Search Results (45)

The conductors of high-speed railway OCSs (Overhead Contact Systems) are susceptible to conductor galloping due to the impact of natural elements such as strong winds, rain, and snow, resulting in conductor fatigue damage and significantly compromising train operational safety. Consequently, monitoring the galloping status of conductors is crucial, and instance segmentation techniques, by delineating the pixel-level contours of each conductor, can significantly aid in the identification and study of galloping phenomena. This work expands upon the YOLO11-seg model and introduces an instance segmentation approach for galloping video and image sensor data of OCS conductors. The algorithm, designed for the stripe-like distribution of OCS conductors in the data, employs four-direction Sobel filters to extract edge features in horizontal, vertical, and diagonal orientations. These features are subsequently integrated with the original convolutional branch to form the FDSE (Four Direction Sobel Enhancement) module. It integrates the ECA (Efficient Channel Attention) mechanism for the adaptive augmentation of conductor characteristics and utilizes the FL (Focal Loss) function to mitigate the class-imbalance issue between positive and negative samples, hence enhancing the model’s sensitivity to conductors. Consequently, segmentation outcomes from neighboring frames are utilized, and mask-difference analysis is performed to autonomously detect conductor galloping locations, emphasizing their contours for the clear depiction of galloping characteristics. Experimental results demonstrate that the enhanced YOLO11-seg model achieves 85.38% precision, 77.30% recall, 84.25% AP@0.5, 81.14% F1-score, and a real-time processing speed of 44.78 FPS. When combined with the galloping visualization module, it can issue real-time alerts of conductor galloping anomalies, providing robust technical support for railway OCS safety monitoring. Full article

Non-active current in the time domain is considered for application to the diagnostics and classification of loads in power grids based on waveform-distortion characteristics, taking as a working example several recordings of the pantograph current in an AC railway system. Data are processed with a deep autoencoder for feature extraction and then clustered via k-means to allow identification of patterns in the latent space. Clustering enables the evaluation of the relationship between the physical meaning and operation of the system and the distortion phenomena emerging in the waveforms during operation. Euclidean distance (ED) is used to measure the diversity and pertinence of observations within pattern groups and to identify anomalies (abnormal distortion, transients, …). This approach allows the classification of new data by assigning data to clusters based on proximity to centroids. This unsupervised method exploiting non-active current is novel and has proven useful for providing data with labels for later supervised learning performed with the 1D-CNN, which achieved a balanced accuracy of 96.46% under normal conditions. ED and 1D-CNN methods were tested on an additional unlabeled dataset and achieved 89.56% agreement in identifying normal states. Additionally, Grad-CAM, when applied to the 1D-CNN, quantitatively identifies the waveform parts that influence the model predictions, significantly enhancing the interpretability of the classification results. This is particularly useful for obtaining a better understanding of load operation, including anomalies that affect grid stability and energy efficiency. Finally, the method has been also successfully further validated for general applicability with data from a different scenario (charging of electric vehicles). The method can be applied to load identification and classification for non-intrusive load monitoring, with the aim of implementing automatic and unsupervised assessment of load behavior, including transient detection, power-quality issues and improvement in energy efficiency. Full article

The integrity and stability of railway fasteners are of vital importance to railway safety. To address the challenges of limited anomaly samples, irregular defect geometries, and complex operational conditions in rail fastener anomaly detection, this paper proposes an unsupervised anomaly detection method using a knowledge-distilled generative adversarial network. First, the proposed method employs collaborative teacher–student learning to model normal sample distributions, where the student network reconstructs input images as normal outputs while a discriminator identifies anomalies by comparing input and reconstructed images. Second, a multi-scale attention-coupling feature-enhancement mechanism is proposed, effectively integrating hierarchical semantic information with spatial-channel attention to achieve both precise target localization and robust background suppression in the teacher network. Third, an enhanced anomaly discriminator is designed to incorporate an enhanced pyramid upsampling module, through which fine-grained details are preserved via multi-level feature map aggregation, resulting in significantly improved sensitivity for small-sized anomaly detection. Finally, the proposed method achieved an AUC of 94.0%, an ACC of 92.5%, and an F1 score of 91.6% on the MNIST dataset, and an AUC of 94.7%, an ACC of 90.1%, and an F1 score of 87.8% on the railway fastener dataset, which proves the superior anomaly detection ability of this method. Full article

