Structural health monitoring (SHM) is a critical concern, especially for metropolitan cities exposed to natural hazards and environmental influences. Remote sensing plays a vital role in monitoring surface movements by taking advantage of regional image acquisition at regular intervals. In this study, the persistent scatterer interferometric synthetic aperture radar (PSInSAR) method was applied to Sentinel-1 satellite data acquired between 2018 and 2021 to monitor surface deformations in the lagoon region between the Büyükçekmece and Küçükçekmece Lakes, Istanbul. This region is particularly significant due to its proximity to the North Anatolian Fault Zone, the presence of active landslides, and clay-rich geological formations. Analysis of line-of-sight (LOS) deformation results from both ascending and descending orbits revealed deformation rates ranging from −9.5 to +7.3 mm/year and −9.9 to +6.6 mm/year, respectively. Six regions containing various engineering structures were selected for investigation. The study identified the impacts of active landslides on structures as well as the seasonal effects associated with clayey geological formations. Additionally, the deformations observed in coastal structures (e.g., piers and breakwaters) and thermal effects on bridge structures (e.g., viaducts) were evaluated based on the obtained results. Overall, the findings demonstrate that PSI effectively captures both regional and structure-specific deformation patterns. The study highlights the potential of PSI-based multi-temporal InSAR techniques for establishing continuous monitoring frameworks that support early warning, maintenance planning, and informed decision-making in urban renewal and infrastructure management in Istanbul.

Modal frequencies are widely used as damage-sensitive features in vibration-based unsupervised anomaly detection for full-scale civil structures. However, environmental variations introduce nonlinear and non-Gaussian behaviors in modal frequencies, characterized by multicentricity and substantial variance differences. These behaviors inevitably mask damage-induced frequency shifts, leading to increased false-detection rates and degraded real-time performance. To address this, this paper presents a novel unsupervised framework for early-stage anomaly detection that integrates exponential slow feature analysis (SFA) with just-in-time learning-aided standard K -nearest neighbor metrics. First, nonlinear modal features with minimum temporal variation are extracted from modal frequencies via exponential SFA for dimensionality reduction. Afterward, under the standard measurement criteria, the Euclidean squared distances between each sample in the entire training set and its nearest neighbors are computed. The standard distance for each neighbor is then modeled, with the number of neighbors automatically selected through Bayesian optimization and cross-validation. The primary purpose of this step is to ensure that the nearest neighbor distances between samples under different environmental conditions and damage conditions including abrupt anomalies or gradual degradations, are measured on a consistent scale, thereby minimizing environmental variability. After that, a robust detection indicator is derived from the mean of squared nearest neighbor distances and combined with kernel density estimation to determine an online detection threshold for outliers. The results indicate the proposed method effectively mitigates strong environmental variability without prior environmental measurements and accurately identifies anomalies in bridge modal frequencies with negligible false-positive and false-negative errors.

As key load-bearing elements, bridge piers are highly sensitive to concrete construction quality, which directly affects structural safety and durability. Surface defects such as color differences and bugholes are common and, if not properly assessed, may accelerate deterioration. To overcome the limitations of conventional inspection methods in precision, efficiency, and objectivity, this study proposes an intelligent framework integrating high-resolution panoramic imaging, deep semantic segmentation, and spatial color-difference analysis. The main contributions include: (1) A segment-based matrix image acquisition strategy is developed using unmanned aerial vehicles, combined with a DSeam-driven deep stitching algorithm to generate panoramic images of piers with geometric consistency and luminance continuity, serving as a unified visual basis for defect recognition. (2) For bughole segmentation, a PoreMamba-Net semantic network based on a state space modeling architecture is constructed. The network employs multi-branch convolution and selective scanning mechanisms to enhance fine-grained perception and contextual representation of small targets, enabling high-precision segmentation. A quantitative evaluation system composed of the apparent bughole area ratio, maximum bughole diameter, and bughole distribution standard deviation is further proposed to comprehensively characterize bughole quantity, size, and spatial uniformity. (3) For unbounded and non-structural color difference defects, a pixel-level analysis method is proposed by combining multi-scale retinex with color restoration-based illumination normalization and the CIE94 color difference model. Moran’s I spatial autocorrelation index is introduced to jointly quantify color deviation intensity and spatial consistency, supporting classification of color differences and detection of clustered anomalies. The proposed framework is validated on typical piers from the Shiziyang Channel Bridge construction project in Guangdong Province. Results demonstrate strong adaptability, robustness, and accuracy in complex field conditions, providing reliable technical support for concrete quality acceptance and lifecycle management.

