Search

We report a photothermally-induced liquid-solid/gas-solid-decoupling photocatalytic water-splitting system, where a carbonized melamine foam (CMF) and a porous g-C3N4(PCN) serve as the photothermal substrate and model photocatalyst, respectively. Specifically, liquid water is transformed into the gaseous phase over the CMF due to the photothermal effect, and the generated vapor can be split into hydrogen by PCN via the photocatalysis. This unique biphasic photocatalytic system exhibits a high hydrogen production rate of 368.1 µmol h−1, which is 2.4 and 25.6 times larger than those of the traditional triphasic PCN system (151.7 µmol h−1) and g-C3N4(CN) system (14.4 µmol h−1), respectively. The improved photocatalytic performance is mainly attributed to the optimized energy and mass transfer at the gas-liquid-solid reaction interface, where gas products are rapidly desorbed in the photocatalytic process. This work provides a novel strategy to enhance the photocatalytic performance from the perspectives of energy and mass flow.

In this article, in order to improve the tracking ability of traditional active disturbance rejection control (ADRC) to the reference signal, the parallel multiple extended state observers (multi-ESOs) based ADRC is proposed. First, the way in which the estimation error of the ESO affects the tracking error of closed-loop system is studied, and it is shown that the tracking accuracy is not completely consistent with the estimation accuracy. Next, a novel switching ADRC approach with parallel multi-ESOs is proposed, in which the control law is constructed by automatically selecting different ESOs' estimations from the perspective of reducing tracking error, instead of reducing estimation error. Furthermore, the stability of the proposed algorithm is rigorously proved under some reasonable assumptions. In addition, the proposed ADRC is applied to the high-speed precision motion stage, which has complicated nonlinear uncertainties and elastic deformation disturbances. The experimental results show that the parallel multi-ESOs based ADRC has higher tracking performance than the traditional single ESO based ADRC.

Emerging evidence indicates a significant role of remnant cholesterol in contributing to the residual risk associated with major adverse cardiovascular events (MACE). This study aims to evaluate the dose–response relationship between remnant cholesterol and the risk of MACE. PubMed, Embase and Cochrane databases were reviewed to identify cohort studies published in English up to 1 August 2023. Twenty-eight articles were selected. Pooled hazard ratios (HR) and their 95% confidence intervals (CIs) were calculated using fixed or random-effects models to evaluate the association between remnant cholesterol and the risk of MACE. The dose–response relationship between remnant cholesterol levels and the risk of MACE was analyzed using the linear model and restricted cubic spline regression models. For calculated remnant cholesterol levels, the pooled HR (95% CI) of MACE for per 1-SD increase was 1.13 (1.08, 1.17); HR (95% CI) for the second quartile (Q2), the third quartile (Q3) and the highest quartile (Q4) of remnant cholesterol levels were 1.14 (1.03, 1.25), 1.43 (1.23, 1.68) and 1.68 (1.44, 1.97), respectively, compared with the lowest quartile (Q1). For measured remnant cholesterol levels, the pooled HR (95% CI) of MACE per 1-SD increase was 1.67 (1.39, 2.01). The dose–response meta-analysis showed a dose–response relationship between remnant cholesterol levels and the risk of MACE, both on a linear trend (P < 0.0001) and a nonlinear trend (P < 0.0001). The risk of MACE is associated with increased levels of remnant cholesterol, and the dose–response relationship between remnant cholesterol levels and the risk of MACE showed both linear and nonlinear trends.

Currently, photodynamic therapy (PDT) is restricted by the laser penetration depth. Except for PDT at 1064 nm wavelength excitation, the development of other NIR-II-activated nanomaterials with a higher response depth is still hindered and rarely reported in the literature. To overcome these problems, we fabricated a nanoplatform with heterostructures that generate reactive oxygen species (ROS) and ferrite nanoparticles under a high concentration of zinc doping (ZnxFe3–xO4NPs), which can achieve oxidative damage of tumor cells under near-infrared (NIR) illumination. The recombination of photoelectrons and holes has been markedly inhibited due to the formation of heterostructures in the interfaces, thus greatly enhancing the capability for ROS and oxygen production by modulating the single-component doping content. The efficiency of PDT was verified by in vivoand in vitroassays under NIR light. Our results revealed that NIR-II (1208 nm) light irradiation of ZnxFe3–xO4NPs exerted a remarkable antitumor activity, superior to NIR-I light (808 nm). More importantly, the reported ZnxFe3–xO4NPs strategy provides an opportunity for the success of comparison with light in the first and second near-infrared regions.

