Search

Accidents caused by vehicles changing lanes occur frequently on highways. Moreover, frequent lane changes can severely impact traffic flow during peak commuting hours and on busy roads. A novel framework based on a multi-relational graph convolutional network (MR-GCN) is herein proposed to address these challenges. First, a dynamic multilevel relational graph was designed to describe interactions between vehicles and road objects at different spatio-temporal granularities, with real-time updates to edge weights to enhance understanding of complex traffic scenarios. Second, an improved spatio-temporal interaction graph generation method was introduced, focusing on spatio-temporal variations and capturing complex interaction patterns to enhance prediction accuracy and adaptability. Finally, by integrating a dynamic multi-relational graph convolutional network (DMR-GCN) with dynamic scene sensing and interaction learning mechanisms, the framework enables real-time updates of complex vehicle relationships, thereby improving behavior prediction’s accuracy and real-time performance. Experimental validation on multiple benchmark datasets, including KITTI, Apollo, and Indian, showed that our algorithmic framework achieves significant performance improvements in vehicle behavior prediction tasks, with Map, Recall, and F1 scores reaching 90%, 88%, and 89%, respectively, outperforming existing algorithms. Additionally, the model achieved a Map of 91%, a Recall of 89%, and an F1 score of 90% under congested road conditions in a self-collected high-speed traffic scenario dataset, further demonstrating its robustness and adaptability in high-speed traffic conditions. These results show that the proposed model is highly practical and stable in real-world applications such as traffic control systems and self-driving vehicles, providing strong support for efficient vehicle behavior prediction.

In intelligent traffic monitoring systems, the significant distance between cameras and their non-overlapping fields of view leads to several issues. These include incomplete tracking results from individual cameras, difficulty in matching targets across multiple cameras, and the complexity of inferring the global trajectory of a target. In response to the challenges above, a deep learning-based vehicle tracking algorithm called FairMOT-MCVT is proposed. This algorithm con-siders the vehicles’ characteristics as rigid targets from a roadside perspective. Firstly, a Block-Efficient module is designed to enhance the network’s ability to capture and characterize image features across different layers by integrating a multi-branch structure and depth-separable convolutions. Secondly, the Multi-scale Dilated Attention (MSDA) module is introduced to improve the feature extraction capability and computational efficiency by combining multi-scale feature fusion and attention mechanisms. Finally, a joint loss function is crafted to better distinguish between vehicles with similar appearances by combining the trajectory smoothing loss and velocity consistency loss, thereby considering both position and velocity continuity during the optimization process. The proposed method was evaluated on the public UA-DETRAC dataset, which comprises 1210 video sequences and over 140,000 frames captured under various weather and lighting conditions. The experimental results demonstrate that the FairMOT-MCVT algorithm significantly enhances multi-target tracking accuracy (MOTA) to 79.0, IDF1 to 84.5, and FPS to 29.03, surpassing the performance of previous algorithms. Additionally, this algorithm expands the detection range and reduces the deployment cost of roadside equipment, effectively meeting the practical application requirements.

The yellowfin tuna represents a significant fishery resource in the Pacific Ocean. Its resource endowment status and spatial variation mechanisms are intricately influenced by marine environments, particularly under varying climate events. Consequently, investigating the spatial variation patterns of dominant environmental factors under diverse climate conditions, and understanding the response of yellowfin tuna catch volume based on the spatial heterogeneity among these environmental factors, presents a formidable challenge. This paper utilizes comprehensive 5°×5° yellowfin tuna longline fishing data and environmental data, including seawater temperature and salinity, published by the Western and Central Pacific Fisheries Commission (WCPFC) and the Inter-American Tropical Tuna Commission (IATTC) for the period 2000–2021 in the Pacific Ocean. In conjunction with the Niño index, a multiscale geographically weighted regression model based on geographically weighted principal component analysis (GWPCA-MGWR) and spatial association between zones (SABZ) is employed for this study. The results indicate the following: (1) The spatial distribution of dominant environmental factors affecting the catch of Pacific yellowfin tuna is primarily divided into two types: seawater temperature dominates in the western Pacific Ocean, while salinity dominates in the eastern Pacific Ocean. When El Niño occurs, the area with seawater temperature as the dominant environmental factor in the western Pacific Ocean further extends eastward, and the water layers where the dominant environmental factors are located develop to deeper depths; when La Niña occurs, there is a clear westward expansion in the area with seawater salinity as the dominant factor in the eastern Pacific Ocean. This change in the spatial distribution pattern of dominant factors is closely related to the movement of the position of the warm pool and cold tongue under ENSO events. (2) The areas with a higher catch of Pacific yellowfin tuna are spatially associated with the dominant environmental factor of mid-deep seawater temperature (105–155 m temperature) to a greater extent than other factors, the highest correlation exceeds 70%, and remain relatively stable under different ENSO events. The formation of this spatial association pattern is related to the vertical movement of yellowfin tuna as affected by subsurface seawater temperature. (3) The GWPCA-MGWR model can fully capture the differences in environmental variability among subregions in the Pacific Ocean under different climatic backgrounds, intuitively reflect the changing areas and influencing boundaries from a macro perspective, and has a relatively accurate prediction on the trend of yellowfin tuna catch in the Pacific Ocean.