Dynamic deflection monitoring with distributed photoelectric deflectometers (PDs) is pivotal to lightweight bridge structural health monitoring systems (SHMS). With few sensors yet theoretically unlimited monitoring markers, such systems maximize coverage under constrained investment; however, minimal redundancy makes them acutely vulnerable to sensor-level anomalies This vulnerability is amplified in small- and medium-span bridges where high-frequency, multi-channel streams must be fused in real time, demanding anomaly detection (AD) that is both robust to environmental variation and computationally efficient with minimal preprocessing. To address this, we propose SGFormer, a transformer-based AD framework featuring similarity-grouped fusion and dynamic prototype learning for multi-sensor deflection signals. A group-level feedforward module organizes channels with aligned temporal embeddings to exploit strong intra-span correlations while preserving cross-span differences, reducing redundancy and stabilizing fusion. Self-updating prototypes, learned directly from lightly preprocessed raw signals, encode operational and environmental effects, thereby avoiding heavy handcrafted calibration. We evaluate SGFormer using representative PD anomalies that reflect the essential characteristics of in-service bridge monitoring systems in lightweight SHMS. Experiments show that SGFormer consistently outperforms seven state-of-the-art baselines in unsupervised and zero-shot settings, achieving average F1-scores of 91.49% and 89.05%, with robustness across temporal granularities.

Timely and accurate fault diagnosis of suspension systems is paramount for ensuring the operational safety of rail vehicles. Recent years have witnessed extensive research in this field, primarily categorized into model-based and data-driven methods based on their underlying knowledge sources. Model-based methods rely on precise mathematical models to achieve fault detection and isolation via state or parameter estimation. Data-driven methods leverage historical and real-time data to extract fault features using statistical analysis, traditional machine learning, or deep learning techniques. This article presents a systematic review of these methods, offering a detailed comparison of their advantages and disadvantages, while summarizing current development trends and existing challenges. Finally, several future research directions are proposed, including model-data fusion diagnostic strategies, enhancement of real-time performance and robustness, enhancement of reliability prediction and uncertainty quantification, translation into engineering applications, and intelligent diagnostic techniques under small-sample conditions, aiming to provide references for the further development of suspension system fault diagnosis technology.

This paper proposes a new fault extraction method to address the challenge in rolling bearing fault diagnosis, where traditional denoising methods inadequately balance noise suppression and impulsive feature preservation. The method utilizes a discrete wavelet autoencoder (DWAE) model that incorporates local importance pooling (LIP) and a joint time–frequency (TF) optimization loss. First, a DWAE network with a TF decoupling mechanism constructs a mapping between noisy and clean signals in feature space, thereby enhancing its blind denoising capability. Second, an attention-based dynamic importance strategy for the LIP module enhances fault sparsity, suppresses noise, and preserves time-domain impulses. Finally, a multi-domain composite loss joint optimization that combines Pearson correlation coefficient loss for time-domain fidelity and Hilbert spectrum-based loss for frequency-domain feature preservation is used to improve diagnosis interpretability via TF joint supervision. Simulation and experimental results demonstrate that the method maintains strong denoising performance while significantly improving time-domain sparsity and enhancing fault features in the frequency domain and demodulation spectra, offering an effective solution for rotating machinery condition monitoring.

When the operating conditions of a hydraulic turbine bearing are dynamically adjusted under varying loads, it often leads to aggravated shaft vibrations and seal failures, which can result in unplanned downtime and substantial economic losses. To address this issue, a fault diagnosis and maintenance system, named signal large language model (LLM), has been developed. In terms of enhancing diagnostic reliability, a dual-channel fusion strategy, integrating both long-term and short-term time windows, is employed to simultaneously capture global and instantaneous features, improving diagnostic accuracy by 5% compared to previous methods. To improve maintenance efficiency, this study proposes a trusted maintenance planning strategy that combines LLMs with neural network-based communication, creating a closed-loop process from diagnosis to maintenance execution. Comparative experiments confirm the system’s superior performance in terms of maintenance efficiency.