Temperature and pressure of natural gas hydrate (NGH) reservoirs influence CH4recovery and CO2sequestration through the CO2–CH4hydrate replacement reaction. The typical temperature and pressure in marine NGH reservoirs are mainly 275.65–294.51 K and 8.41–22.25 MPa, respectively. However, the corresponding CO2–CH4hydrate replacement reaction under the temperature above 281.15 K has been studied little in previous studies. In this work, six experiments were conducted at 281.15–289.15 K and 15–17 MPa using a self-developed experimental apparatus. These experiments utilized techniques such as general photography, microscopy, temperature and pressure measurements, computational analysis, and gas chromatography to study the dynamic changes in the morphology and distribution of CH4, CO2, and H2O under varying temperatures, pressures, and compositional conditions. The results showed that liquid CO2replaced CH4in CH4hydrates or CH4–CO2mixed hydrates. This replacement process was featured with multiple mechanisms during the soaking stage and governed by the CH4–CO2mixed hydrate phase equilibrium, which was controlled by CH4–CO2mixtures and their compositions. Moreover, the relative increases of the CH4molar ratio in the CO2-rich fluid phase, the efficiency of CO2sequestration, and the CO2sequestration amount in comparison with the CH4reserves reached 24.97%, 19.53%, and 2.17 times, respectively. CO2injection pressure and the initial volume fraction of liquid water and the CO2-rich fluid were found to have larger impacts on the CO2–CH4hydrate replacement reaction above 281.15 K than temperature and the initial volume of CH4hydrate. Besides, controlling the composition of the CH4–CO2–H2O mixed system and depressurization were proposed to balance CH4recovery and CO2sequestration. This work may offer some references for CH4recovery and CO2sequestration under real settings.

Electrochemiluminescence (ECL) luminophores with wavelength-tunable multicolor emissions are essential for multicolor ECL imaging detection and multiplexed analysis. In this work, silver nanoparticle (Ag NP)-decorated graphitic carbon nitride (g-CN@Ag) nanocomposites were synthesized. The morphology, chemical composition, structure, and ECL property of g-CN@Ag were investigated. The prepared g-CN, g-CN@Ag1, g-CN@Ag5, and g-CN@Ag10 can produce blue, blue-green, chartreuse, and yellow colored ECL emissions, respectively, by using K2S2O8as the coreagent. The ECL emission wavelength of g-CN@Ag can be regulated from 460 to 565 nm by modulating the content of the immobilized Ag NPs. Then, a multicolor ECL detection array was fabricated by using g-CN, g-CN@Ag1, g-CN@Ag5, and g-CN@Ag10 as four ECL luminophores. Dopamine was detected based on its inhibition effect on the multicolor ECL emissions. The linear range is from 0.1 nM to 1 mM with the lowest detection limit of 44 pM. Then, machine learning-assisted multiparameter concentration prediction of dopamine was further carried out by combining the deep neural network (DNN) algorithm. This work provides a new avenue to regulate the ECL emission wavelength of g-CN by using the metal nanoparticle modification strategy and presents an effective machine learning-assisted multicolor ECL detection strategy for accurate multiparameter quantitative detection.

Excessive water production impacted marine natural gas hydrate production. Cationic polyacrylamide (CPAM), one kind of chemical reagent, was used for the first time to selectively reduce water production while having less impact on gas production. After quartz sands with an absolute permeability of 100–1000 mD were treated with CPAM, the water phase permeability was reduced to 1/15, while the gas phase permeability almost had no change. Furthermore, CPAM was effective in controlling water flow in quartz sands with an absolute permeability range of 100–1000 mD. The water phase permeability initially decreased and then stabilized as the concentration of CPAM increased from 500 to 20,000 ppm. Additionally, the formation of methane hydrate in pretreated quartz sands with CPAM led to a greater reduction in quartz sands’ permeability than the sum of the respective reduction for CPAM treatment and methane hydrate formation. In addition, compared to quartz sands treated only with CPAM, the dissociation of 5% methane hydrate significantly increased the permeability of CPAM-treated quartz sands, but the permeability was smaller than that before CPAM treatment.