The Building Information Modeling (BIM) concept has been recently implemented in railway infrastructure, assisting mainly in the project elaboration, and further, the facility management aspect. The present study addresses the inspection activity of the railway geometry, in a BIM context, using a rigorous modeling process of the railway track components, and the development of a Dynamo script for the evaluation of the degree of geometric irregularity detected during inspection works. The monitoring phase of the rail tracks involves a planned railway inspection schedule, normally supported by human analyses of data collected in a railway geometric inspection. The created script allows for evaluating the inspection data and categorizes the data by alert levels that are associated with a color code, visualized over the railway components of the BIM model. The Dynamo script uses new BIM parameters considering the maintenance activity, allowing for analyzing inspection data and visualizing the colored alerts. This capacity alerts the maintenance engineer about the urgency of planning a retrofitting action, according to the severity level of the detected geometric anomaly. An illustrative real railway track segment is considered supporting the modeling process, the inspection data collection and the efficiency analyses of the script application. This research intends to contribute to increment knowledge of BIM adoption in railway infrastructures, emphasizing the potential of using Dynamo programming on BIM model database management. Full article

With the ease of acquiring RGB-D images from line-scan 3D cameras and the development of computer vision, anomaly detection is now widely applied to railway inspection. As 2D anomaly detection is susceptible to capturing condition, a combination of depth maps is now being explored in industrial inspection to reduce these interferences. In this case, this paper proposes a novel approach for RGB-D anomaly detection called Dual-Branch Cross-Fusion Normalizing Flow (DCNF). In this work, we aim to exploit the fusion strategy for dual-branch normalizing flow with multi-modal inputs to be applied in the field of track detection. On the one hand, we introduce the mutual perception module to acquire cross-complementary prior knowledge in the early stage. On the other hand, we exploit the effectiveness of the fusion flow to fuse the dual-branch of RGB-D inputs. We experiment on the real-world Track Anomaly (TA) dataset. The performance evaluation of DCNF on TA dataset achieves an impressive AUROC score of 98.49%, which is 3.74% higher than the second-best method. Full article

Railway tight and wide gauges are critical factors affecting the safety and reliability of railway systems. Undetected tight and wide gauges can lead to derailments, posing significant risks to operations and passenger safety. This study explores a novel approach to detecting railway tight and wide gauges by integrating accelerometer data, machine-learning techniques, and building information modeling (BIM). Accelerometers installed on axle boxes provide real-time dynamic data, capturing anomalies indicative of tight and wide gauges. These data are processed and analyzed using supervised machine-learning algorithms to classify and predict potential tight- and wide-gauge events. The integration with BIM offers a spatial and temporal framework, enhancing the visualization and contextualization of detected issues. BIM’s capabilities allow for the precise mapping of tight- and wide-gauge locations, streamlining maintenance workflows and resource allocation. Results demonstrate high accuracy in detecting and predicting tight and wide gauges, emphasizing the reliability of machine-learning models when coupled with accelerometer data. This research contributes to railway maintenance practices by providing an automated, data-driven methodology that enhances the proactive identification of tight and wide gauges, reducing the risk of derailments and maintenance costs. Additionally, the integration of machine learning and BIM highlights the potential for comprehensive digital solutions in railway asset management. Full article

This study presents a comparative analysis of three AutoEncoder (AE) models—Variational AutoEncoder (VAE), Sparse AutoEncoder (SAE), and Convolutional AutoEncoder (CAE)—to detect and quantify structural anomalies in railway vehicle wheels, such as polygonization. Vertical acceleration data from a virtual wayside monitoring system serve as input for training the AE models, which are coupled with Hotelling’s T² Control Charts to differentiate normal and abnormal railway component behaviors. The results indicate that the SAE-T² model outperforms its counterparts, achieving 16.67% higher accuracy than the CAE-T² model in identifying distinct structural conditions, although with a 35.78% higher computational cost. Conversely, the VAE-T² model is outperformed in 100% of the analyzed scenarios when compared to SAE-T² in identifying distinct structural conditions while also exhibiting a 21.97% higher average computational cost. Across all scenarios, the SAE-T² methodology consistently provided better classifications of wheel damage, showing its capability to extract relevant features from dynamic signals for Structural Health Monitoring (SHM) applications. These findings highlight SAE’s potential as an interesting tool for predictive maintenance, offering improved efficiency and safety in railway operations. Full article

