Search Results (120)

The converter valve is the core component of the ultra-high voltage direct current (UHVDC) transmission system, and its fault detection is very important to ensure the safe and stable operation of the transmission system. However, the voiceprint signals collected by converter stations under complex operating conditions are often affected by background noise, spikes, and nonlinear interference. Traditional methods make it difficult to achieve high-precision detection due to the lack of feature extraction ability and poor noise robustness. This paper proposes a fault-aware variational self-encoder model (FAVAE-AS) based on a weak correlation between attention and self-supervised learning. It extracts probability features via a conditional variational autoencoder, strengthens feature representation using multi-layer convolution and residual connections, and introduces a weak correlation attention mechanism to capture global time point relationships. A self-supervised learning module with six signal transformations improves generalization, while KL divergence-based correlation inconsistency quantization with dynamic thresholds enables accurate anomaly detection. Experiments show that FAVAE-AS achieves 0.925 accuracy in fault detection, which is 5% higher than previous methods, and has strong robustness. This research provides critical technical support for UHVDC system safety by addressing converter valve acoustic anomaly detection. It proposes an extensible framework for industrial intelligent maintenance. Full article

This paper presents a robust hardware system designed for future detection of faults in wind turbines by analyzing audible noise signals. Predictive maintenance strategies have increasingly relied on acoustic monitoring as a non-invasive method for identifying anomalies that may indicate component wear, misalignment, or impending mechanical failures. The proposed device captures and processes sound signals in real-time using strategically positioned microphones, ensuring high-fidelity data acquisition without interfering with turbine operation. Signal processing techniques are applied to extract relevant acoustic features, facilitating future identification of abnormal sound patterns that may indicate mechanical issues. The system’s effectiveness was validated through rigorous field tests, demonstrating its capability to enhance the reliability and efficiency of wind turbine maintenance. Experimental results showed an average transmission latency of 131.8 milliseconds, validating the system’s applicability for near real-time audible noise monitoring in wind turbines operating under limited connectivity conditions. Full article

Growing concern over anthropogenic underwater noise, highlighted by initiatives like the Marine Strategy Framework Directive (MSFD) and its Technical Group on Underwater Noise (TG Noise), emphasizes regions like the Western Black Sea, where increasing activities threaten marine habitats. This region is experiencing rapid growth in maritime traffic and resource exploitation, which is intensifying concerns over the noise impacts on its unique marine habitats. While machine learning offers promising solutions, a research gap persists in comprehensively evaluating diverse ML models within an integrated framework for complex underwater acoustic data, particularly concerning real-world data limitations like class imbalance. This paper addresses this by presenting a multi-faceted framework using passive acoustic monitoring (PAM) data from fixed locations (50–100 m depth). Acoustic data are processed using advanced signal processing (broadband Sound Pressure Level (SPL), Power Spectral Density (PSD)) for feature extraction (Mel-spectrograms for deep learning; PSD statistical moments for classical/unsupervised ML). The framework evaluates Convolutional Neural Networks (CNNs), Random Forest, and Support Vector Machines (SVMs) for noise event classification, alongside Gaussian Mixture Models (GMMs) for anomaly detection. Our results demonstrate that the CNN achieved the highest classification accuracy of 0.9359, significantly outperforming Random Forest (0.8494) and SVM (0.8397) on the test dataset. These findings emphasize the capability of deep learning in automatically extracting discriminative features, highlighting its potential for enhanced automated underwater acoustic monitoring. Full article

