Search

In this research, unique Tourmaline@ZnO micro-nano structure were constructed by surface-coating using uniform precipitation, with tourmaline particles as the inner core and nano-zinc oxide as the outer layer. In order to explore the optimized preparation conditions towards the highly efficient degradation of organic pollutant, the preparative experiments were implemented based on the two most crucial processing parameters: growth temperature and dosage of tourmaline. The structure, morphology and surface analysis of as-prepared core–shell composites were systemically characterized by x-ray diffraction, scanning electron microscopy, transmission electron microscopy, Brunauer–Emmett–Teller and x-ray photoelectron spectroscopy measurements, then the optical and optoelectrical properties were tested by ultraviolet–visible spectroscopy, photoluminescence spectroscopy and electrochemical performance. The photocatalysis performances were evaluated by selecting methyl orange as the model dye under the irradiation of a xenon lamp. In the case of a growth temperature of 100°C and tourmaline introduction amount of 3 wt.%, the correlated investigations confirmed that the hybrid tourmaline@nano-ZnO product achieved the most efficient degradation efficiency of 90% and the fastest reaction velocity of 0.8889 h−1 respectively within 2 h, with an enhancement of 9% and 36% compared to pure ZnO (the degradation rate was 81% and reaction rate was 0.6521 h−1). The tourmaline core was confirmed to make important contributions to the growth morphology and degradation performance of ZnO-based photocatalytic materials. The introduction of tourmaline will play a key role in the properties modification of metal oxide.

As a crucial task for video analysis, social relation recognition for characters not only provides semantically rich description of video content but also supports intelligent applications, e.g., video retrieval and visual question answering. Unfortunately, due to the semantic gap between visual and semantic features, traditional solutions may fail to reveal the accurate relations among characters. At the same time, the development of social media platforms has now promoted the emergence of crowdsourced comments, which may enhance the recognition task with semantic and descriptive cues. To that end, in this article, we propose a novel multimodal-based solution to deal with the character relation recognition task. Specifically, we capture the target character pairs via a search module and then design a multistream architecture for jointly embedding the visual and textual information, in which feature fusion and attention mechanism are adapted for better integrating the multimodal inputs. Finally, supervised learning is applied to classify character relations. Experiments on real-world data sets validate that our solution outperforms several competitive baselines.

This paper presents model tests (macro aspect) and microstructure tests (micro aspect) for investigating the consolidation behavior of Tianjin dredged clay using the prefabricated vertical drain air-booster vacuum preloading (PAVP) and tube air-booster vacuum preloading (TAVP) methods. The mechanism of air-booster vacuum preloading (AVP) using a spring-like system is explained. The main difference between these two methods is the air-boosting equipment. A new anticlogging air-booster prefabricated vertical drain (PVD) is used in the PAVP technique and a self-designed air-booster tube is used in the TAVP technique. In the model tests, a comparison of the variables that are monitored during reinforcement (vacuum pressure, surface settlement, water discharge, and pore-water pressure) and after reinforcement (water content, dry density, and vane shear strength) is conducted. The results indicate that the consolidation behavior of Tianjin dredged clay using the PAVP method is better than that using the TAVP method. PAVP more efficiently mitigates the issue of water-draining PVD clogging and significantly accelerates drainage consolidation. In addition, in the microstructure tests, a comparison of the variables that are monitored after reinforcement (via scanning electron microscopy (SEM) and mercury intrusion porosimetry (MIP)) is conducted, and the results further explain the model test results.

Micro steel pipe pile was used for existing foundation reinforcement and renovation. An energy micro pile-raft foundation equipped with heat exchange tube was constructed in silty clay. The diameter and the length of the energy micro pile are 160 mm and 13.0 m, respectively. A series of in situ thermal performance tests were carried out by controlling cycle heating, in which the inlet and outlet water temperatures, flow rate, and thermomechanical properties of the energy micro pile were measured. Combined with a numerical simulation method, the thermomechanical stresses and displacement of the raft were also analyzed and discussed. The energy micro pile-raft foundation was also analyzed for different combinations of energy piles and nonenergy piles in the group. Results show that the micro pile-raft foundation can provide sufficient heat exchange compared with other types of ground heat exchangers. Differential settlement at both the pile top and tip were observed for the groups that contained both energy piles and nonenergy piles.