A model transformation method based on state refinement and semantic mapping is proposed to address the challenges of high modeling complexity and resource consumption in symbolic validation of industrial software requirements. First, a rule-based semantic mapping system is constructed through the explicit definition of element correspondence between statechart components and verification models, coupled with a composite state-level refinement strategy to structurally optimize model hierarchy. Second, an automated transformation algorithm is developed to bridge graphical modeling tools with formal verification environments, supported by quantitative evaluation metrics for mapping validity. To demonstrate its practical applicability, the methodology is systematically applied to railway infrastructure safety—specifically the railroad turnout control system—as a critical case study. The experimental implementation converts operational statecharts of turnout control logic into optimized NuSMV models. Not only did the models remain intact, but the state space was also effectively reduced through the optimization of the hierarchical structure. In the validation phase, the converted model is tested for robustness using the fault injection method, and boundary condition anomalies that are not explicitly stated in the requirement specification are successfully detected. The experimental results show that the validation model generated by this method has improved validation efficiency in the NuSMV tool, which is significantly better than the traditional conversion method. Full article

Cables and tendons are crucial elements in bridge engineering but also are vulnerable structural elements because they are usually subjected to fatigue and corrosion problems. Thus, vibration-based non-destructive techniques have been used for external post-tensioning tendon assessment. Regarding continuous monitoring systems, tendon assessment is carried out through the continuous tracking of its natural frequencies and the subsequent estimation of the tension force, as this parameter is essential for the bridge’s overall structural performance, thus providing useful information about bridge safety. However, for long-term monitoring assessment, two main challenges have to be addressed regarding practical applications: (i) double-peak spectra and other spurious factors that affect the frequency estimation, and (ii) temperature dependency, which needs to be carefully treated since frequency/tension variation may be explained by temperature variation, thus masking potential structural anomalies. On this subject, this paper presents the experimental long-term monitoring of several post-tensioning external tendons in a high-speed railway bridge in which a sectorized weighted peak-picking frequency identification procedure is proposed for frequency estimation, alongside a cascade clustering process, which allows meaningful frequency estimates to be selected. Finally, the selected frequency estimates, which show variations from 1 to 2% for all analyzed frequencies, are used for the long-term assessment of the tension force. Full article

The abnormal structural state of the pantograph skateboard is a significant and highly concerning issue that has a significant impact on the safety of high-speed railway operation. In order to obtain real-time information on the abnormal state of the skateboard in advance, an intelligent defect identification model suitable to be used as a monitoring device for the pantograph skateboard was designed using a computer vision-based intelligent detection technology for pantograph skateboard defects, combined with an improved YOLO v8 model and traditional image processing algorithms such as edge extraction. The results show that the anomaly detection algorithm for the pantograph sliding plate structure has good robustness, maintaining recognition accuracy of 90% or above in complex scenes, and the average runtime is 12.32 ms. Railway field experiments have proven that the intelligent recognition model meets the actual detection requirements of railway sites and has strong practical application value. Full article