Termites pose significant threats to the structural integrity of embankments due to their nesting and tunneling behavior, which leads to internal voids, water leakage, or even dam failure. This review systematically classifies and evaluates current termite detection techniques in the context of embankment maintenance, focusing on physical sensing technologies and biological characteristic-based methods. Physical sensing methods enable non-invasive localization of subsurface anomalies, including ground-penetrating radar, acoustic detection, and electrical resistivity imaging. Biological characteristic-based methods, such as electronic noses, sniffer dogs, visual inspection, intelligent monitoring, and UAV-based image analysis, are capable of detecting volatile compounds and surface activity signs associated with termites. The review summarizes key principles, application scenarios, advantages, and limitations of each technique. It also highlights integrated multi-sensor frameworks and artificial intelligence algorithms as emerging solutions to enhance detection accuracy, adaptability, and automation. The findings suggest that future termite detection in embankments will rely on interdisciplinary integration and intelligent monitoring systems to support early warning, rapid response, and long-term structural resilience. This work provides a scientific foundation and practical reference for advancing termite management and embankment safety strategies. Full article

This study investigates pre-seismic ionospheric anomalies preceding the 2023 Jishishan Ms6.2 earthquake using total electron content (TEC) data derived from BDS geostationary orbit (GEO) satellites. Multi-scale analysis integrating Butterworth filtering and wavelet transforms resolved TEC disturbances into three distinct frequency regimes: (1) high-frequency perturbations (0.56–3.33 mHz) showed localized disturbances (amplitude ≤ 4 TECU, range < 300 km), potentially associated with near-field acoustic waves from crustal stress adjustments; (2) mid-frequency signals (0.28–0.56 mHz) exhibited anisotropic propagation (>1200 km) with azimuth-dependent N-shaped waveforms, consistent with the characteristics of acoustic–gravity waves (AGWs); and (3) low-frequency components (0.18–0.28 mHz) demonstrated phase reversal and power-law amplitude attenuation, suggesting possible lithosphere–atmosphere–ionosphere (LAI) coupling oscillations. The stark contrast between near-field residuals and far-field weak fluctuations highlighted the dominance of large-scale atmospheric gravity waves over localized acoustic disturbances. Geometry-based velocity inversion revealed incoherent high-frequency dynamics (5–30 min) versus anisotropic mid/low-frequency traveling ionospheric disturbance (TID) propagation (30–90 min) at 175–270 m/s, aligning with theoretical AGW behavior. During concurrent G1-class geomagnetic storm activity, spatial attenuation gradients and velocity anisotropy appear primarily consistent with seismogenic sources, providing insights for precursor discrimination and contributing to understanding multi-scale coupling in seismo-ionospheric systems. Full article

The sound speed profile (SSP) is an important feature in the field of ocean acoustics. The accurate estimation of SSP is significant for the development of underwater position, communication, and associated fundamental marine research. The Res-SACNN model is proposed for SSP inversion based on the convolutional neural network (CNN) embedded with the residual network and self-attention mechanism. It combines the spatiotemporal characteristics of sea level anomaly (SLA) and sea surface temperature anomaly (SSTA) data and establishes a nonlinear relationship between satellite remote sensing data and sound speed field by deep learning. The single empirical orthogonal function regression (sEOF-r) method is used in a comparative experiment to confirm the model’s performance in both the time domain and the region. Experimental results demonstrate that the proposed model outperforms sEOF-r regarding both spatiotemporal generalization ability and inversion accuracy. The average root mean square error (RMSE) is decreased by 0.92 m/s in the time-domain experiment in the South China Sea, and the inversion results for each month are more consistent. The optimization ratio hits 71.8% and the average RMSE decreases by 7.39 m/s in the six-region experiment. The Res-SACNN model not only shows more superior inversion ability in the comparison with other deep-learning models, but also achieves strong generalization and real-time performance while maintaining low complexity, providing an improved technical tool for SSP estimation and sound field perception. Full article