Multifunctional mesoporous silica-based nanocarriers able to efficiently encapsulate drugs for stimuli-responsive release and display rapid biodegradation are highly desirable. In this work, we dope disulfide bonds and calcium into silica framework by one step method to obtain a redox and pH dual-responsive biodegradable mesoporous silica nanoparticle (BT-Ca-MSN) as a potential drug carrier for synergistic cancer therapy. TEM and ICP-OES are used to assess the biodegradation behavior of BT-Ca-MSN. The results show that BT-Ca-MSN can significantly biodegrade in a concurrent reductive and acidic environment due to the simultaneous disulfide bonds cleavage and Ca2+ release. In addition, BT-Ca-MSN shows efficient drug loading capacity and significant biodegradation-mediated drug release. Moreover, the in-vitro cytotoxicity indicates that BT-Ca-MSN can not only exhibit significant cancer cell killing effect without obvious toxicity on healthy cells via the way of released Ca2+-mediated apoptosis, but also can combine with its loaded doxorubicin hydrochloride for synergistic cancer therapy. This work demonstrates that BT-Ca-MSN is a promising platform as drug carrier, providing a paradigm to rationally design biodegradable silica-based carriers for highly efficient cancer therapy.

Quaternary compound BiAgSeS shows an excellent thermoelectric performance with ultralow thermal conductivity. However, competition between the trigonal and cubic phases, which is strongly dependent on experimental processing, results in ambiguous transport behavior. The highly stable n-type polycrystalline BiAgSe2−xSx (x = 1.2) sample was successfully synthesized, guided by phase boundary engineering. Whilst maintaining promising thermoelectric properties, significant enhancement of the thermodynamical stability was obtained due to the strengthened chemical bonding caused by the substitution of Se with high-electronegativity S. Cubic BiAgSe0.8S1.2 can be well preserved after thermoelectric measurements. Our work demonstrates that phase boundary engineering is an applicable route to stabilize some compounds with moderate thermodynamical stability.

The microenvironmental humidity at the interface between the bioelectrical electrode and the human skin can affect the quality of the bioelectrical signal collected by the electrode. This paper utilized superabsorbent polyacrylic fiber as the base material to fabricate a novel moisture slow-releasing embroidered electrode. The moisture-locking capacity of polyester non-woven fabric (PENWF) substrate, polyacrylic non-woven fabric (PANWF) substrate, polyester-based electrode (PE-E) and polyacrylic-based electrode (PA-E) were compared in this paper. The non-woven fabric with polyacrylic fiber owned excellent moisture-locking capacity, so it could realize the slow-releasing of moisture and provide a suitable wet environment for the dry bioelectric electrode to collect ECG signals. With the extension of time, the amplitude of the ECG signal collected by PA-E did not change much, and only decreased by 17.1 % after evaporating for 10 h; but after evaporating for 10 h, the amplitude of the ECG signal collected by the PE-E decreased by 48.9 %. In addition, 5 mm, 10 mm, and 15 mm thick polyurethane sponge (PUS) filling materials were used to make the ECG monitoring belts, and their signal-to-noise ratios were analyzed under different states of static, swinging arms and walking. The 5 mm thick elastic filler material has the highest signal-to-noise ratio among the three thicknesses. It could collect ECG signals stably under swing arm and walking at a constant speed, and the signal-to-noise ratio (SNR) were 25.393 dB and 30.086 dB respectively. The polyurethane sponge filling materials with a thickness of 5 mm provided an appropriate pre-stress for the ECG signal dynamic measurement, which could provide a reference parameter for the production of smart ECG garment in the future.

We proposed and developed a small bionic amphibious spherical robot system for tasks such as coastal environment monitoring and offshore autonomous search and rescue. Our third-generation bionic small amphibious spherical robots have many disadvantages, such as the lack of maneuverability and a small operating range. It is difficult to accomplish underwater autonomous motion control with these robots. Therefore, we proposed a fourth-generation amphibious spherical robot. However, the amphibious spherical robot developed in this project has a small and compact design, with limited sensors and external sensing options. This means that the robot has weak external information collection capabilities. We need to make the real time operation of the robot’s underwater motion control system more reliable. In this paper, we mainly used a fuzzy Proportional-Integral-Derivative (PID) control algorithm to design an underwater motion control system for a novel robot. Moreover, we compared PID with fuzzy PID control methods by carrying out experiments on heading and turning bow motions to verify that the fuzzy PID is more robust and exhibits good dynamic performance. We also carried out experiments on the three-dimensional (3D) motion control to validate the design of the underwater motion control system.

