目标优化，以确保提取的成分一致；由此产生的乘法更新算法产生物理上可解释的分解。仿真和实际振动数据验证表明，ASNMF在强耦合和强噪声条件下有效地分离了车辆和轨道引起的响应。与现有的矩阵分解和盲源分离方法相比，ASNMF实现了更高的信号保真度和更精确的轨道相关特征估计，为非平稳工况下耦合车辆-轨道动态响应的解耦和解释提供了一种鲁棒的新解决方案。

{"title":"Aligned sparse non-negative matrix factorization for vehicle-track features decoupling","authors":"Jiyuan Huo , Jianwei Yang , Dechen Yao , Zhongshuo Hu , Yuanting Dai , Bin Zhu","doi":"10.1016/j.ymssp.2026.113907","DOIUrl":"10.1016/j.ymssp.2026.113907","url":null,"abstract":"<div><div>Vibration signals collected from in-service urban rail vehicles exhibit strong coupling between vehicle dynamics and track geometry excitations, often compounded by environmental noise. This poses a significant challenge for the accurate decoupling of sources and the estimation of track geometric parameters, particularly curve superelevation, from vehicle acceleration data. To address this, we propose an Aligned Sparse Non-negative Matrix Factorization (ASNMF) framework to decouple of vehicle-track features: A Kurtosis-Spectral Peak (KSP) criterion is first applied to construct a Hankel matrix that enhances the representation of non-stationary features; A multi-objective optimization is then formulated by integrating a Gini-based sparsity constraint and a Maximum Mean Discrepancy (MMD) alignment term to ensure consistent component extraction; The resulting multiplicative updating algorithm yields physically interpretable decompositions. Validation using both simulated and real-world vibration data demonstrates that ASNMF effectively separates vehicle and track-induced responses under strong coupling and noise. Compared with existing matrix factorization and blind source separation methods, ASNMF achieves higher signal fidelity and more accurate track-related feature estimation, offering a robust and novel solution for decoupling and interpreting coupled vehicle–track dynamic responses under non-stationary operating conditions.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113907"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 novel multi-scale dense residual shrinkage GAN for data-limited rotating machinery fault diagnosis 基于多尺度密集残余收缩GAN的旋转机械故障诊断

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113906

Tongqiang Yi , Yongjie Shi , Xiangnan Jing , Jiang Guo , Fang Yuan , Wenyang Lei

Fault diagnosis of rotating machinery is crucial for industrial safety, yet practical applications face significant challenges, including high data acquisition costs, scarce fault samples, and severe class imbalance, which severely limit the performance of deep learning diagnostic models. This paper proposes a multi-scale dense residual shrinkage generative adversarial network (MDRS-GAN) specifically designed to address fault diagnosis under limited data conditions. The method innovatively introduces a multi-scale generator that employs multiple sub-generators working collaboratively with a multi-head attention mechanism to achieve dynamic fusion of multi-scale features, significantly enhancing sample distribution simulation capabilities. Simultaneously, a hybrid time–frequency discriminator based on dense-block deep residual shrinkage networks is constructed, integrating dense connections, an improved efficient channel attention mechanism, and soft-thresholding denoising techniques to enhance sensitivity to critical features in both time and frequency domains, achieving dual functionality of sample authenticity recognition and fault classification. Additionally, a Bayesian optimization strategy is introduced to adaptively adjust discriminator hyperparameters, improving model training stability and efficiency. Extensive experiments on CWRU and XJTU-SY bearing datasets demonstrate that: generated samples achieve feature distribution similarity exceeding 0.8 (maximum 0.91) with real samples; under extreme small-sample conditions (only 2 samples per class), fault diagnosis accuracies reach 94.32% and 95.83% respectively; under severe class imbalance (100:1), accuracies maintain at 96.46% and 96.06%. Compared with existing methods, MDRS-GAN shows significant advantages across all evaluation metrics, providing an effective solution for data-limited rotating machinery fault diagnosis in industrial scenarios.