Situation awareness (SA) involves analyzing sensory data, such as audio signals, to identify anomalies. While acoustic features are widely used in audio analysis, existing methods face critical limitations; they often overlook the relevance of SA audio segments, failing to capture the complex relational patterns in audio data that are essential for SA. In this study, we first propose a graph neural network (GNN) with an attention mechanism that models SA audio features through graph structures, capturing both node attributes and their relationships for richer representations than traditional methods. Our analysis identifies suitable audio feature combinations and graph constructions for SA tasks. Building on this, we introduce a situation awareness gated-attention GNN (SAGA-GNN), which dynamically filters irrelevant nodes through max-relevance neighbor sampling to reduce redundant connections, and a learnable edge gated-attention mechanism that suppresses noise while amplifying critical events. The proposed method employs sigmoid-activated attention weights conditioned on both node features and temporal relationships, enabling adaptive node emphasizing for different acoustic environments. Experiments reveal that the proposed graph-based audio features demonstrate superior representation capacity compared to traditional methods. Additionally, both proposed graph-based methods outperform existing approaches. Specifically, owing to the combination of graph-based audio features and dynamic selection of audio nodes based on gated-attention, SAGA-GNN achieved superior results on two real datasets. This work underscores the importance and potential value of graph-based audio features and attention mechanism-based GNNs, particularly in situational awareness applications. Full article

This paper introduces an acoustic-based monitoring system for high-speed CNC drilling, aimed at optimizing processes and enabling real-time machine state detection. High-fidelity acoustic sensors capture sound signals during drilling operations, allowing the identification of critical events such as tool engagement, material breakthrough, and tool withdrawal. Advanced signal processing techniques, including spectrogram analysis and Fast Fourier Transform, extract dominant frequencies and acoustic patterns, while machine learning algorithms like DBSCAN clustering classify operational states such as cutting, breakthrough, and returning. Experimental studies on materials including acrylic, PTFE, and hardwood reveal distinct acoustic profiles influenced by material properties and drilling conditions. Smoother sound patterns and lower dominant frequencies characterize PTFE drilling, whereas hardwood produces higher frequencies and rougher patterns due to its density and resistance. These findings demonstrate the correlation between acoustic emissions and machining dynamics, enabling non-invasive real-time monitoring and predictive maintenance. As AI power increases, it is expected to extract in-situ process information and achieve higher resolution, enhancing precision in data interpretation and decision-making. A key contribution of this project is the creation of an open sound library for drilling processes, fostering collaboration and innovation in intelligent manufacturing. By integrating big data concepts and intelligent algorithms, the system supports continuous monitoring, anomaly detection, and process optimization. This AI-ready hardware enhances the accuracy and efficiency of drilling operations, improving quality, reducing tool wear, and minimizing downtime. The research establishes acoustic monitoring as a transformative approach to advancing CNC drilling processes and intelligent manufacturing systems. Full article

Anomaly detection in oil and gas pipelines based on acoustic signals currently faces challenges, including limited anomalous samples, varying audio data distributions across different operating conditions, and interference from background noise. These challenges lead to reduced accuracy and efficiency in pipeline anomaly detection. The primary challenge in reconstruction-based pipeline audio anomaly detection is to prevent the loss of critical information and ensure the high-quality reconstruction of feature maps. This paper proposes a pipeline anomaly detection method termed Multi-scale Feature Fusion GANomaly with Dilated Neighborhood Attention. Firstly, to mitigate information loss during network deepening, a Multi-scale Feature Fusion module is proposed to merge the encoded and decoded feature maps at different dimensions, enhancing low-level detail and high-level semantic information. Secondly, a Dilated Neighborhood Attention module is introduced to assign varying weights to neighborhoods at various dilation rates, extracting channel interactions and spatial relationships between the current pixel and its neighborhoods. Finally, to enhance the quality of the reconstructed spectrum, a loss function based on the Structure Similarity Index Measure is designed, considering both pixel-level and structural differences to maintain the structural characteristics of the reconstructed spectrum. MFDNA-GANomaly achieved 92.06% AUC, 93.96% Accuracy, and 0.955 F1-score on the test set, demonstrating that the proposed method can effectively enhance pipeline anomaly detection performance. Additionally, MFDNA-GANomaly exhibited competitive performance on the ToyTrain and Bearing subsets of the development dataset in the DCASE Challenge 2023 Task 2, confirming the generalization capability of the model. Full article