Air pollution by dust particles is a major health concern in urban and industrialized areas. Actual methods to reduce dust contamination, such as water spraying, are limited, calling for new dust suppressants. Here, we synthesized a liquid starch-based dust suppressant by grafting polyacrylic acid onto NaOH-pretreated potato starch and then crosslinking the grafted copolymer with gelatine. This dust suppressant was characterized by Fourier-transform infrared spectrometry, thermogravimetry, viscosity and contact angle measurements. We also measured permeability, water retention, anti-wind erosion and degradability of the dust suppressant. Field experiments tested the dust suppressing effect on particle matter 2.5 (PM2.5) and particulate particle matter 10.0 (PM10). We found that the addition of NaOH to potato starch improved the reaction efficiency and simplified the preparation process. The dust suppressant demonstrated excellent degradability, as shown by weight loss of 82 wt % after 60 days of soil burial, and the total dust suppression rates reached 57% for PM2.5 and 56% for PM10 in the field. Compared with spraying water, total dust suppression was higher by 25% and 33%, respectively.

A 4Kg grade AlCoCrFeNi2.2 high entropy alloy was prepared by low vacuum medium frequency induction melting. The microstructure results showed that the microstructure and composition of the ingot were uniform. The microstructures were composed of primary Ni-rich FCC phase and eutectic structure. The eutectic structure was composed of Ni-rich FCC phase and Al-rich B2 phase. The hardness results showed no significant difference in Vickers hardness between different parts of the ingot, and fluctuated within the range of HV260 ± 20. The compression test results showed that there was no significant difference in compression performance between different parts of the ingot. The AlCoCrFeNi2.2 high entropy alloy had excellent compression performance, the yield strength reached 600 MPa, the fracture strength reached 2200 MPa, and the compression ratio was greater than 40%. Above all, the near-eutectic AlCoCrFeNi2.2 high-entropy alloy had the characteristics of good casting ability, high strength, good plasticity, and was easy to realize the preparation of large-volume non-segregation ingot.

Graph-based semi-supervised learning (GSSL) studies the problem where in addition to a set of data points with few available labels, there also exists a graph structure that describes the underlying relationship between data items. In practice, structure uncertainty often occurs in graphs when edges exist between data with different labels, which may further results in prediction uncertainty of labels. Considering that Gaussian process generalizes well with few labels and can naturally model uncertainty, in this paper, we propose an Uncertainty aware Graph Gaussian Process based approach (UaGGP) for GSSL. UaGGP exploits the prediction uncertainty and label smooth regularization to guide each other during learning. To further subdue the effect of irrelevant neighbors, UaGGP also aggregates the clean representation in the original space and the learned representation. Experiments on benchmarks demonstrate the effectiveness of the proposed approach.

In multi-view multi-label learning (MVML), each training example is represented by different feature vectors and associated with multiple labels simultaneously. Nonetheless, the labeling quality of training examples is tend to be affected by annotation noises. In this paper, the problem of multi-view partial multi-label learning (MVPML) is studied, where the set of associated labels are assumed to be candidate ones and only partially valid. To solve the MVPML problem, a two-stage graph-based disambiguation approach is proposed. Firstly, the ground-truth labels of each training example are estimated by disambiguating the candidate labels with fused similarity graph. After that, the predictive model for each label is learned from embedding features generated from disambiguation-guided clustering analysis. Extensive experimental studies clearly validate the effectiveness of the proposed approach in solving the MVPML problem.