旋转机械的故障诊断对工业安全至关重要，但实际应用面临着巨大的挑战，包括数据采集成本高、故障样本稀缺以及严重的类别不平衡，这些都严重限制了深度学习诊断模型的性能。针对有限数据条件下的故障诊断问题，提出了一种多尺度密集剩余收缩生成对抗网络（MDRS-GAN）。该方法创新性地引入了一种多尺度发生器，利用多个子发生器协同工作，采用多头注意机制实现多尺度特征的动态融合，显著增强了样本分布仿真能力。同时，构建了基于密集块深度残差收缩网络的混合时频鉴别器，将密集连接、改进的高效通道关注机制和软阈值去噪技术相结合，增强了对时域和频域关键特征的敏感性，实现了样本真实性识别和故障分类的双重功能。此外，引入贝叶斯优化策略自适应调整鉴别器超参数，提高了模型训练的稳定性和效率。在CWRU和XJTU-SY轴承数据集上的大量实验表明：生成的样本与真实样本的特征分布相似度超过0.8（最大0.91）；在极端小样本条件下（每类只有2个样本），故障诊断准确率分别达到94.32%和95.83%；在严重的类不平衡（100:1）下，准确率维持在96.46%和96.06%。与现有方法相比，MDRS-GAN在所有评估指标上都显示出显著的优势，为工业场景下数据有限的旋转机械故障诊断提供了有效的解决方案。

{"title":"A novel multi-scale dense residual shrinkage GAN for data-limited rotating machinery fault diagnosis","authors":"Tongqiang Yi , Yongjie Shi , Xiangnan Jing , Jiang Guo , Fang Yuan , Wenyang Lei","doi":"10.1016/j.ymssp.2026.113906","DOIUrl":"10.1016/j.ymssp.2026.113906","url":null,"abstract":"<div><div>Fault diagnosis of rotating machinery is crucial for industrial safety, yet practical applications face significant challenges, including high data acquisition costs, scarce fault samples, and severe class imbalance, which severely limit the performance of deep learning diagnostic models. This paper proposes a multi-scale dense residual shrinkage generative adversarial network (MDRS-GAN) specifically designed to address fault diagnosis under limited data conditions. The method innovatively introduces a multi-scale generator that employs multiple sub-generators working collaboratively with a multi-head attention mechanism to achieve dynamic fusion of multi-scale features, significantly enhancing sample distribution simulation capabilities. Simultaneously, a hybrid time–frequency discriminator based on dense-block deep residual shrinkage networks is constructed, integrating dense connections, an improved efficient channel attention mechanism, and soft-thresholding denoising techniques to enhance sensitivity to critical features in both time and frequency domains, achieving dual functionality of sample authenticity recognition and fault classification. Additionally, a Bayesian optimization strategy is introduced to adaptively adjust discriminator hyperparameters, improving model training stability and efficiency. Extensive experiments on CWRU and XJTU-SY bearing datasets demonstrate that: generated samples achieve feature distribution similarity exceeding 0.8 (maximum 0.91) with real samples; under extreme small-sample conditions (only 2 samples per class), fault diagnosis accuracies reach 94.32% and 95.83% respectively; under severe class imbalance (100:1), accuracies maintain at 96.46% and 96.06%. Compared with existing methods, MDRS-GAN shows significant advantages across all evaluation metrics, providing an effective solution for data-limited rotating machinery fault diagnosis in industrial scenarios.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113906"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072593","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Nonlinear dynamics of tristable galloping-based energy harvesters and their application in weak signal enhancement 三稳驰振能量采集器的非线性动力学及其在弱信号增强中的应用

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113933

Haitao Xu , Shengxi Zhou

In nonlinear systems, the multistable potential function has been demonstrated to be an effective means to broaden the work frequency bandwidth. Taking the energy harvester as an example, the function can help it efficiently capture the broadband energy under low-level excitations. However, it is necessary to discuss the effect of asymmetry of the multistable potential function on nonlinear dynamics. Firstly, this paper designs the piecewise tristable potential function, which can produce three types of asymmetries. Secondly, according to tristable galloping-based energy harvesters, influence of asymmetries on their responses under harmonic excitation, and on the stochastic resonance phenomenon under harmonic and random excitations are investigated by numerical simulations. In addition, the phase trajectories, Poincaré maps and Lyapunov exponents are also employed to exam the system responses, such as the chaotic motion, quasi-periodic motion, and periodic motion. Thirdly, according to the experimental signal, the proposed signal enhancement methods based on the stochastic resonance of tristable galloping-based energy harvesters are successfully validated. The output signal-to-noise ratios are also calculated to compare their performance. Overall, this paper explores the effect of asymmetry on nonlinear dynamics, as well as the potential application in signal processing.

