As offshore wind turbines develop into deepwater operations, accurately quantifying the impact of stochastic excitation in complex sea environments on offshore wind turbines and conducting structural fatigue reliabili...
详细信息
As offshore wind turbines develop into deepwater operations, accurately quantifying the impact of stochastic excitation in complex sea environments on offshore wind turbines and conducting structural fatigue reliability analysis has become challenging. In this paper, based on long-term wind-wave reanalysis data from the South China Sea, a novel direct probability integral method (DPIM) is developed for the stochastic response and fatigue reliability analysis of the key components for the floating offshore wind turbine structures, under combined wind-wave excitation. A 5 MW floating offshore wind turbine is considered as the research object, and a comprehensive analysis of the wind turbine system is performed to assess the short-term fatigue damage at the tower base and blade root. The proposed method's accuracy and efficiency are validated by comparing the results to those obtained from Monte Carlo simulations (MCS) and a subset simulation (SSM). Additionally, a sensitivity analysis is conducted to evaluate the impact of different environmental parameters on fatigue damage, providing valuable insights for the design and operation of FOWTs in varying sea conditions. Furthermore, the results indicate that the fatigue life of floating offshore wind turbine (FOWT) structures under combined wind-wave excitation meets the design requirements. Notably, the fatigue reliability of the wind turbine under aligned wind-wave conditions is lower compared to misaligned wind-wave conditions.
Despite the similar global structures in Chest X-ray (CXR) images, the same anatomy exhibits varying appearances across images, including differences in local textures, shapes, colors, etc. Learning consistent represe...
详细信息
Despite the similar global structures in Chest X-ray (CXR) images, the same anatomy exhibits varying appearances across images, including differences in local textures, shapes, colors, etc. Learning consistent representations for anatomical semantics through these diverse appearances poses a great challenge for self-supervised pre-training in CXR images. To address this challenge, we propose two new pre-training tasks: inner-image anatomy localization (IIAL) and cross-image anatomy localization (CIAL). Leveraging the relatively stable positions of identical anatomy across images, we utilize position information directly as supervision to learn consistent semantic representations. Specifically, IIAL adopts a coarse-to-fine heatmap localization approach to correlate anatomical semantics with positions, while CIAL leverages feature affine alignment and heatmap localization to establish a correspondence between identical anatomical semantics across varying images, despite their appearance diversity. Furthermore, we introduce a unified end-to-end pre-training framework, anatomy-aware representation learning (AARL), integrating IIAL, CIAL, and a pixel restoration task. The advantages of AARL are: 1) preserving the appearance diversity and 2) training in a simple end-to-end way avoiding complicated preprocessing. Extensive experiments on six downstream tasks, including classification and segmentation tasks in various application scenarios, demonstrate that our AARL: 1) has more powerful representation and transferring ability;2) is annotation-efficient, reducing the demand for labeled data and 3) improves the sensitivity to detecting various pathological and anatomical patterns.
作者：
Gan, ZhihuaWu, ShihaoPang, ZilongChai, XiuliSong, YalinHenan Univ
Inst Intelligent Network Syst Intelligent Data Proc Engn Res Ctr Henan Prov Sch Software Kaifeng 475004 Peoples R China Henan Univ
Henan Engn Res Ctr Ind Internet Things Sch Artificial Intelligence Zhengzhou 450046 Peoples R China
Selective encryption algorithms have currently become an important method for protecting image privacy. Visual security evaluation of selective encrypted images plays an important role in measuring the effectiveness o...
详细信息
Selective encryption algorithms have currently become an important method for protecting image privacy. Visual security evaluation of selective encrypted images plays an important role in measuring the effectiveness of these algorithms, yet such studies are scarce. In this paper, we propose a visual security index combining superpixel segmentation and block variance calculation for selective encrypted images (SBVSI). Specifically, we propose a method based on superpixel segmentation and block variance calculation to find blocks of pixels that include valid information. These blocks can better represent the image regions of interest to the human visual system, thereby improving the evaluation accuracy. Next, to obtain features with stability, global features and local features of the image are extracted to represent the overall changes in the encrypted image. After that, the global similarity index and local similarity index are constructed using the above two features. Finally, the support vector regression model is used to integrate the similarity indices of global and local features, which effectively combines the information of different features and improves the accuracy and robustness of the visual security assessment. Experimental results on two public selective encrypted databases demonstrate that compared with existing state-of-the-art work, the proposed SBVSI exhibits better performance, especially in handling middle and high quality images.
Multimodal large language models (MLLMs) have demonstrated remarkable capabilities in various tasks. However, effectively evaluating these MLLMs on face perception remains largely unexplored. To address this gap, we i...
Post-stack data are susceptible to noise interference and have low resolution, which impacts the accuracy and efficiency of subsequent seismic data interpretation. To address this issue, we propose a deep learning app...
详细信息
Post-stack data are susceptible to noise interference and have low resolution, which impacts the accuracy and efficiency of subsequent seismic data interpretation. To address this issue, we propose a deep learning approach called Seis-SUnet, which achieves simultaneous random noise suppression and super-resolution reconstruction of seismic data. First, the Conv-Swin-Block is designed to utilize ordinary convolution and Swin transformer to capture the long-distance dependencies in the spatial location of seismic data, enabling the network to comprehensively comprehend the overall structure of seismic data. Second, to address the problem of weakening the effective signal during network mapping, we use a hybrid training strategy of L1 loss, edge loss and multi-scale structural similarity loss. The edge loss function directs the network training to focus more on the high-frequency information at the edges of seismic data by amplifying the weight. Additionally, the verification of synthetic and field seismic datasets confirms that Seis-SUnet can effectively improve the signal-to-noise ratio and resolution of seismic data. By comparing it with traditional methods and two deep learning reconstruction methods, experimental results demonstrate that Seis-SUnet excels in removing random noise, preserving the continuity of rock layers and maintaining faults as well as being strong robustness.