The trajectory planning module generates optimal collision-free paths, which is essential to automated driving. Existing approaches have focused on traffic flow situations and the geometrical feasibility of local trajectories. However, high nonlinearity of vehicle dynamics could cause handling instability or even severe accidents in cases of tracking dynamic infeasible paths. Meanwhile, the variety and complexity of driving environments bring intractable challenges to safe trajectory planning. This article aims to address this crucial issue and proposes a chassis global dynamics-oriented trajectory planning scheme. This work designs comprehensive performance indices to represent handling dynamic status based on chassis dynamics modeling. As the basis of trajectory planning, a reference path is extracted from global digital map data, and the curvilinear coordinate frame is exploited. Numerical optimization is proposed to solve the trajectory planning problem. Thus, an optimal and dynamically feasible trajectory, which satisfies geometrical smoothness, chassis global dynamics indices, and feasible safety, can be generated and transformed. Various scenarios are involved in carrying out simulation tests, and those results demonstrate the excellent capability and effectiveness of the proposed scheme to provide optimal trajectories in a variety of driving situations.

This paper focuses on point-based single-stage 3D object detection from point clouds and proposes a novel elegant detector CPC-3Det. Pyramid and confidence-guided backbones are widely used in point-based methods. However, the limitation of neighborhood points and negative sample construction bring obstacles to the discriminative feature learning and cost. Also, Scene-level spatial information loss should be noted. This paper presents the repository-based backbone consisting of a feature repository and partial knowledge to meet the issues. Additionally, explicit class-aware statistics are designed to raise robust features. Moreover, statistics-embedded detection heads through feature modulation and parameter control enhance CPC-3Det performance. Furthermore, The misalignment in IoU optimization caused by center offset is explored in this paper. The paper proposes a center-weighted IoU and designs hybrid losses to drive network parameter optimization. Extensive experiments on both the KITTI and Waymo Open datasets demonstrate the superiority of CPC-3Det over state-of-the-art methods.

The effect of different post-harvest storage pressures on the quality of strawberries was investigated. Strawberries were stored at 10 kPa, 50 kPa, and 90 kPa. The alterations in quality indicators were recorded, including weight loss, vitamin C, titratable acid, breathing intensity, and sensory scores. Kinetic models were used to fit the quality indicators. The results showed that the shelf life of strawberries refrigerated was only 6 days. The shelf life of strawberries at 50 kPa and 5 °C was 12 days. On the 15th day, weight loss, vitamin C content, titratable acidity content, and respiration intensity of strawberries stored at 50 kPa and 5 °C were 1.73%, 82 mg/100 g, 0.54%, and 32 mg/(kg h), respectively. When compared with strawberries under normal atmospheric conditions (the control), hypobaric storage maintained the texture and flavor of strawberries and inhibited structural depression and mold. 50 kPa is the optimum storage pressure for strawberries. Degradation of quality indicators followed zero-order and first-order kinetic models, with vitamin C and respiration intensity of strawberries stored under reduced pressure being more suitable for fitting using the latter. The coefficients of determination were 0.78497–0.99817 and 0.74064–0.98621, respectively. The kinetic model developed in this study revealed the changes in quality indicators during strawberry storage, which can effectively assess the nutritional value of the fruit.

This paper proposes a robust image encryption algorithm that utilizes a Novel three-dimensional (3D) Chaotic map and an Enhanced Logistic Sine System (ELSS). We leverage the unpredictability of 3D chaotic dynamics alongside the complexity of ELSS and DNA Sequence to forge a formidable image encryption scheme. Firstly, the image pixels are converted from decimal to hexadecimal notation and sorted in a 1D pixel array carrying a unique sequence of three channels of the RGB image. Secondly, the secret key is appended to XOR, the values with that 1-D pixels array. Thirdly, values are sorted by performing the binary right shift operation and encoded into DNA. Fourthly, a novel chaotic map is used to perform scrambling operations. Lastly, a novel enormous keyspace ELSSis used to perform efficient Border and Cross-Border (B&CB) pixel exchange, further enhancing the encryption quality of the proposed algorithm. Comprehensive security analysis proved that the proposed algorithm exhibits remarkable resilience against powerful known and chosen plaintext attacks and other prevalent cryptanalysis attacks, including differential attacks and exhaustive key search attacks. Henceforth, the proposed algorithm’s superior security and low computational cost make it an ideal choice for real-time secure image communication across various platforms, including satellite, multimedia, and military communications.