The building-damper system designed by a seismic code is usually considered to be able to withstand the attack of strong earthquakes. However, near-fault earthquakes, especially those with the forward-directivity effect, might cause early and unexpected failure of code-designed dampers and consequent severe structural damage. In this paper, by taking into account near-fault earthquakes, seismic performance of the building-damper system and damper failure’s influence are evaluated systematically. A 9-storey steel building is designed by the Chinese seismic code as the benchmark model, and five typical dampers, including buckling-restrained brace damper (BRB), friction damper (FD), self-centering damper (SCD), viscous damper (VD), and viscoelastic damper (VED), are adopted. It was found that the building-damper systems show a large response and possible damper failure under the near-fault earthquake excitations. Then, the influence of damper failure is investigated, which reveals that damper failure would significantly affect seismic performance of the building-damper system, especially for the building-SCD system. Subsequently, by introducing the artificial near-fault earthquake excitation, the influences of different pulse parameters, such as pulse velocity amplitude, pulse period, and the number of significant pulses, are studied. It shows that the pulse velocity amplitude and pulse period obviously affect the seismic performance, while the number of significant pulses presents little influence.

Seismic loads pose a potential threat to the high-speed railway bridges in China, which have been rapidly developing in recent years, especially for those subjected to the near-fault earthquakes. The previous researches on high-speed railway bridges usually concern the far-field earthquake, and the damage of high-speed railway bridge–track system subjected to the near-fault earthquake has not been well studied. In this article, a seven-span high-speed railway simply supported bridge–track system is selected to explore the seismic damage features under the excitation of near-fault earthquake which possesses characteristics of obvious velocity pulse and high-frequency vibration. First, a detailed finite element model of the selected bridge–track system is established and calibrated by the experimental data and design code. Then the low-frequency pulse-type portion and the high-frequency background portion are separated from the selected eight original near-fault records, and a series of nonlinear dynamic analysis is conducted. The results show that the background portion leads to more serious damage of the bridge–track system than the pulse-type portion. Due to the high stiffness of high-speed railway bridge–track system, the background portion with high-frequency vibration characteristic produces the main part of seismic response of system. As for the damage part of system, the weakest component of the bridge–track system is the sliding layer, followed by the shear alveolar.

Integrated zoom and image stabilization system based on deformable mirrors (DMs) has advantages of miniaturization, rapid response and low energy consumption. Integrating the two capabilities on one DM-based imaging system poses considerable challenges. First, limited DM stroke will result in limited changeable aberration correction value for zoom and image stabilization. Second, the DM-based off-axis imaging system suffers a simple relative movement between optical axis and objects caused by carrier vibration may result in complicated optical system aberrations. To address these challenges, a reasonable assignment of the changeable aberration correction value for zoom and image stabilization is needed. Image stabilization and aberration correction in DM-based off-axis imaging system, should be considered when translation or rotation occurs. To troubleshoot the assignment issues of the correcting aberrations, we propose a graded optimization method which comprises two steps. To troubleshoot the image stabilization and aberration correction issues, we introduce vectorial ray tracing method and image point freezing principle to transform the aberration correction problem into an optimization problem of image coordinates. With the proposed method, we construct practical integrated freeform surfaces of DMs for a space camera with three mirrors. The sagittal heights of designed DM surface profiles are limited within their available stroke.

Infection is a main cause of implant failure. Early implant-related infections often occur in the first 4 weeks post-operation. Inhibiting bacterial adhesion and biofilm formation at the early stage and promoting subsequent implant osseointegration are important for implant success. Our previous studies demonstrated that dimethylaminododecyl methacrylate (DMADDM) provided dental materials with antibacterial effects. In the present study, DMADDM and hydroxyapatite (HA) are loaded on to the titanium (Ti) surface via poly dopamine (PDA) self-polymerization. This local DMADDM-delivery Ti is referred as Ti-PHD. Here we report the two-staged capability of Ti-PHD: (1) in the first stage, releasing DMADDM during the high-infection-risk initial period post-implantation for 4 weeks; (2) then in the second stage, enhancing osteogenesis and promoting osseointegration. Ti-PHD has a porous surface with higher average roughness and greater hydrophilicity than pure Ti. Its biocompatibility is verified in vitro and in vivo. During the first 4 weeks of release, both DMADDM remaining on Ti surface and DMADDM released into the soaking medium greatly reduced the adherence and growth of pathogens. This is further confirmed by the prevention of bone destruction in a rat osteomyelitis model. After releasing DMADDM for 4 weeks, Ti-PHD promotes osteogenic differentiation of human bone marrow mesenchymal stem cells (hBMSCs) and new bone formation around the implants in vivo. This article represents the first report on the two-staged, time-dependent antibacterial and osteogenesis effects of Ti-PHD, demonstrating its potential for clinical applications to inhibit implant-associated infections. STATEMENT OF SIGNIFICANCE: The present study develops a two-staged time-dependent system for local dimethylaminododecyl methacrylate (DMADDM) delivery via Ti implant (referred to as Ti-PHD). DMADDM and hydroxyapatite (HA) are loaded on to the Ti surface with poly dopamine (PDA). Ti-PHD can release DMADDM during the high-risk period of infection in the first stage, and then promote osseointegration and new bone formation in the second stage. This bioactive and therapeutic Ti is promising to inhibit infections and enhance implant success.