The quality of the processed products in CNC machining centers is a critical factor in manufacturing equipment. The anomaly detection and predictive maintenance functions are essential for improving efficiency and reducing time and costs. This study aims to strengthen service competitiveness by reducing quality assurance costs and implementing AI-based predictive maintenance services, as well as establishing a predictive maintenance system for CNC manufacturing equipment. The proposed system integrates preventive maintenance, time-based maintenance, and condition-based maintenance strategies. Using continuous learning based on long short-term memory (LSTM), the system enables anomaly detection, failure prediction, cause analysis, root cause identification, remaining useful life (RUL) prediction, and optimal maintenance timing decisions. In addition, this study focuses on roller-cutting devices that are essential in packaging processes, such as food, pharmaceutical, and cosmetic production. When rolling pins are machining with CNC equipment, a sensor system is installed to collect acoustic data, analyze failure patterns, and apply RUL prediction algorithms. The AI-based predictive maintenance system developed ensures the reliability and operational efficiency of CNC equipment, while also laying the foundation for a smart factory monitoring platform, thus enhancing competitiveness in intelligent manufacturing environments. Full article

This work uses recurrence plots (RPs) to identify nonlinearities and non-stationary conditions in wind turbines. Traditionally, recurrence plots have been applied to vibration or acoustic data; this paper applies them to magnetometer and accelerometer data to compare the sensitivity. The recurrence plots are generated by plotting points in the phase space and identifying those points where the dynamic system returns to a similar configuration, meaning that the state variables are similar to previous conditions. The state variables for the acceleration data are the position and velocity, whereas, for the magnetometer data, they are the magnitude of the magnetic field and its integral. The time series are integrated by combining the shifting principle of harmonic functions and the empirical mode decomposition. The EMD method separates the original signal into several modes, shifts them, and combines them back. The time series were obtained from an accelerometer and a magnetometer mounted in a wind turbine. The results showed that the RP presents different patterns depending on the signal; magnetometer signals identify low-frequency components, such as magnetic field anomalies, and accelerometer signals identify high-frequency components, such as bearings and gears. Full article

This study proposes a framework for anomaly detection in industrial machines with a focus on robust multiclass classification using acoustic data. Many state-of-the-art methods only have binary classification capabilities for each machine, and suffer from poor scalability and noise robustness. In this context, we propose the use of Smoothed Pseudo Wigner–Ville Distribution-based Mel-Frequency Cepstral Coefficients (SPWVD-MFCCs) in the framework which are specifically tailored for noisy environments. SPWVD-MFCCs, with better time–frequency resolution and perceptual audio features, improve the accuracy of detecting anomalies in a more generalized way under variable signal-to-noise ratio (SNR) conditions. This framework integrates a CNN-LSTM model that efficiently and accurately analyzes spectral and temporal information separately for anomaly detection. Meanwhile, the dimensionality reduction strategy ensures good computational efficiency without losing critical information. On the MIMII dataset involving multiple machine types and noise levels, it has shown robustness and scalability. Key findings include significant improvements in classification accuracy and F1-scores, particularly in low-SNR scenarios, showcasing its adaptability to real-world industrial environments. This study represents the first application of SPWVD-MFCCs in industrial diagnostics and provides a noise-robust and scalable method for the detection of anomalies and fault classification, which is bound to improve operational safety and efficiency within complex industrial scenarios. Full article