In an ever expanding railway network all around the world, the need for track maintenance grows steadily. Traditionally, one major part of track maintenance is ramming large vibrating steel picks into the gravel between and under railway sleepers to compress the gravel and generate a safe substructure. Even today, maintenance personnel still have to manually locate the sleepers if they cannot be detected by computer vision systems or visually by the operator. Here we developed a first of its kind magnetic sleeper detector, even able to find sleepers, buried in gravel, undetectable by vision based systems. Our approach of magnetic detection is based on a DC magnetic field excitation and a detector moving with respect to the rail system, including the sleepers and fasteners for mounting the rails. Due to railway application constraints a large air gap between the sensor and the sleeper structure is required, which significantly complicates the magnetic sensing task for robust sleeper detection. The design and optimization of the magnetic circuit was based on extensive 3D simulation studies to ensure highest possible variation in magnetic flux density at the sensor locations for absence and presence of a sleeper. Furthermore, a low noise and high sensitivity electronic circuit has been realized to cope with sensor signal offsets from unknown or changing sensor orientations with respect to the earth’s magnetic field, or magnetic interferences from other trains potentially passing by during active measurements. Since we only want to detect sleepers in close vicinity of the moving sensor system, digital signal processing of the acquired signals can easily compensate for disturbing slowly changing or static field components within real world application scenarios. We demonstrate that magnetic detection of even buried sleepers on railway tracks is possible for distances of up to 172 mm between the sensor and the sleeper. This enables an even higher level of railway maintenance automation previously impossible in certain scenarios. Full article

The straddle-type monorail is an electric-powered public vehicle widely known for its versatility and ease of maintenance. The finger-band is a critical connecting structure for the straddle-type monorail, but issues such as loose bolts are inevitable over time. Manual inspection is the primary method for detecting bolt looseness in the finger-band, but this approach could be more efficient and resistant to missed detections. In this study, we conducted a straddle-type monorail finger-band-anomaly-monitoring experiment using Distributed Acoustic Sensing (DAS), a distributed multi-point-monitoring system widely used in railway monitoring. We analyzed track vibration signals’ time-domain and frequency-domain characteristics under different monorail operating conditions. Our findings revealed the following: 1. DAS can effectively identify the monorail’s operating status, including travel direction, starting and braking, and real-time train speed measurement. 2. Time-domain signals can accurately pinpoint special track structures such as turnouts and finger-bands. Passing trains over finger-bands also results in notable energy reflections in the frequency domain. 3. After the finger-band bolts loosen, there is a significant increase in vibration energy at the finger-band position, with the degree of energy increase corresponding to the extent of loosening. Full article

The impact of the quality of electricity on the pantograph is an important parameter for the supply of locomotives in railway companies (RCs). The subject of this research is the analysis of the quality of electricity on the pantograph of the 441-series locomotivelocated at distances of 1 km or 35 km from the power station in the electric traction system of Serbian Railways. The analysis included the simulation of the system in the MATLAB-Simulink software package (R2016a), which resulted in data that were often difficult to measure due to the complexity of the electric traction system. The obtained values indicate that the total harmonic voltage distortion on the pantograph of the 441 locomotive is 16.34% for 1 km and 51.06% for 35 km, while the EN 50160 standard prescribes a maximum of 8%. The total harmonic distortion current in the electric traction substation and through the locomotive pantograph is 33.42% (for 1 km) and 32.53% (for 35 km), showing anomalies in the supply of locomotives in RCs. Full article

In central Sakha (Yakutia) Republic, groundwater icings, primarily formed by intrapermafrost water, are less prone to contamination and serve as a stable freshwater resource. The periodic growth of icings threatens infrastructure such as roads, railways, and bridges in permafrost areas. Therefore, research in this field has become urgently necessary. This study aims to analyze the impacts of various factors on the scale of icing formation using Landsat satellite data, Gravity Recovery and Climate Experiment (GRACE)/GRACE Follow-On (GRACE-FO) data, Global Land Data Assimilation System (GLDAS) data, and field observation results. The results showed that the surface area of icings in the study area showed an overall increasing trend from 2002 to 2022, with an average growth rate of 0.06 km²/year. Suprapermafrost water and intrapermafrost water are the main sources of icings in the study area. The total Groundwater Storage Anomaly (GWSA) values from October to April showed a strong correlation with the maximum icing areas. Icings fed by suprapermafrost water were influenced by precipitation in early autumn, while those fed by intrapermafrost water were more affected by talik size and distribution. Climate warming contributed to the degradation of the continuous permafrost covering an area of 166 km² to discontinuous permafrost, releasing additional groundwater. This may also be one of the reasons for the observed increasing trend in icing areas. This study can provide valuable insights into water resource management and infrastructure construction in permafrost regions. Full article