在非线性系统中，多稳定势函数已被证明是拓宽工作带宽的有效手段。以能量采集器为例，该函数可以有效地捕获低能级激励下的宽带能量。然而，有必要讨论多稳定势函数的不对称性对非线性动力学的影响。首先，本文设计了分段三稳定势函数，它可以产生三种不对称。其次，针对三稳驰振能量采集器，通过数值模拟研究了不对称性对其谐波激励下响应的影响，以及谐波和随机激励下随机共振现象的影响。此外，还采用相轨迹、poincar映射和Lyapunov指数来检测系统的混沌运动、准周期运动和周期运动等响应。第三，根据实验信号，成功验证了基于三稳驰骋能量采集器随机共振的信号增强方法。还计算了输出信噪比，以比较它们的性能。总的来说，本文探讨了不对称对非线性动力学的影响，以及在信号处理中的潜在应用。

{"title":"Nonlinear dynamics of tristable galloping-based energy harvesters and their application in weak signal enhancement","authors":"Haitao Xu , Shengxi Zhou","doi":"10.1016/j.ymssp.2026.113933","DOIUrl":"10.1016/j.ymssp.2026.113933","url":null,"abstract":"<div><div>In nonlinear systems, the multistable potential function has been demonstrated to be an effective means to broaden the work frequency bandwidth. Taking the energy harvester as an example, the function can help it efficiently capture the broadband energy under low-level excitations. However, it is necessary to discuss the effect of asymmetry of the multistable potential function on nonlinear dynamics. Firstly, this paper designs the piecewise tristable potential function, which can produce three types of asymmetries. Secondly, according to tristable galloping-based energy harvesters, influence of asymmetries on their responses under harmonic excitation, and on the stochastic resonance phenomenon under harmonic and random excitations are investigated by numerical simulations. In addition, the phase trajectories, Poincaré maps and Lyapunov exponents are also employed to exam the system responses, such as the chaotic motion, quasi-periodic motion, and periodic motion. Thirdly, according to the experimental signal, the proposed signal enhancement methods based on the stochastic resonance of tristable galloping-based energy harvesters are successfully validated. The output signal-to-noise ratios are also calculated to compare their performance. Overall, this paper explores the effect of asymmetry on nonlinear dynamics, as well as the potential application in signal processing.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113933"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072595","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 comprehensive review of indirect bridge health monitoring 桥梁间接健康监测技术综述

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113918

Zhenkun Li , Weiwei Lin , Chul-Woo Kim , Maria Pina Limongelli , Eleni Chatzi

Indirect Bridge Health Monitoring (BHM) using indirect measurements of the response from passing vehicles has recently gained significant attention from researchers within the Structural Health Monitoring (SHM) domain. This approach requires only one or a few sensors installed on the vehicle, making it more cost-effective, efficient, and easier to implement than traditional methods, which demand numerous sensors on bridges. Recent advancements in both algorithms and hardware have further accelerated progress in this field. This paper aims to provide a comprehensive, one-stop review of indirect BHM using measured vehicle response since 2004. It systematically analyzes the connections and integrations within existing literature, incorporating rapidly emerging state-of-the-art studies. The review initiates with a bibliometric analysis, covering annual publication trends, keyword cooccurrence, and authorship networks, followed by a discussion on the fundamental theories of vehicle–bridge interaction. Subsequently, it summarizes the vehicle, bridge, and road roughness models used in indirect BHM. Furthermore, it explores current techniques and challenges in identifying bridge modal parameters, such as bridge frequencies, mode shapes, and damping ratios, as well as in indirect bridge damage detection using signal processing, modal-based, and data-driven methods. Additionally, this review includes affiliated studies that, while not directly related, contribute to the advancement of indirect BHM. Finally, recent developments in 2025, future investigation directions, and key conclusions are provided. It is intended to serve as a fundamental resource for researchers seeking to advance their studies in the field of indirect BHM.