The role that underwater image translation plays assists in generating rare images for marine applications. However, such translation tasks are still challenging due to data lacking, insufficient feature extraction ability, and the loss of content details. To address these issues, we propose a novel multi-scale image translation model based on style-independent discriminators and attention modules (SID-AM-MSITM), which learns the mapping relationship between two unpaired images for translation. We introduce Convolution Block Attention Modules (CBAM) to the generators and discriminators of SID-AM-MSITM to improve its feature extraction ability. Moreover, we construct style-independent discriminators that enable the discriminant results of SID-AM-MSITM to be not affected by the style of images and retain content details. Through ablation experiments and comparative experiments, we demonstrate that attention modules and style-independent discriminators are introduced reasonably and SID-AM-MSITM performs better than multiple baseline methods.

The ultra-rapid satellite clock product based on the satellite clock batch estimation is commonly used for high-precision and reliable precise point positioning (PPP) services. In order to clarify the effect of different ranging errors on the satellite clock batch estimation accuracy, the source of the satellite clock bias induced by the batch observation model is classified into the initial clock bias (ICB) and time-dependent bias (TDB). In addition to the effect of the ICB and TDB, the analytic relationship between the observation redundancy and the satellite clock batch estimation accuracy are derived and verified. The suitable number of stations is suggested to be 40 for the satellite clock batch estimation to achieve the counterbalance between the efficiency and saturable accuracy. For the PPP based on the batch-estimated satellite clock, the impacts of the ICB and TDB on PPP are clarified. The satellite clock batch estimation and PPP experiments are carried out to investigate the impacts of the ICB and TDB on the satellite clock batch estimation accuracy and the PPP performance. The ICB causes a significant bias for the batch-estimated satellite clock. The TDB is impacted by the assimilation ability of the batch-estimated satellite clock to the satellite orbit error. The convergence time and the positioning accuracy after the convergence of PPP are primarily affected by the ICB and TDB, respectively.

The construction of underwater geodetic networks (UGN) is crucial in marine geodesy. To provide high-precision kinematic positioning for underwater submersibles, an underwater acoustic geodetic network configuration of three seafloor base stations, one subsurface buoy, and one sea surface buoy is proposed. The simulation results show that, for a 3 km-deep sea, based on the proposed UGN, the submersible positioning range and positioning accuracy are primarily affected by the size of the seafloor base station array, while the height of the subsurface buoy has a greater impact on the submersible positioning accuracy than the positioning range. Considering current acoustic ranging technology, the kinematic positioning performance of the UGN is optimal when the seafloor base stations are 9~13 km apart and the subsurface buoy is less than 2.5 km above the seafloor, which can achieve a submersible positioning accuracy of less than 30 m within an underwater space of 25 km × 25 km × 3 km. The proposed cost-effective UGN configuration can provide high-precision submersible kinematic positioning performance for seafloor surveying and ocean precision engineering. The impact of the underwater environment on the acoustic transmission characteristics should be further investigated.