In this paper, we investigate the stability problem of 1-D wave equations with delayed feedback control on the boundary. By a delicate spectral analysis, the sufficient and necessary conditions for the feedback gain a...
详细信息
In this paper, we investigate the stability problem of 1-D wave equations with delayed feedback control on the boundary. By a delicate spectral analysis, the sufficient and necessary conditions for the feedback gain and the time delay are derived to guarantee the exponential stability of the closed-loop system. We discuss about all the situations for the time delay r > 0 including the case that r is irrational. The stability region of the feedback gain exists if and only if the time delay r is an even number. In this case, an explicit formula of the stability region is obtained accordingly and it characterizes the shrink of the stability region as r tends to infinity. In addition, we find that the small perturbation of magnitude in the time delay can only trigger the excitation of high frequency modes. That completely proves the judgement in [3, Page 5, Remark] and gives a mathematical explanation why numerical experiments usually do not demonstrate the non-robustness when a small perturbation is added to the time delay. (c) 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Unlike the post-buckling behaviors of classical piezoelectric cylindrical shell, the size-dependent effect of flexoelectric material and high strain gradient in the post-buckling process play an important role in the ...
详细信息
Unlike the post-buckling behaviors of classical piezoelectric cylindrical shell, the size-dependent effect of flexoelectric material and high strain gradient in the post-buckling process play an important role in the stability analysis of the micro/nano cylindrical shells. To reveal the impacts on the post-buckling of flexoelectric cylindrical shells, an accurate post-buckling model for the flexoelectric cylindrical shells under axial compression is proposed based on the higher-order shear deformation shell theory and von Karman geometrical nonlinearity. The size-dependent post-buckling equilibrium path with mode-jumping phenomena is obtained by using Galerkin's method and Newton-Raphson method. The predicted results are in agreement with those reported in the open literature. A detailed parametric study is also carried out to investigate the influence of geometrical parameters, flexoelectric coefficients, and electric voltage on the size-dependent post-buckling behaviors of flexoelectric cylindrical shells.
This paper introduces a memristor-based neural network with bidirectional cyclic constructed from square wave pulse functions. Additionally, we design an image encryption and compression network framework incorporatin...
详细信息
Pulse coding technology can effectively improve the performance of the phase-sensitive optical time domain reflectometry (Phi-OTDR) system, but the cross-correlation decoding for mass data along a long sensing distanc...
详细信息
Pulse coding technology can effectively improve the performance of the phase-sensitive optical time domain reflectometry (Phi-OTDR) system, but the cross-correlation decoding for mass data along a long sensing distance brings a lengthy extra time cost. In this paper, a pre-frequency-shift method is proposed for the lightweight demodulation in the random number coding Phi-OTDR system. By using the first-order sideband generated by the phase modulation, this method can effectively reduce the frequency of the beating signal, thus reducing the requirement of the data sampling rate. The principle of the cross-correlation decoding algorithm based on sequence interpolation is studied, and the extra computation caused by the cross-correlation operation is analyzed theoretically. In the experiment, the validity of the pre-frequency-shift scheme is verified, and the sampling rate of the system is reduced from 500 MSps to 100 MSps. Comparative experiments using conventional down-sampling techniques with sampling rates of 250 MSps, 125 MSps, and 60 MSps validate the superiority of the proposed method. Finally, the vibrations of different frequencies are successfully located at 62.24 km with a spatial resolution of 5 m. At this distance, the algorithmic time cost of the proposed method is only 13.28 % of that of the traditional scheme.
High-resolution remote sensing imagery is indispensable for applications ranging from environmental monitoring to national defense, where secure data transmission is paramount. In this study, we introduce a novel steg...
详细信息
High-resolution remote sensing imagery is indispensable for applications ranging from environmental monitoring to national defense, where secure data transmission is paramount. In this study, we introduce a novel steganography network specifically engineered for embedding remote sensing images within natural images, addressing the longstanding challenges of achieving high payload capacity, imperceptibility, and robustness. Our approach leverages an encoder-decoder architecture with independent encryption and decryption modules to reduce computational overhead and accelerate convergence. Key innovations include an enhanced Residual Convolutional Block Attention Module (Res-CBAM) and a Discrete Wavelet Transform (DWT)-based frequencydomain transformation, which together optimize feature extraction and reduce the visibility of secret image contours. A dual-stage training process further refines the model, with the first stage ensuring high-quality encryption and decryption and the second stage mitigating rounding errors to enable accurate recovery under lossy conditions. Comprehensive evaluations on standard benchmark datasets (DIV2K, COCO, and ImageNet) and our custom remote sensing-natural image dataset demonstrate that our method consistently outperforms 12 state-of-the-art techniques in terms of Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM), while achieving steganalysis detection accuracy near random guessing levels. These results underscore the model's superior performance, enhanced security, and broad applicability, making it a promising solution for robust, high-resolution image hiding in real-world scenarios.
暂无评论