Research on electronic skin (e-skin) is dedicated to simulating natural skin for the perception of external mechanical stimuli. Currently, e-skin is ineffective in analyzing a single stimulus from different directions. This work successfully fabricates an integrated electronic skin (IES) with biaxial sensing capability through the combination of a biphasic liquid metal and porous foam. Remarkably different from traditional e-skin, the IES can analyze the type, strength, and area of an external mechanical stimulus from vertical and horizontal dimensions with a dual response (capacitive and resistive change, respectively). As a multifunctional sensor, the IES simultaneously responds to compression viacapacitive change and tension viaresistive change. Furthermore, 1000 cyclic compressions were conducted to confirm the electrical stability of the IES. Very subtle stimuli (e.g.thawing ice and touch) can be detected by the IES viabiaxial detection. This work provides a new protocol for the development of future intelligent flexible electronics.

Designing a query-efficient attack strategy to generate high-quality adversarial examples under the hard-label black-box setting is a fundamental yet challenging problem, especially in natural language processing (NLP). The process of searching for adversarial examples has many uncertainties (e.g., an unknown impact on the target model's prediction of the added perturbation) when confidence scores cannot be accessed, which must be compensated for with a large number of queries. To address this issue, we propose TextCheater, a decision-based metaheuristic search method that performs a query-efficient textual adversarial attack task by prohibiting invalid searches. The strategies of multiple initialization points and Tabu search are also introduced to keep the search process from falling into a local optimum. We apply our approach to three state-of-the-art language models (i.e., BERT, wordLSTM, and wordCNN) across six benchmark datasets and eight real-world commercial sentiment analysis platforms/models. Furthermore, we evaluate the Robustly optimized BERT pretraining Approach (RoBERTa) and models that enhance their robustness by adversarial training on toxicity detection and text classification tasks. The results demonstrate that our method minimizes the number of queries required for crafting plausible adversarial text while outperforming existing attack methods in the attack success rate, fluency of output sentences, and similarity between the original text and its adversary.

Event extraction (EE) is a crucial research task for promptly apprehending event information from massive textual data. With the rapid development of deep learning, EE based on deep learning technology has become a research hotspot. Numerous methods, datasets, and evaluation metrics have been proposed in the literature, raising the need for a comprehensive and updated survey. This article fills the research gap by reviewing the state-of-the-art approaches, especially focusing on the general domain EE based on deep learning models. We introduce a new literature classification of current general domain EE research according to the task definition. Afterward, we summarize the paradigm and models of EE approaches, and then discuss each of them in detail. As an important aspect, we summarize the benchmarks that support tests of predictions and evaluation metrics. A comprehensive comparison among different approaches is also provided in this survey. Finally, we conclude by summarizing future research directions facing the research area.

Real-world social events typically exhibit a severe class-imbalance distribution, which makes the trained detection model encounter a serious generalization challenge. Most studies solve this problem from the frequency perspective and emphasize the representation or classifier learning for tail classes. While in our observation, compared to the rarity of classes, the calibrated uncertainty estimated from well-trained evidential deep learning networks better reflects model performance. To this end, we propose a novel uncertainty-guided class imbalance learning framework—UCL$_{SED}$SED, and its variant—UCL-EC$_{SED}$SED, for imbalanced social event detection tasks. We aim to improve the overall model performance by enhancing model generalization to those uncertain classes. Considering performance degradation usually comes from misclassifying samples as their confusing neighboring classes, we focus on boundary learning in latent space and classifier learning with high-quality uncertainty estimation. First, we design a novel uncertainty-guided contrastive learning loss, namely UCL and its variant—UCL-EC, to manipulate distinguishable representation distribution for imbalanced data. During training, they force all classes, especially uncertain ones, to adaptively adjust a clear separable boundary in the feature space. Second, to obtain more robust and accurate class uncertainty, we combine the results of multi-view evidential classifiers via the Dempster-Shafer theory under the supervision of an additional calibration method. We conduct experiments on three severely imbalanced social event datasets including Events2012_100, Events2018_100, and CrisisLexT_7. Our model significantly improves social event representation and classification tasks in almost all classes, especially those uncertain ones.

