Based on the sedimentary geochemical analysis of sediments from core ZK6 in the muddy sedimentary area of the Changjiang River Estuary and the collection of literature data, this study comprehensively evaluated the accumulation characteristics and influencing factors of four heavy metals (Cu, Pb, Zn, and Cr) in the sediments of the Changjiang River Estuary from interdecadal to millennial scales. The results of the enrichment factor (EF) and the index of geoaccumulation method indicate that the hazard quotient of heavy metals with the whole core is overall low. On the millennial scale (from ca. 1500 aBP to the present), the heavy metal contents were primarily determined by the changes of natural background values in the catchment, while Cu and Pb contents decreased and then increased from the bottom to the top, and Zn and Cr contents displayed decreasing trends. Before 1899, two historical periods, Tang and Song Dynasties, were mainly identified. Due to the social and economic prosperity and strong human activities during these periods, the degree of heavy metal enrichment fluctuated under certain influences. After 1899, the degree of heavy metal enrichment increased significantly due to the influence of global industrial development. On the centennial scale, the degree of heavy metal enrichment was significantly correlated with the stage of economic development in China. The accumulation degree of heavy metals was low before 1950 and intensified with the rapid industrial development from 1950 to 1990. The content of some elements declined due to the environmental protection policy after 1990. On the interdecadal scale (from 2003 to 2019), Pb and Cr element accumulation in sediments of the Changjiang River Estuary showed small variations after 2010, and Cu emission was controlled effectively, but the continuous Zn enrichment deserved more attention. This study revealed the natural and anthropogenic driving mechanisms of the spatial and temporal changes of heavy metals in the Changjiang River Estuary and provided scientific evidence for estuarine environmental management and governance.

This paper aims to investigate the regulatory role of RNA binding protein (RBP) fused in sarcoma (FUS) in the mechanism by which circ-ZNF609 (circBase ID: mmu_circ_0001797) alleviates cardiomyocyte apoptosis caused by ischemia/reperfusion injury (I/RI). Bioinformatics analysis revealed that FUS was a potential splicing factor binding to the flanking reverse complementary sequences of circ-ZNF609. Subsequently, an I/RI animal model and an oxygen-glucose deprivation/recovery (OGD/R) cell model were constructed. Methods such as terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) staining and quantitative reverse transcription polymerase chain reaction (qRT-PCR) were employed to validate that FUS can regulate the biogenesis of circ-ZNF609 in myocardial cells. The protective effect of FUS on OGD/R-induced cardiomyocyte apoptosis is reversed by circ-ZNF609 overexpression. Therefore, inhibiting FUS is a promising therapeutic strategy against the pathological process of myocardial ischemia/reperfusion injury.

This paper proposed a multirobot collaborative location algorithm based on a joint semantic constraint model using the characteristics of image semantics, which contained stable environment information. Aided by a semantic segmentation network and the ORB (oriented features from accelerated segment test (FAST) and rotated binary robust independent elementary feature (BRIEF)) extraction algorithm, the proposed algorithm obtained the semantic map points of scenes, used the semantic map points to construct the semantic error function, and constructed a joint semantic constraint model by combining semantic and geometric reprojection errors. Subsequently, the semantic label of feature points combined with feature bag of words (BOW) technology was used to estimate the relative pose of multiple robots, and the multirobot pose trajectory was unified based on the relative pose. Finally, the global pose was optimized by combining the joint semantic constraint model to realize multirobot collaborative localization. Experimental results showed that compared with the current mainstream multirobot collaborative localization algorithm，the proposed algorithm reduced the absolute pose error (APE) by 17.4%, thus demonstrating the applicability of the proposed algorithm to multirobot collaborative scenes.

The watermark image generated by the current watermarking algorithm exhibits poor robustness to non-differentiable noise processing and inadequate general image quality. This study introduces an improved model for a robust image watermarking based on generative adversarial networks. The model integrates the spatial and channel attention mechanism blocks into an encoder with dense structure type to enhance feature extraction ability. A high-pass filter is added to the front of the discriminator to enhance the imperceptibility of the generated watermark image. For end-to-end training, mini-batch of real and simulated (MBRS) JPEG processing is performed in the noise layer, adding various types of noise of different strengths. The experimental results show that compared with HiDDeN and two-stage deep learning robust watermarking model, the watermarked images generated by this model have better robustness and imperceptibility. The average peak signal-to-noise ratio (PSNR) increased by 1.20 dB; average structural similarity increased by 4.71%; and average information extraction bit error rate was reduced by 13.55%.