间接桥梁健康监测（BHM）是一种利用间接测量过往车辆对桥梁的反应的方法，近年来受到了结构健康监测（SHM）领域研究人员的广泛关注。这种方法只需要在车辆上安装一个或几个传感器，与需要在桥梁上安装多个传感器的传统方法相比，它更具成本效益、效率更高，也更容易实施。最近在算法和硬件方面的进步进一步加速了这一领域的进展。本文旨在提供一个全面的，一站式的间接BHM审查使用测量车辆响应自2004年以来。它系统地分析了现有文献中的联系和整合，结合了快速出现的最先进的研究。回顾从文献计量分析开始，包括年度出版趋势、关键词协同和作者网络，随后讨论了车桥相互作用的基本理论。随后，总结了间接BHM中使用的车辆、桥梁和道路粗糙度模型。此外，它还探讨了识别桥梁模态参数的当前技术和挑战，例如桥梁频率，模态振型和阻尼比，以及使用信号处理，基于模态和数据驱动方法的间接桥梁损伤检测。此外，本综述还包括了与间接BHM进展相关的附属研究，尽管这些研究没有直接关系。最后，给出了2025年的最新发展、未来的研究方向和关键结论。它旨在为寻求推进间接BHM领域研究的研究人员提供基础资源。

{"title":"A comprehensive review of indirect bridge health monitoring","authors":"Zhenkun Li , Weiwei Lin , Chul-Woo Kim , Maria Pina Limongelli , Eleni Chatzi","doi":"10.1016/j.ymssp.2026.113918","DOIUrl":"10.1016/j.ymssp.2026.113918","url":null,"abstract":"<div><div>Indirect Bridge Health Monitoring (BHM) using indirect measurements of the response from passing vehicles has recently gained significant attention from researchers within the Structural Health Monitoring (SHM) domain. This approach requires only one or a few sensors installed on the vehicle, making it more cost-effective, efficient, and easier to implement than traditional methods, which demand numerous sensors on bridges. Recent advancements in both algorithms and hardware have further accelerated progress in this field. This paper aims to provide a comprehensive, one-stop review of indirect BHM using measured vehicle response since 2004. It systematically analyzes the connections and integrations within existing literature, incorporating rapidly emerging state-of-the-art studies. The review initiates with a bibliometric analysis, covering annual publication trends, keyword cooccurrence, and authorship networks, followed by a discussion on the fundamental theories of vehicle–bridge interaction. Subsequently, it summarizes the vehicle, bridge, and road roughness models used in indirect BHM. Furthermore, it explores current techniques and challenges in identifying bridge modal parameters, such as bridge frequencies, mode shapes, and damping ratios, as well as in indirect bridge damage detection using signal processing, modal-based, and data-driven methods. Additionally, this review includes affiliated studies that, while not directly related, contribute to the advancement of indirect BHM. Finally, recent developments in 2025, future investigation directions, and key conclusions are provided. It is intended to serve as a fundamental resource for researchers seeking to advance their studies in the field of indirect BHM.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113918"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072600","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Anomalous data diagnosis in bridge strain monitoring by fusing multi-modal data feature 融合多模态数据特征的桥梁应变监测异常数据诊断

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113922

Pengtao Chen , Gan Yang , Junfeng Wang , Xiuping Liu , Shizhi Chen , Wanshui Han

Identifying anomalous monitoring data caused by unstable sensor performance is crucial for accurately assessing the operational condition of bridges. In practical monitoring, such anomalies often exhibit various complex patterns, such as slow-varying trends and missing data. However, traditional analysis methods based on unimodal data features struggle to simultaneously consider both the transient dynamics and the global evolutionary features of time-series data, which leads to insufficient identification capability for slow-varying anomalies such as drift and trend. To address this, a framework for diagnosing anomalous bridge data based on multimodal data feature fusion is proposed, which achieves fine-grained identification of complex anomaly patterns by fusing Markov Transition Field (MTF) image features with one-dimensional (1D) time-series features. This fusion dynamically combines features from two parallel branches: one branch extracts global state transition patterns from the MTF images, while the other captures key transient dynamics from the 1D time-series data. Experimental results show that the method achieves an overall mean Average Precision (mAP) of 99.83% on the main girder strain monitoring data from a highway cable-stayed bridge (across seven data classes), representing a significant improvement compared to models using only unimodal data features, with the image-only model achieving 94.63% and the time-series-only model achieving 91.34%. Notably, the F1-Scores for minority slow-varying anomalies (trend, drift) are improved by over 15%. Furthermore, the model demonstrates strong generalization, achieving 97.97% accuracy on a large-scale dataset collected from sensor locations that were used during training.