In this article, an innovative design for the substrate integrated waveguide (SIW) equalizer operating in the Ka-band (millimeter waves) is proposed. This equalizer incorporates a capacitive loading technology and leverages the tiny gap effect to achieve a significantly enhanced equalization value. An equivalent capacitance can be created between the surface resistance layer of indium tin oxide (ITO) conductive film and the top metal layer of SIW body, due to tiny gap filled with a thin sheet of poly tetra fluoroethylene (PTFE). In this case, the area where surface current flows, characterized by surface resistance, has been enlarged. At lower operating frequencies, the expanded area is larger. Hence, the introductions of capacitive loading technology and tiny gap effect lead to a more prominent decrease in the attenuation constant $\vert \alpha \vert $ . This improvement is evident when compared to scenarios, where such technologies are not employed in previous studies. The measurement results show that the minimum/maximum equalization values are 3.95/14.77 dB ( $27\Omega $ /sq), 5.15/15.72 dB ( $37\Omega $ /sq), and 7.48/17.38 dB ( $50\Omega $ /sq), respectively. Significantly, the cost of SIW equalizer in this study is considerably lower than that of ceramic substrates. Moreover, the stability of the equalization value surpasses that of absorbing materials, which tend to exhibit slight variations under physical pressure.

Learning the representation of sentences is fundamental work in the field of Natural Language Processing. Although BERT-like transformers have achieved new SOTAs for sentence embedding in many tasks, they have been proven difficult to capture semantic similarity without proper fine-tuning. A common idea to measure Semantic Textual Similarity (STS) is considering the distance between two text embeddings defined by the dot product or cosine function. However, the semantic embedding spaces induced by pretrained transformers are generally non-smooth and tend to deviate from a normal distribution, which makes traditional distance metrics imprecise. In this paper, we first empirically explain the failure of cosine similarity in semantic textual similarity measuring, and present CoSENT, a novel Consistent SENTence embedding framework. Concretely, a supervised objective function is designed to optimize the Siamese BERT network by exploiting ranked similarity labels of sample pairs. The loss function utilizes uniform cosine similarity-based optimization for both the training and prediction phases, improving the consistency of the learned semantic space. Additionally, the unified objective function can be adaptively applied to different datasets with various types of annotations and different comparison schemes of the STS tasks only by using sortable labels. Empirical evaluations on 14 common textual similarity benchmarks demonstrate that the proposed CoSENT excels in performance and reduces training time cost.

ObjectiveThe management of blood pressure (BP) in acute ischaemic stroke remains a subject of controversy. This investigation aimed to explore the relationship between 24-hour BP patterns following ischaemic stroke and clinical outcomes.MethodsA cohort of 4069 patients who had an acute ischaemic stroke from 26 hospitals was examined. Five systolic BP trajectories were identified by using latent mixture modelling: trajectory category 5 (190–170 mm Hg), trajectory category 4 (180–140 mm Hg), trajectory category 3 (170–160 mm Hg), trajectory category 2 (155–145 mm Hg) and trajectory category 1 (150–130 mm Hg). The primary outcome was a composite outcome of death and major disability at 3 months poststroke.ResultsPatients with trajectory category 5 exhibited the highest risk, while those with trajectory category 1 had the lowest risk of adverse outcomes at 3-month follow-up. Compared with the patients in the trajectory category 5, adjusted ORs (95% CIs) for the primary outcome were 0.79 (0.58 to 1.10), 0.70 (0.53 to 0.93), 0.64 (0.47 to 0.86) and 0.47 (0.33 to 0.66) among patients in trajectory category 4, trajectory category 3, trajectory category 2 and trajectory category 1, respectively. Similar trends were observed for death, vascular events and the composite outcome of death and vascular events.ConclusionPatients with persistently high BP at 180 mm Hg within 24 hours of ischaemic stroke onset had the highest risk, while those maintaining stable BP at a moderate-low level (150 mm Hg) or even a low level (137 mm Hg) had more favourable outcomes.