The periodic noise exists in BeiDou navigation satellite system (BDS) clock offsets. As a commonly used satellite clock prediction model, the spectral analysis model (SAM) typically detects and identifies the periodic terms by the Fast Fourier transform (FFT) according to long-term clock offset series. The FFT makes an aggregate assessment in frequency domain but cannot characterize the periodic noise in a time domain. Due to space environment changes, temperature variations, and various disturbances, the periodic noise is time-varying, and the spectral peaks vary over time, which will affect the prediction accuracy of the SAM. In this paper, we investigate the periodic noise and its variations present in BDS clock offsets, and improve the clock prediction model by considering the periodic variations. The periodic noise and its variations over time are analyzed and quantified by short time Fourier transform (STFT). The results show that both the amplitude and frequency of the main periodic term in BDS clock offsets vary with time. To minimize the impact of periodic variations on clock prediction, a time frequency analysis model (TFAM) based on STFT is constructed, in which the periodic term can be quantified and compensated accurately. The experiment results show that both the fitting and prediction accuracy of TFAM are better than SAM. Compared with SAM, the average improvement of the prediction accuracy using TFAM of the 6 h, 12 h, 18 h and 24 h is in the range of 6.4% to 10% for the GNSS Research Center of Wuhan University (WHU) clock offsets, and 11.1% to 14.4% for the Geo Forschungs Zentrum (GFZ) clock offsets. For the satellites C06, C14, and C32 with marked periodic variations, the prediction accuracy is improved by 26.7%, 16.2%, and 16.3% for WHU clock offsets, and 29.8%, 16.0%, 21.0%, and 9.0% of C06, C14, C28, and C32 for GFZ clock offsets.

Global Navigation Satellite System—Acoustic (GNSS-A) positioning is the main technique for seafloor geodetic positioning. A transceiver lever arm offset and sound velocity bias in seawater are the main systematic errors of the GNSS-A positioning technique. Based on data from a sea trial in shallow water, this paper studies the functional model of GNSS-A positioning. The impact of the two systematic errors on seafloor positioning is analysed and corresponding processing methods are proposed. The results show that the offset in the lever arm measurement should be parameterised in the observation equation. Given the high correlation between the vertical lever arm offset and the vertical coordinate of the seafloor station, a sample search method was introduced to fix the vertical offset correction. If the calibration of the sound velocity profiler cannot be ensured, the correction parameter of the sound velocity bias should be solved. According to the refined functional model and corrections, the position of a seafloor station in shallow water can be determined with a precision of better than 1 cm.

Hierarchical structures and organizational models highly affect the sustainable operation of the spatial network of tourism destinations. This paper adopted the modified tourism gravity model and social network analysis method to reveal the spatial characteristics of the tourism network on the Tibetan Plateau based on tourism flow data in 2019. The results are as follows: (1) the spatial organization of tourism flows was unbalanced, showing the characteristics of “high in the east and low in the west”; (2) The county tourism flow formed a multidimensional system of spatial hierarchy with Chengguan District of Lhasa, Golmud City, and Pengzhou City as the top nodes, the spatial polarization characteristics, and zonal distribution differences were evident; (3) The inter-county tourism linkage conformed to the law of distance decay, and the multiple network structure hierarchy features highlighted the complex linkage pattern; (4) The tourism network had a more distinctive dominant flow and was influenced by county-level administrative divisions; (5) The tourism linkage network formed eight subgroups with significant geographical characteristics. This study provides recommendations for optimizing the spatial structure of the Tibetan Plateau county tourism network for the government.

The proton transfer mechanism on the carbon cathode surface has been considered as an effective way to boost the electrochemical performance of Zn-ion hybrid supercapacitors (SCs) with both ionic liquid and organic electrolytes. However, cheaper, potentially safer, and more environmental friendly supercapacitor can be achieved by using aqueous electrolyte. Herein, we introduce the proton transfer mechanism into a Zn-ion hybrid supercapacitor with the ZnSO4 aqueous electrolyte and functionalized activated carbon cathode materials (FACs). We reveal both experimentally and theoretically an enhanced performance by controlling the micropores structure and hydrogen-containing functional groups (−OH and −NH functions) of the activated carbon materials. The Zn-ion SCs with FACs exhibit a high capacitance of 435 F g−1 and good stability with 89% capacity retention over 10,000 cycles. Moreover, the proton transfer effect can be further enhanced by introducing extra hydrogen ions in the electrolyte with low pH value. The highest capacitance of 544 F g−1 is obtained at pH = 3. The proton transfer process tends to take place preferentially on the hydroxyl-groups based on the density functional theory (DFT) calculation. The results would help to develop carbon materials for cheaper and safer Zn-ion hybrid SCs with higher energy.