识别由传感器性能不稳定引起的异常监测数据对于准确评估桥梁运行状况至关重要。在实际监测中，这种异常常常表现出各种复杂的模式，如缓慢变化的趋势和丢失的数据。然而，传统的基于单峰数据特征的分析方法难以同时考虑时间序列数据的瞬态动力学和全局演化特征，导致对漂移和趋势等慢变异常的识别能力不足。针对这一问题，提出了一种基于多模态数据特征融合的桥梁异常数据诊断框架，该框架通过将马尔可夫过渡场（MTF）图像特征与一维时间序列特征融合，实现了复杂异常模式的细粒度识别。这种融合动态地结合了两个并行分支的特征：一个分支从MTF图像中提取全局状态转移模式，而另一个分支从一维时间序列数据中捕获关键的瞬态动态。实验结果表明，该方法对某公路斜拉桥主梁应变监测数据（跨越7个数据类别）的总体平均平均精度（mAP）达到99.83%，与仅使用单峰数据特征的模型相比有显著提高，其中仅图像模型达到94.63%，仅时间序列模型达到91.34%。值得注意的是，少数慢变异常（趋势，漂移）的f1分数提高了15%以上。此外，该模型具有很强的泛化能力，在训练期间使用的传感器位置收集的大规模数据集上达到97.97%的准确率。

{"title":"Anomalous data diagnosis in bridge strain monitoring by fusing multi-modal data feature","authors":"Pengtao Chen , Gan Yang , Junfeng Wang , Xiuping Liu , Shizhi Chen , Wanshui Han","doi":"10.1016/j.ymssp.2026.113922","DOIUrl":"10.1016/j.ymssp.2026.113922","url":null,"abstract":"<div><div>Identifying anomalous monitoring data caused by unstable sensor performance is crucial for accurately assessing the operational condition of bridges. In practical monitoring, such anomalies often exhibit various complex patterns, such as slow-varying trends and missing data. However, traditional analysis methods based on unimodal data features struggle to simultaneously consider both the transient dynamics and the global evolutionary features of time-series data, which leads to insufficient identification capability for slow-varying anomalies such as drift and trend. To address this, a framework for diagnosing anomalous bridge data based on multimodal data feature fusion is proposed, which achieves fine-grained identification of complex anomaly patterns by fusing Markov Transition Field (MTF) image features with one-dimensional (1D) time-series features. This fusion dynamically combines features from two parallel branches: one branch extracts global state transition patterns from the MTF images, while the other captures key transient dynamics from the 1D time-series data. Experimental results show that the method achieves an overall mean Average Precision (mAP) of 99.83% on the main girder strain monitoring data from a highway cable-stayed bridge (across seven data classes), representing a significant improvement compared to models using only unimodal data features, with the image-only model achieving 94.63% and the time-series-only model achieving 91.34%. Notably, the F1-Scores for minority slow-varying anomalies (trend, drift) are improved by over 15%. Furthermore, the model demonstrates strong generalization, achieving 97.97% accuracy on a large-scale dataset collected from sensor locations that were used during training.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113922"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072598","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 modified Levenberg–Marquardt method for estimating the elastic material parameters of polymer waveguides using residuals between autocorrelated frequency responses 基于自相关频率响应间残差估计聚合物波导弹性材料参数的改进Levenberg-Marquardt方法

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113904

Dominik Itner , Dmitrij Dreiling , Hauke Gravenkamp , Bernd Henning , Carolin Birk

In this contribution, we address the estimation of the frequency-dependent elastic parameters of polymers in the ultrasound range, which is formulated as an inverse problem. This inverse problem is implemented as a nonlinear regression-type optimization problem, in which the simulation signals are fitted to the measurement signals. These signals consist of displacement responses in waveguides, focusing on hollow cylindrical geometries to enhance the simulation efficiency. To accelerate the optimization and reduce the number of model evaluations and wait times, we propose two novel methods. First, we introduce an adaptation of the Levenberg–Marquardt method derived from a geometrical interpretation of the least-squares optimization problem. Second, we introduce an improved objective function based on the autocorrelated envelopes of the measurement and simulation signals. Given that this study primarily relies on simulation data to quantify optimization convergence, we aggregate the expected ranges of realistic material parameters and derive their distributions to ensure the reproducibility of optimizations with proper measurements. We demonstrate the effectiveness of our objective function modification and step adaptation for various materials with isotropic material symmetry by comparing them with the Broyden–Fletcher–Goldfarb–Shanno method. In all cases, our method reduces the total number of model evaluations, thereby shortening the time to identify the material parameters.