Vehicle chassis coordinated control always has been an appealing topic in academia and industry because of the increasing number of chassis electronic actuators with the rapid development of automated vehicles. The optimization of multiple performance targets with multiactuators is intractable, which involves trajectory tracking and handling stability. Additionally, the optimization of tire friction usage remains a knotty problem. Therefore, this article develops a global chassis multiactuator integrated control framework, named by the chassis domain controller (CDC), to realize chassis global dynamics optimization for automated vehicles. Aiming at realizing more efficient, reliable, and flexible mobility, this framework defines each individual wheel to be fully adjustable and controllable to overcome the individual actuation limitation of traditional chassis structures. Global chassis dynamic modeling is formulated based on the analysis of distributed and controllable tire modules and vehicle dynamics motions. A game-theoretical control scheme is proposed to formulate chassis multiactuator integrated control, and the chassis global dynamics can be optimized by guaranteeing a Nash equilibrium for this game. Various experimental results demonstrate the feasibility and effectiveness of the proposed control method, and it suggests the CDC merits further studies to enhance the dynamics performance of automated vehicles in full situations.

Graph representation learning (GRL) is critical for graph-structured data analysis. However, most of the existing graph neural networks (GNNs) heavily rely on labeling information, which is normally expensive to obtain in the real world. Although some existing works aim to effectively learn graph representations in an unsupervised manner, they suffer from certain limitations, such as the heavy reliance on monotone contrastiveness and limited scalability. To overcome the aforementioned problems, in light of the recent advancements in graph contrastive learning, we introduce a novel self-supervised GRL algorithm via graph contrastive adjusted zooming, namely, G-Zoom, to learn node representations by leveraging the proposed adjusted zooming scheme. Specifically, this mechanism enables G-Zoom to explore and extract self-supervision signals from a graph from multiple scales: micro (i.e., node level), meso (i.e., neighborhood level), and macro (i.e., subgraph level). First, we generate two augmented views of the input graph via two different graph augmentations. Then, we establish three different contrastiveness on the above three scales progressively, from node, neighboring, to subgraph level, where we maximize the agreement between graph representations across scales. While we can extract valuable clues from a given graph on the micro and macro perspectives, the neighboring-level contrastiveness offers G-Zoom the capability of a customizable option based on our adjusted zooming scheme to manually choose an optimal viewpoint that lies between the micro and macro perspectives to better understand the graph data. In addition, to make our model scalable to large graphs, we use a parallel graph diffusion approach to decouple model training from the graph size. We have conducted extensive experiments on real-world datasets, and the results demonstrate that our proposed model outperforms the state-of-the-art methods consistently.

Seismic noise suppression refers to a data processing technique utilized to bolster the signal-to-noise ratio of recorded seismic signals. This enhancement in clarity can significantly amplify the efficacy of subsequent analysis and processing endeavors. With the advancement of deep learning, neural networks have significantly outperformed traditional denoising methods in seismic noise reduction tasks. They can process more data in less time and achieve better denoising results, moreover, they do not require limiting assumptions. In this article, we design an end-to-end seismic noise separation model based on the Wave-Unet, which is an adaptation of Unet for the 1-D time domain. Our designed model introduces the selective Kernel convolution into the Wave-Unet to enhance its performance and incorporates an attention-enhanced skip connection to bridge the semantic gap caused by the concatenation of features computed by networks of varying depths. Meanwhile, a joint loss function is used to enhance the model’s ability to learn signal frequency information. We tested our network using synthetic noisy signals and real seismic records from the STEAD dataset and compared the results with the wavelet threshold denoising method and the advanced omni-dimensional dynamic convolution (ODConv network) modules. The results demonstrate that our network performance surpasses the other two methods and can restore clean signals with high resolution without damaging the effective signal.

Existing wind turbine (WT) control strategies realize damping support mainly by active power modulation, which may affect wind energy harvest. To improve energy efficiency and robustness, this article proposes a two-stage coordinated control of type-4 WT for damping support using the cost-effective WT resources, where the grid-forming control is applied. In the first stage, two reactive power-based dampers are designed using DC voltage synchronization control. The impacts of the dampers on stability are analytically evaluated. In this stage, the reactive power modulation, along with the partial DC capacitor energy, are jointly utilized for damping provision, while the WT is operated at maximum power point tracking (MPPT) mode. The second stage that seeks active power modulation is only activated when the DC voltage exceeds the threshold. An MPPT recovery process is further designed to eliminate the droop errors induced by the damper. The proposed scheme is validated under various contingencies compared with existing controls. Numerical results show it does not require a frequency estimator or the active power modulation for WT synchronization and grid-forming purposes; meanwhile is relatively energy-efficient and robust to provide strong damping support.