To improve the efficiency and accuracy of ultrasonic phased array (PA) scanning, a nondestructive testing approach that combines sector scanning and the total focusing method (TFM) is proposed with the assistance of a deep learning model, namely, YOLOv5s. Using this method, the self-trained YOLOv5s model was applied for preliminary defect recognition on sector scanning images to enable the rapid identification of possible defect regions. Software was developed for the total focusing method imaging system to image possible defects in possible defect regions based on the Qt platform. The results obtained from the test of the assessment block indicate that this new proposed approach can take advantage of not only the high imaging efficiency of ultrasonic phased array sector scanning but also the high accuracy of the total focusing method to achieve improvements in both test efficiency and accuracy.

Optical coherence tomography (OCT) uses a low-coherence optical source, and the images obtained are inevitably affected by scattering noise. To obtain OCT images with high signal-to-noise ratio and high resolution, a super-resolution reconstruction network model is proposed based on the generative adversarial network to simultaneously achieve denoising and super-resolution reconstruction of OCT images. This model is evaluated on an OCT image dataset and compared with some well-established models quantitatively and qualitatively. The results show that the average peak signal-to-noise ratio (PSNR) of this model is in the intermediate range and that the average similarity value of the learned perceptual image block is superior. This indicates that this model effectively recovers the image details and is conducive to the diagnosis of medical images, thus improving the accuracy of clinical diagnosis.

Multilabel learning is highly conducive to the accurate classification of remote sensing (RS) images that contain rich feature information, and the deep-learning-based multilabel classification of RS images has recently gained popularity owing to its exceptional performance. However, the categorical annotation of training samples in large-scale datasets is a time-consuming and labor-intensive task whose quality cannot be guaranteed, leading to the potential occurrence of missing labels. Incomplete label information can mislead the model training process, thereby negatively impacting classification performance. To address this problem, an incomplete multilabel classification method for RS images based on local semantics (LS) is proposed in this paper. First, a semantic representation learning (SRL) module is used to extract deep features from input images, which are then used to generate pseudo-labels. Next, a local semantic learning module is deployed to capture local semantic correlations within the features. These correlations are weighted and subsequently used to enhance the pseudo-labels. Finally, a weighted combination of pseudo-labels and enhanced pseudo-labels is used as the final predicted label, and the network is optimizes using the cross-entropy loss function. The experimental results demonstrate that the proposed method outperforms existing multilabel classification methods with different missing label rates.

Based on unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC), the flexible construction of line-of-sight links significantly improves the communication quality of a system and plays an important role in handling computationally intensive and latency-sensitive tasks. However, single UAV-assisted MEC suffers from a limited coverage range and long task processing time. This study considers multiple UAVs with MEC servers to provide offloading computing services for multiple ground users, and it presents the problem of maximizing the minimum number of computation bits. Under the constraints of limited energy consumption and no-fly zones, this study jointly optimized user scheduling, user upload power, task offloading time, local computation time, and UAV trajectory. An iterative optimization algorithm based on block coordinate descent was introduced to provide users with a fairer computation offloading service. The original problem is divided into four subproblems, and the non-convex subproblem is transformed into a convex optimization subproblem via successive convex approximations. The simulation results show that compared with other benchmark schemes, the proposed joint optimization scheme can significantly increase the number of maximum-minimum computation bits.

Spacecraft identification and tracking, which are core technologies in the aero-space field, are crucial for ensuring the safety and effectiveness of spacecraft operation in celestial bodies. However, in the space environment, dense stars interfere with the recognition of targets, and complex and variable lighting conditions may result in incomplete imaging, thereby affecting the clear presentation of targets. Therefore, a spacecraft recognition and positioning algorithm based on star-point recognition and multiframe aggregation is proposed in this study to accurately identify spacecraft targets within the proximity of a camera. This algorithm first maps star-point information to the spacecraft imaging plane via star-map recognition, thus eliminating dense interfering star points in the background of celestial bodies. Subsequently, a target-aggregation algorithm and a continuous-frame local-area localization method are introduced to address the incomplete spacecraft imaging and limited information of single-frame spacecraft due to lighting. Experimental results show that the proposed algorithm effectively eliminates more than 97% of interfering star points and satisfies the spacecraft positioning accuracy within a deviation range of $0.2^{\circ}$, thus demonstrating its high accuracy and robustness.