在这一贡献中，我们解决了超声范围内聚合物的频率相关弹性参数的估计，这是一个逆问题。该反问题是一个非线性回归型优化问题，其中仿真信号拟合到测量信号。这些信号由波导中的位移响应组成，重点关注空心圆柱几何形状，以提高仿真效率。为了加快优化速度，减少模型评估的次数和等待时间，我们提出了两种新的方法。首先，我们从最小二乘优化问题的几何解释中引入了Levenberg-Marquardt方法的适应性。其次，基于测量信号和仿真信号的自相关包络引入了一种改进的目标函数。鉴于本研究主要依赖于模拟数据来量化优化收敛，我们汇总了实际材料参数的预期范围，并推导了它们的分布，以确保优化的可重复性。通过与Broyden-Fletcher-Goldfarb-Shanno方法的比较，证明了我们的目标函数修正和步进自适应方法对具有各向同性材料对称性的各种材料的有效性。在所有情况下，我们的方法减少了模型评估的总数，从而缩短了识别材料参数的时间。

{"title":"A modified Levenberg–Marquardt method for estimating the elastic material parameters of polymer waveguides using residuals between autocorrelated frequency responses","authors":"Dominik Itner , Dmitrij Dreiling , Hauke Gravenkamp , Bernd Henning , Carolin Birk","doi":"10.1016/j.ymssp.2026.113904","DOIUrl":"10.1016/j.ymssp.2026.113904","url":null,"abstract":"<div><div>In this contribution, we address the estimation of the frequency-dependent elastic parameters of polymers in the ultrasound range, which is formulated as an inverse problem. This inverse problem is implemented as a nonlinear regression-type optimization problem, in which the simulation signals are fitted to the measurement signals. These signals consist of displacement responses in waveguides, focusing on hollow cylindrical geometries to enhance the simulation efficiency. To accelerate the optimization and reduce the number of model evaluations and wait times, we propose two novel methods. First, we introduce an adaptation of the Levenberg–Marquardt method derived from a geometrical interpretation of the least-squares optimization problem. Second, we introduce an improved objective function based on the autocorrelated envelopes of the measurement and simulation signals. Given that this study primarily relies on simulation data to quantify optimization convergence, we aggregate the expected ranges of realistic material parameters and derive their distributions to ensure the reproducibility of optimizations with proper measurements. We demonstrate the effectiveness of our objective function modification and step adaptation for various materials with isotropic material symmetry by comparing them with the Broyden–Fletcher–Goldfarb–Shanno method. In all cases, our method reduces the total number of model evaluations, thereby shortening the time to identify the material parameters.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113904"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072599","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Repetitive error selected iterative learning contouring control for precision multiaxis systems 精密多轴系统的重复误差选择迭代学习轮廓控制

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2025.113583

Ze Wang , Fang Peng , Min Li , Taotao Chen , Chuxiong Hu , Yu Zhu

The contour error control method based on Iterative Learning Control (ILC) has gained widespread application in multi-axis precision motion systems due to its excellent control accuracy. Fundamentally, ILC works by filtering both repetitive and non-repetitive errors over multiple repetitive control tasks, and iteratively compensating for the repetitive errors to achieve extremely high control precision. In engineering applications, the low-frequency components of errors are typically considered repetitive, while the high-frequency components are regarded as random and non-repetitive. Based on this, ILC often utilizes low-pass filters to filter out repetitive errors. However, the causes of multi-axis contour errors are far more complex than those of single-axis tracking errors. We believe that directly using frequency characteristics to distinguish whether contour errors are repetitive is insufficiently accurate, despite the extensive use of this approach in prior studies. Therefore, this paper proposes a novel ILC method for contour error control based on the selection of repetitive errors. The method determines whether the errors in corresponding frequency bands exhibit repetitive characteristics based on the spectral features of the contour errors, thus enabling more precise filtering of repetitive errors. This approach effectively avoids the issues of convergence speed and precision degradation caused by the introduction of non-repetitive components during the iteration process in traditional ILC. Furthermore, we conducted a series of validation experiments on a multi-axis motion platform, which fully demonstrate that the proposed method outperforms traditional methods under various experimental conditions, effectively addressing the shortcomings of ILC in error filtering.