This article presents an adaptive control strategy for lane-keeping task of automatic steering systems via sliding-mode technique. First, considering the road-vehicle lateral dynamics, a standard single track steering model has been developed for the lane-keeping task. To cope with the time-variant nature of the longitudinal velocity and uncertainty of measurement, a class of interval type-2 fuzzy sets considered in this article are employed to reconstruct the steering system dynamics mathematical model. Based on the fuzzy system, an integral sliding surface is proposed and the asymptotic convergence criterion for the overall system is derived with extended dissipation. Furthermore, an adaptive control law is provided to achieve the reachability of the assigned sliding surface and improve the attenuation ability to unknown curvature and exogenous disturbance. Finally, several scenarios with different path-following tasks are given in the simulations. Results demonstrate that the proposed sliding-mode control method has the capability to track the road centerline and is robust to external unknown disturbances.

Local electric field induced by the lightning-rod effect attracts great attention for regulating the local microenvironment and electronic properties of active sites. Nevertheless, local electric-field-assisted applications are mainly limited to metals with strong surface plasmonic resonance properties (e.g., Au, Ag, and Cu). Herein, we fabricate RuCu snow-like nanosheets (SNSs) with high-curvature nanotips for enhancing the hydrogen oxidation reaction (HOR) and hydrogen evolution reaction (HER). Theoretical simulations show that RuCu SNSs can induce a strong local electric field around the sharp nanotips, which favors the accumulation of OH–for HOR and H+for HER. Cu incorporation can modulate the binding strength of OH* and H*, leading to significantly enhanced HOR and HER performance. Impressively, the mass activity of RuCu SNSs for alkaline HOR is 31.3 times higher than that of RuCu nanocrystals without sharp tips. Besides, the required overpotential for reaching 10 mA cm–2during HER over RuCu SNSs is 14.0 mV.

View synthesis methods using implicit continuous shape representations learned from a set of images, such as the Neural Radiance Field (NeRF) method, have gained increasing attention due to their high quality imagery and scalability to high resolution. However, the heavy computation required by its volumetric approach prevents NeRF from being useful in practice; minutes are taken to render a single image of a few megapixels. Now, an image of a scene can be rendered in a level-of-detail manner, so we posit that a complicated region of the scene should be represented by a large neural network while a small neural network is capable of encoding a simple region, enabling a balance between efficiency and quality. Recursive-NeRF is our embodiment of this idea, providing an efficient and adaptive rendering and training approach for NeRF. The core of Recursive-NeRF learns uncertainties for query coordinates, representing the quality of the predicted color and volumetric intensity at each level. Only query coordinates with high uncertainties are forwarded to the next level to a bigger neural network with a more powerful representational capability. The final rendered image is a composition of results from neural networks of all levels. Our evaluation on public datasets and a large-scale scene dataset we collected shows that Recursive-NeRF is more efficient than NeRF while providing state-of-the-art quality. The code will be available at https://github.com/Gword/Recursive-NeRF

Feature extraction of the retinal blood vessel is one of the crucial tasks in the prediction of ophthalmologic diseases. Important features are extracted based on image segmentation results. The efficiency of vessel segmentation methods could help doctors in the diagnostic of several relevant diseases as early as possible. Recently, U-Net has achieved good results in many medical image segmentation tasks, especially for images of blood vessels. However, due to the limitation of the network structure, some small features could be lost in the transmission process. As a result, there are still many research gaps for U-Net-based retinal vessel segmentation works. In this paper, we propose an improved U-Net based model to segment images of retinal vessels. The improvement focuses on U-Net from two aspects: designing a local feature enhancement module composed of dilated convolution and 1×1convolution to enhance the feature extraction of tiny vessels; integrating an attention mechanism with skip connection of the network to highlight features related to vessel segmentation information taken from the down-sampling part to the up-sampling part. The performance of the proposed model was evaluated and compared with several published state-of-the-art approaches on the same public dataset—DRIVE, and the proposed method achieved an accuracy of 0.9563, F1-score of 0.823, TPR of 0.7983, and TNR of 0.9793. The AUC of PRC is 0.9109 and the AUC of ROC is 0.9794. The results proved the potential for clinical applications.