Oceanographic plankton classification for many images is time consuming, especially for low-contrast images obtained when the water is dirty and opaque due to impurities. A novel, overcomplete dictionary algorithm is studied by analyzing the sparse characteristics of the image matrix using pixel values. The features in hyperspace are mapped onto a specifically designed vector space. Thus, mathematically, the algorithm exhibits faster calculating convergence and has a strong expressivity for the selective signal in the vector space with a lower signal loss rate. The clustering method based on the dictionary can classify planktons for species counting, which can enable high-speed, multi-object recognition of planktons in turbid water. The experimental results demonstrate that if less data (up to 60%) is processed for each image, a recall rate and accuracy greater than 75% and a structural similarity for the reconstructed image greater than 0.9 can be achieved.

Poly(3,4-ethenedioxythiophene):poly (styrenesulfonate) (PEDOT:PSS) and graphene oxide (GO) as well as their hybrids have been widely used to prepare hole-transport layer (HTL) for inverted planar perovskite solar cells (PeSCs). However, their acidic nature is detrimental to the stability of PeSC. In this study, we report a simple and general strategy to prepare non-corrosive HTL by treating GO and PEDOT:PSS with ammonia or ammonium (i.e., (NH4)2HPO4), respectively. The PeSC based on ammonia-treated GO (a-GO) yields a power conversion efficiency (PCE) over 14%, remarkably outperforming those with pristine GO (∼12%) or PEDOT:PSS (∼13%). Similar improvement is also observed for the devices with ammonia or ammonium-treated PEDOT:PSS. The mechanism behind such improvement is thoroughly studied, demonstrating the significance of ammonia treatment on the HTL performance. More importantly, the devices based on ammonia-treated HTL exhibit significantly improved stability, as compared to those based on pristine PEDOT:PSS and GO.

Newly dredged fills feature a high moisture content and a high floating mud content. Based on the conventional vacuum preloading (CVP) method, this article presents a new method of air-booster vacu...

High-power high-efficiency broad-area diode lasers are highly efficient light sources. A study on the design of large-optical-cavity 976-nm diode lasers is presented here. Efforts were made to suppress the higher-order modes and to explore new design methods for high-power high-efficiency diode lasers. The mode distribution, confinement factor, internal optical loss and far-field pattern were numerically calculated to analyze the waveguide design. To suppress the higher-order transverse modes, both gain and loss discrimination mechanisms were experimentally investigated. By adjusting the position of the active region, successful suppression of higher-order modes can be achieved. Alternatively, higher-order modes can be suppressed by increasing the cavity length, which is attributed to the discrimination of modal loss between the higher-order modes and the fundamental mode. A high-efficiency 4-mm-long broad-area single emitter with an ultralow internal optical loss of 0.27 cm–1 and an internal quantum efficiency of 93.5% was designed and fabricated. A high slope efficiency of 1.1 W/A and >66% wall-plug efficiency was obtained at 25 °C under continuous current injection. The far-field divergence angles with 95% power content were 38° and 9.7° for the fast and slow axes, respectively, at an output power of 10.0 W.