基于迭代学习控制（ILC）的轮廓误差控制方法以其优异的控制精度在多轴精密运动系统中得到了广泛的应用。从根本上说，ILC的工作原理是在多个重复控制任务上过滤重复和非重复误差，并迭代地补偿重复误差，以实现极高的控制精度。在工程应用中，误差的低频分量通常被认为是重复的，而高频分量则被认为是随机的和非重复的。基于此，ILC通常使用低通滤波器滤除重复误差。然而，多轴轮廓误差产生的原因远比单轴跟踪误差产生的原因复杂。我们认为，尽管在先前的研究中广泛使用这种方法，但直接使用频率特性来区分轮廓误差是否重复是不够准确的。因此，本文提出了一种基于重复误差选择的轮廓误差控制方法。该方法根据轮廓误差的频谱特征判断相应频带的误差是否具有重复特征，从而能够更精确地滤波重复误差。该方法有效地避免了传统ILC在迭代过程中由于引入非重复元件而导致的收敛速度和精度下降问题。此外，我们在多轴运动平台上进行了一系列验证实验，充分证明了该方法在各种实验条件下都优于传统方法，有效地解决了ILC在误差滤波方面的不足。

{"title":"Repetitive error selected iterative learning contouring control for precision multiaxis systems","authors":"Ze Wang , Fang Peng , Min Li , Taotao Chen , Chuxiong Hu , Yu Zhu","doi":"10.1016/j.ymssp.2025.113583","DOIUrl":"10.1016/j.ymssp.2025.113583","url":null,"abstract":"<div><div>The contour error control method based on Iterative Learning Control (ILC) has gained widespread application in multi-axis precision motion systems due to its excellent control accuracy. Fundamentally, ILC works by filtering both repetitive and non-repetitive errors over multiple repetitive control tasks, and iteratively compensating for the repetitive errors to achieve extremely high control precision. In engineering applications, the low-frequency components of errors are typically considered repetitive, while the high-frequency components are regarded as random and non-repetitive. Based on this, ILC often utilizes low-pass filters to filter out repetitive errors. However, the causes of multi-axis contour errors are far more complex than those of single-axis tracking errors. We believe that directly using frequency characteristics to distinguish whether contour errors are repetitive is insufficiently accurate, despite the extensive use of this approach in prior studies. Therefore, this paper proposes a novel ILC method for contour error control based on the selection of repetitive errors. The method determines whether the errors in corresponding frequency bands exhibit repetitive characteristics based on the spectral features of the contour errors, thus enabling more precise filtering of repetitive errors. This approach effectively avoids the issues of convergence speed and precision degradation caused by the introduction of non-repetitive components during the iteration process in traditional ILC. Furthermore, we conducted a series of validation experiments on a multi-axis motion platform, which fully demonstrate that the proposed method outperforms traditional methods under various experimental conditions, effectively addressing the shortcomings of ILC in error filtering.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113583"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072592","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 vision-based vibration measurement method of bridge structure using swin transformer motion magnification and improved recurrent all-pairs field transform algorithms 基于旋转变压器运动放大和改进的循环全对场变换算法的桥梁结构视觉振动测量方法