The sound speed in the ocean has a considerable impact on the characteristics of underwater acoustic propagation. The swift gathering of the underwater three-dimensional (3D) sound speed field is essential for target detection, underwater acoustic communication, and navigation. Currently, the reconstruction of the underwater sound speed utilizing satellite remote sensing data of the sea surface has emerged as a significant area of research. However, dynamic activities within the ocean result in varying degrees of perturbation in the sound speed structure. Relying solely on sea surface information will restrict the accuracy of sound speed reconstruction. In response to this issue, by utilizing multi-source satellite remote sensing data alongside Argo profiles, we first implemented the random forest (RF) algorithm to establish the statistical mapping relationship from the sea surface temperature (SST), sea level anomaly (SLA), and absolute dynamic topography (ADT) to the density, and thus, reconstructed a 3D density field. Subsequently, based on the sea surface environmental information, we introduced the underwater vertical density as a novel input for sound speed calculations and proposed a new model for 3D sound speed field reconstruction (RF-SDR). The experimental results indicate that utilizing both the sea surface environmental variables and underwater density as inputs yielded an average root-mean-square error (RMSE) of 1.51 m/s for the reconstructed sound speed, along with an average mean absolute error (MAE) of 0.85 m/s. Following the incorporation of density into the reconstruction inputs, the two error metrics exhibited reductions of 31% and 35%, respectively. And the proposed RF-SDR model demonstrated a reduction in the RMSE by 36% and in the MAE by 43% when compared with the commonly utilized single Empirical Orthogonal Function regression (sEOF-r) method. Furthermore, simulations of the sound propagation with both the reconstructed sound speed and Argo sound speed demonstrated a high degree of consistency in the computed acoustic propagation losses. The correlation coefficients consistently exceeded 0.7, thereby reinforcing the validity of the reconstructed sound speed. Full article

This study investigates the potential of deploying a neural network model on an advanced programmable logic controller (PLC), specifically the Finder Opta™, for real-time inference within the predictive maintenance framework. In the context of Industry 4.0, edge computing aims to process data directly on local devices rather than relying on a cloud infrastructure. This approach minimizes latency, enhances data security, and reduces the bandwidth required for data transmission, making it ideal for industrial applications that demand immediate response times. Despite the limited memory and processing power inherent to many edge devices, this proof-of-concept demonstrates the suitability of the Finder Opta™ for such applications. Using acoustic data, a convolutional neural network (CNN) is deployed to infer the rotational speed of a mechanical test bench. The findings underscore the potential of the Finder Opta™ to support scalable and efficient predictive maintenance solutions, laying the groundwork for future research in real-time anomaly detection. By enabling machine learning capabilities on compact, resource-constrained hardware, this approach promises a cost-effective, adaptable solution for diverse industrial environments. Full article

As medical imaging continues to expand, concerns about the potential risks of ionizing radiation to the developing fetus have led to a preference for non-radiation-based alternatives such as ultrasonography and fetal MRI. This review examines the current evidence on the safety of MRI during pregnancy, with a focus on 3 T MRI and contrast agents, aiming to provide a comprehensive synthesis that informs clinical decision-making, ensures fetal safety and supports the safe use of all available modalities that could impact management. We conducted a comprehensive review of studies from 2000 to 2024 on MRI safety during pregnancy, focusing on 3 T MRI and gadolinium use. The review included peer-reviewed articles and large database studies, summarizing key findings and identifying areas for further research. Fetal MRI, used alongside ultrasound, enhances diagnostic accuracy for fetal anomalies, particularly in the brain, thorax, gastrointestinal and genitourinary systems, with no conclusive evidence of adverse effects on fetal development. While theoretical risks such as tissue heating and acoustic damage exist, studies show no significant harm at 1.5 T or 3 T, though caution is still advised in the first trimester. Regarding gadolinium-based contrast agents, the evidence is conflicting: while some studies suggest risks such as stillbirth and rheumatological conditions, animal studies show minimal fetal retention and no significant toxicity, and later clinical research has not substantiated these risks. The existing literature on fetal MRI is encouraging, suggesting minimal risks; however, further investigation through larger, prospective and long-term follow-up studies is essential to comprehensively determine its safety and late effects. Full article