Few-shot cross-domain bearing fault diagnosis faces two critical challenges: the scarcity of labeled training samples and significant distribution shifts between source and target domains, which severely degrade the generalization capability of diagnostic models. To address these issues, this study proposes a novel hyperspherical contrastive prototype adaptation (HSCPA) framework. The framework encompasses three key innovations. First, a hyperspherical feature extractor projects raw features onto a unit hypersphere, enforcing geometric constraints—unit normalization, ordered structure, and uniform distribution—to simultaneously enhance inter-class discriminability and intra-class compactness. Second, a contrastive prototype network is developed, integrating four complementary loss functions: intra-class compactness loss, query-prototype contrast loss, prototype uniformity loss, and prototype correlation loss. This design effectively mitigates prototype deviation and overcomes the limitations of conventional metric learning in few-shot scenarios. Third, a cross-domain feature alignment strategy based on Kullback–Leibler divergence is introduced to minimize inter-domain distribution discrepancy while promoting flat minima convergence for robust generalization. Extensive experiments under varying working conditions and equipment configurations demonstrate that the proposed HSCPA framework achieves state-of-the-art diagnostic accuracies of 98.81 and 94.67% in cross-condition and cross-machine tasks, respectively. Compared with eight benchmark methods, HSCPA exhibits superior performance, validating its efficacy for bearing fault diagnosis under data scarcity and domain shift conditions.

Conventional signal decomposition methods often rely on static time-domain indicators for filter design, overlooking the periodic frequency-domain structure of signals. This drawback conceals the spectral allocation of fault-related frequencies and prevents the comprehensive capture of diagnostic information. To address this issue, this article proposes v -Ramanujan spectral signal-to-noise ratio ( v-RSSNR )-driven impact spectrum aware mode decomposition (ISAMD), a method that introduces the v-RSSNR as a dynamic optimization criterion. During each iteration, the v -RSSNR index dynamically adjusts the generalized envelope exponent v , thereby adaptively guiding the update of filter parameters toward its optimal configuration. This enables continuous refinement of the filter’s response to enhance periodic impulse components in the frequency domain. Furthermore, a hybrid criterion combining cross-correlation and v-RSSNR is employed to eliminate redundant components, ensuring that the resulting modes retain highly concentrated periodic impact information. Experimental results based on both simulated signals and real rolling bearing fault data demonstrate that ISAMD achieves superior performance in extracting weak fault features under noisy conditions, confirming its effectiveness as an adaptive and robust decomposition framework for mechanical fault diagnosis.

Stochastic resonance (SR), with its unique noise utilization mechanism, has demonstrated remarkable advantages in the field of mechanical fault diagnosis. However, the performance of SR is highly dependent on the precise matching of noise intensity and system parameters. In practical engineering applications, noise is often uncontrollable and dynamically changing, which limits its applicability. By contrast, vibration resonance (VR) enhances weak signal detection by introducing a controllable external high-frequency excitation. The stiffness-softening effect induced by this excitation can modulate the potential-well characteristics, alter the system’s equilibrium states, and induce bifurcation, thereby effectively amplifying weak fault signals. Based on this, this paper proposes a potential-well parameter decoupled mechanical fault detection method based on time-delay feedback fractional VR (TDF-FVR). By independently adjusting the width and depth of the potential well, the method effectively optimizes the resonance effect to amplify weak fault signals. To achieve this, TDF-FVR system with tunable parameters is constructed, and a quantum genetic algorithm (QGA) is employed to adaptively optimize the system parameters, using the system’s response amplitude gain Q as the objective function to achieve optimal resonance and improve weak fault signal detection performance. Finally, bearing fault experiments verify the effectiveness of the proposed method, which shows excellent performance in weak fault signal detection. Furthermore, comparative results with the optimized wavelet packet transform, fast kurtogram, and feature modal decomposition methods further demonstrate the superiority of the proposed method.