IF 8.9 1区工程技术 Q1 ENGINEERING, MECHANICAL

Mechanical Systems and Signal Processing

Pub Date : 2026-01-29 DOI: 10.1016/j.ymssp.2026.113940

Jieqi Li , Wei Ji , Lu Deng , Yong Liu

Accurate measurement of bridge vibrations under load excitation is often impeded using traditional computer vision-based measurement (CVBM) methods in complex environments, particularly when small amplitudes and low-texture surfaces are involved. This paper proposes a CVBM method that estimates the full-field dynamic displacement and modal parameters of bridge structures by integrating the swin transformer motion magnification (STMM) with the improved recurrent all-pairs field transform (IRAFT) algorithm. The STMM algorithm could capture high-frequency information about structural vibrations while effectively suppressing noise, blurring, and motion artifacts during magnification. The IRAFT algorithm could improve the calculation accuracy of full-field optical flow in low-texture and large-motion scenes. First, vibrational videos of bridge structures are acquired and calibrated, and then the pixel motion in the video is magnified by the STMM algorithm. Subsequently, the IRAFT algorithm is employed to compute full-field optical flow from the magnified video. Finally, pixel motion within regions of interest is converted to a displacement time-history curve through motion normalization and a scale-factor method, from which modal parameters are identified based on fast Fourier transform and covariance-driven stochastic subspace identification. The proposed method was validated on a synthetic truss bridge, a laboratory experiment of a Q235 simply-supported beam, and on-site inspection of a pedestrian overpass. The results show that the utilization of the STMM algorithm is not only effective in acquiring high-frequency information of structural vibrations but also has the advantage of suppressing motion noise and artifacts during magnification. By integrating the STMM with the IRAFT algorithm, the identification performance of the full-field dynamic displacement and modal parameters of structures is significantly improved, with better robustness to illumination changes and partial occlusion.

传统的基于计算机视觉的测量（CVBM）方法在复杂环境下，特别是在涉及小振幅和低纹理表面时，往往无法准确测量荷载激励下的桥梁振动。本文提出了一种CVBM方法，通过将swin变压器运动放大（STMM）与改进的循环全对场变换（IRAFT）算法相结合，估计桥梁结构的全场动态位移和模态参数。STMM算法可以捕获结构振动的高频信息，同时有效地抑制放大过程中的噪声、模糊和运动伪影。IRAFT算法可以提高低纹理大运动场景下的全场光流计算精度。首先对桥梁结构的振动视频进行采集和标定，然后利用STMM算法对视频中的像素运动进行放大。随后，利用IRAFT算法计算放大后的视频的全场光流。最后，通过运动归一化和尺度因子方法将感兴趣区域内的像素运动转换为位移时程曲线，并基于快速傅立叶变换和协方差驱动的随机子空间识别方法识别模态参数。通过综合桁架桥、Q235简支梁的室内试验和人行天桥的现场检验，验证了该方法的有效性。结果表明，利用STMM算法不仅可以有效地获取结构振动的高频信息，而且在放大过程中具有抑制运动噪声和伪影的优点。通过将STMM与IRAFT算法相结合，显著提高了结构的全场动态位移和模态参数识别性能，对光照变化和局部遮挡具有更好的鲁棒性。

{"title":"A vision-based vibration measurement method of bridge structure using swin transformer motion magnification and improved recurrent all-pairs field transform algorithms","authors":"Jieqi Li , Wei Ji , Lu Deng , Yong Liu","doi":"10.1016/j.ymssp.2026.113940","DOIUrl":"10.1016/j.ymssp.2026.113940","url":null,"abstract":"<div><div>Accurate measurement of bridge vibrations under load excitation is often impeded using traditional computer vision-based measurement (CVBM) methods in complex environments, particularly when small amplitudes and low-texture surfaces are involved. This paper proposes a CVBM method that estimates the full-field dynamic displacement and modal parameters of bridge structures by integrating the swin transformer motion magnification (STMM) with the improved recurrent all-pairs field transform (IRAFT) algorithm. The STMM algorithm could capture high-frequency information about structural vibrations while effectively suppressing noise, blurring, and motion artifacts during magnification. The IRAFT algorithm could improve the calculation accuracy of full-field optical flow in low-texture and large-motion scenes. First, vibrational videos of bridge structures are acquired and calibrated, and then the pixel motion in the video is magnified by the STMM algorithm. Subsequently, the IRAFT algorithm is employed to compute full-field optical flow from the magnified video. Finally, pixel motion within regions of interest is converted to a displacement time-history curve through motion normalization and a scale-factor method, from which modal parameters are identified based on fast Fourier transform and covariance-driven stochastic subspace identification. The proposed method was validated on a synthetic truss bridge, a laboratory experiment of a Q235 simply-supported beam, and on-site inspection of a pedestrian overpass. The results show that the utilization of the STMM algorithm is not only effective in acquiring high-frequency information of structural vibrations but also has the advantage of suppressing motion noise and artifacts during magnification. By integrating the STMM with the IRAFT algorithm, the identification performance of the full-field dynamic displacement and modal parameters of structures is significantly improved, with better robustness to illumination changes and partial occlusion.</div></div>","PeriodicalId":51124,"journal":{"name":"Mechanical Systems and Signal Processing","volume":"247 ","pages":"Article 113940"},"PeriodicalIF":8.9,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072596","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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