This paper investigates 3D object detection from point clouds scene recovered by lidar and proposes a novel semi-anchor-free 3D object detector. Compared with the point-based method, the voxel-based method has more advantages in feature extraction and avoids the complicated neighborhood search. However, voxel-based methods usually require to preset a large number of anchors related to the objects to be detected, which brings the computational cost to the model inference. This paper introduces a framework with the semi-anchor-free mechanism that utilizes the voxel-based method to aid the learning of point features and achieves anchor-free performance in inference. Actually, feature learning still needs to be strengthened for point cloud encoding, and this paper designs a Directional Slice Attention to augment the discriminability of features. Additionally, the region of interest characterization based on Symmetric Feature Propagation is proposed in this paper to alleviate the obstacle brought by the incomplete scanning of objects in autonomous driving scenarios. Extensive experiments on both the KITTI dataset and Waymo Open dataset demonstrate the superiority of 3SNet over state-of-the-art methods.

Hydrogel‐based materials are widely used in many fields, but most of them compared to tendons with the same water content do not exhibit high toughness, strength or fatigue resistance, which limits their application. The Hofmeister effect can adjust the mechanical properties of hydrogels by ions affecting the aggregation of polymer chains. Herein, we prepared a bioactivated hydrogel through simply mixing polyvinyl alcohol (PVA) and Na2CO3aqueous solutions. PVA chains are aggregated by strong hydrogen bonds between CO32−and OH, at which stage the hydrogel is soft and easy to shape. Further salting out promotes the formation of crystal microphase of PVA chain, resulting in high mechanical strength and excellent toughness. The existence of Na2CO3endow the hydrogel with conductivity, exhibits high and stable sensitivity in strain sensing, and could be used to monitor human actions. On the basis of this method, Ca2+was added to prepare mineralized PVA hydrogels, which had high stability and long‐term swelling resistance in different ionic solutions. In vitro cell experiment, hydrogel had obvious effect of promoting bone cell proliferation and calcium deposition. This work provides a facile and feasible method for preparing bionic hydrogels with high strength, toughness, and adaptable shape.

Graph embedding is essential for graph mining tasks. With the prevalence of graph data in real-world applications, many methods have been proposed in recent years to learn high-quality graph embedding for various types of graphs, among which the Generative Adversarial Networks (GAN) based methods attract increasing attention among researchers. However, most GAN-based generator-discriminator frameworks randomly generate the negative samples from the original graph distributions to enhance the training process of the discriminator without considering the noise. In addition, most of these methods only focus on the explicit graph structures and cannot fully capture complex semantics of edges such as various relationships or asymmetry. In order to address these issues, we propose a robust and generalized framework named AGE. It generates fake neighbors as the enhanced negative samples from the implicit distribution, and enables the discriminator and generator to jointly learn robust and generalized node representations. Based on this framework, we propose three models to handle three types of graph data and derive the corresponding optimization algorithms, namely the UG-AGE and DG-AGE for undirected and directed homogeneous graphs, respectively, and the HIN-AGE for heterogeneous information networks. Extensive experiments show that our methods consistently and significantly outperform existing state-of-the-art methods across multiple graph mining tasks.

Multimodal treatment is an important tool to overcome tumor drug resistance. The reactive oxygen species (ROS) generated by photodynamic therapy (PDT) can directly play a killing role on tumor cells, which has the advantages of repeatable treatment and no drug resistance. However, its therapeutic oxygen consumption and destruction of tumor microvessels lead to hypoxia in tumor tissues, and hypoxia leads to overexpression of the receptor tyrosine kinase (c-MET) and vascular endothelial growth factor receptor (VEGFR). Overexpression of these two receptors leads to increased tumor invasiveness and metastasis. The molecularly targeted drug cabozantinib (CAB) has multiple targets, including anti-c-MET and VEGFR, to inhibit the development of hepatocellular carcinoma (HCC). In this study, our team designed a pH-sensitive nanoparticle CAB/Ce6@ZIF-8@PEG-FA (CCZP) loaded with CAB and Ce6, which exerted a multimodal therapeutic effect of PDT and molecularly targeted therapy by laser irradiation, and the PDT-induced overexpression of MET and VEGFR could also be inhibited by the target of CAB, thus reducing the invasive tumor cells metastasis. In summary, CCZP gives full play to the advantages of both drugs, exerting multimodal treatment while reducing HCC invasion and metastasis, providing a safe, potential approach to clinical treatment.