To solve the small-sample problem, this paper aimed to analyze the mechanism of spectral information transmission across network nodes and established a spectral method for cross-network node classification based on domain adaptation. In theory, this paper proposed an intermediate-frequency graph convolutional kernel to introduce intermediate-frequency information for a graph neural network (GNN) and then analyzed the properties of different frequency information. The results showed that unlike low-frequency and high-frequency information, intermediate-frequency information preserved similar features of nodes both in a single network and across networks. It was demonstrated that intermediate-frequency information was considered more suitable for cross-network node classification than low- frequency and high-frequency information. In practice, an intermediate frequency-domain adaptive graph convolutional network (IFDA-GCN) was proposed for classifying nodes across networks. IFDA-GCN relied on the intermediate-frequency graph convolutional kernel to extract intermediate-frequency information from the source and target networks, while leveraging domain adaptation techniques to mitigate the distributional shift between networks. Experimental results on real-world datasets demonstrated that IFDA-GCN outperformed baselines, and that intermediate-frequency information outperformed low-frequency and high-frequency information in cross-network node classification.

This paper proposed a satellite trajectory prediction method based on the idea of filtering. Drawing on the structure of long short-term memory（LSTM）networks, this method employed optimal estimation as the statistical method and began from the perspective of signal analysis, with a clear physical significance. Through comparative simulation experiments on short-arc orbit extrapolation, the performance differences between the wave model and the currently commonly used orbit extrapolation methods were verified. In the experiments, two wave models with different levels of systematic errors were used. The controlled variable experiments were conducted to discuss the effects of orbit altitude, observation accuracy, and observation arc length on forecast accuracy. For a short-arc observation data of three minutes, if the forecast duration is less than 10 minutes, the J2 wave model has comparable forecast accuracy to the dynamical extrapolation method and is superior to the Chebyshev extrapolation method for low earth orbit (LEO) targets, while the forecast accuracy of the J2 wave model is superior to that of the dynamical extrapolation method for geostationary earth orbit (GEO) targets.

The algal barrier at the water source was considered as the research object. The filtering and interception effects of the barrier on algae were investigated in details. It was explored that the main factors affecting the pollution interception effects of the barrier and the mechanism of the filtering and interception processes of the barrier through experimental simulation. The cleaning and recycling performances of the barrier were studied. By setting up an algal barrier in the field, the total amount of change was compared in algae before and after filtering by the barrier, and the actual algae removal ability of the barrier was verified. The experimental results showed that the water source containment had an interception effect on the algae in the water. There was a large difference in the total amount of algae before and after the containment. In the range of 0$\sim$0.03 m$^3$/(m$^2\cdot$s), the higher the through-water flux, the lower was the algae removal efficiency. Beyond 0.03 m$^3$/(m$^2\cdot$s), the removal rate was less than 5%. In the 10$\sim$60 NTU range, the higher the turbidity, the better was the algae removal effect. Chlorophyll a concentration did not have a significant effect on the perimeter filter interception. The filter cake filtration model was the main pollution interception mechanism in perimeter filtration and had good cleaning and recycling abilities.

We introduce two kinds of new variations of dominant dimensions, which have some advantages in the study of gendo-Gorenstein algebras. One is the $\nu$-stably dominant dimension associated to the Nakayama functor $\nu$. Using it, a criterion for an algebra being gendo-Gorenstein is given, and the gendo-Gorensteiness of algebras are invariant under left-split extensions. Moreover, the difference of the $\nu$-stably dominant dimension is bounded for two derived equivalent gendo-Gorentein algebras. The other is a dominant dimension building from the Gorenstein balance module. An upper bound of this kind of dominant dimension is given for gendo-Gorenstein algebras.

For the two-dimensional distributed-order time fractional diffusion equation with a variable coefficient in this paper, a Gauss integral approximates the distributed-order operator $D^\omega_t u$ and original problem, which is transformed into a multi-term time fractional differential equation. The nonconforming $EQ_1^{\rm rot}$ and zero-order Raviart-Thomas (R-T) elements are employed in a spatial direction, the modified L1 scheme is applied in a temporal direction, the fully discrete scheme of the equation is established, and the stability of the fully discrete scheme is then demonstrated. Using the interpolation operator $\Pi_h$, $I_h$ and projection operator $R_h$, of the elements, the superclose results of the variable $u$ in $H^1$-norm and intermediate variable $\overrightarrow{p}=\hbar (X)\nabla u$ in $L^2$-norm are obtained, respectively. Finally, the global superconvergence results are derived by using the related properties of the interpolation operators $I_{2h}$ and $\Pi_{2h}$.

In this paper, we study the linearizability for the complex 1:-4 resonant quadratic system, and we obtain the necessary and sufficient conditions for the linearizibility of the system.