In the narrow, submarine, unstructured environment, the present localization approaches, such as GPS measurement, dead-reckoning, acoustic positioning, artificial landmarks-based method, are hard to be used for multiple small-scale underwater robots. Therefore, this paper proposes a novel RGB-D camera and Inertial Measurement Unit (IMU) fusion-based cooperative and relative close-range localization approach for special environments, such as underwater caves. Owing to the rotation movement with zero-radius, the cooperative localization of Multiple Turtle-inspired Amphibious Spherical Robot (MTASRs) is realized. Firstly, we present an efficient Histogram of Oriented Gradient (HOG) and Color Names (CNs) fusion feature extracted from color images of TASRs. Then, by training Support Vector Machine (SVM) classifier with this fusion feature, an automatic recognition method of TASRs is developed. Secondly, RGB-D camera-based measurement model is obtained by the depth map. In order to realize the cooperative and relative close-range localization of MTASRs, the MTASRs model is established with RGB-D camera and IMU. Finally, the depth measurement in water is corrected and the efficiency of RGB-D camera for underwater application is validated. Then experiments of our proposed localization method with three robots were conducted and the results verified the feasibility of the proposed method for MTASRs.

Adding dampers is a commonly adopted seismic risk mitigation strategy for modern buildings, and the corresponding design procedure of dampers has been well established by the Chinese Building Code. Even though all types of dampers are designed by the same procedure, actual seismic performance of the building may differ from one to the others. In this study, a nine-story benchmark steel building is established, and three different and typical types of dampers are designed according to the Chinese Building Code to realize structural vibration control under strong earthquake excitation. The seismic response of the prototype building equipped with a viscoelastic damper, viscous damper and buckling-restrained brace (BRB) subjected to 10 earthquake records are calculated, and Incremental Dynamic Analysis (IDA) is performed to describe progressive damage of the structure under increasing earthquake intensity. In the perspective of fragility, it shows that the viscoelastic damper has the highest collapse margin ratio (CMR), and the viscous damper provides the best drift control. Both the BRB and viscoelastic dampers can effectively reduce the floor acceleration responses in the mid-rise building.

The versatility of debranched starch nanoparticles (DBS-NPs) has attracted considerable attention from the food, agricultural, and cosmetic industries. The aim of this study was to investigate the effect of DBSs with low (DP DP DP < 25) DPs (where DP represents the degree of polymerization), as well as reaction temperatures (25, 60, and 90 °C) on size, morphology, crystal structure, and digestibility of the refined DBS-NPs via nanoprecipitation. Spherical DBS-NPs with a controlled size of 40–200 nm were obtained; no nanoparticles were observed when the average DP was

It is difficult to effectively rectify or convert low power in the circuit stage of harvesters. The high-gain antenna can offer a higher level of power and is beneficial to RF power harvesting in a low power density environment. To extend the narrow beam of the conventional high-gain rectenna, the multiport and multibeam antenna is promising. To pursue a simple configuration and avoid extra beamforming networks, a traveling-wave grid-array antenna (GAA) with two isolated ports and two symmetrically tilted beams is proposed. A prototype is designed and fabricated after a detailed analysis of the GAA with tilted beams and the coplanar stripline-based rectifier. The measured results show that the proposed rectenna is sensitive and effective in a wide-angle range. The harvesting angle range is extended by combining two tilted beams and can be greater than 70°, where the level of dc power exceeds $100~\mu \text{W}$ when the power density is as low as $1~\mu \text{W}$ /cm2. A maximum dc output of 3.6– $203.8~\mu \text{W}$ and a maximum RF-to-dc conversion efficiency of 16.3%–45.3% are available at 2.45 GHz, under the condition that the power density ranges from 0.052 to $1~\mu \text{W}$ /cm2.

The refractive index is one of the most important parameters of optical glasses and has a significant effect on optical properties. The measurement of optical glasses, especially for optical elements such as lenses, is urgently needed. However, several presented methods require the immersion of the sample in liquid and provide indirect measurements, while others require structural parameters as priori knowledge, which is complex and time-consuming. In this study, a Brewster-Law-based direct and simple measurement method for the refractive index of glasses with arbitrary shapes is proposed, and a laser beam is focused on the surface of the sample as a probe. The incident angle of the chief ray is close to the Brewster angle. The reflected light is collected by an array detector. The refractive index is calculated from the minimum intensity position obtained with image processing. Additionally, a symmetric measurement scheme is proposed to improve the accuracy. Using these methods, a prism and four spherical lens samples with different refractive indices or radii of curvature are tested and error analyses are carried out. Results indicate that the accuracy can reach 10−4.

Since the label collecting is prohibitive and time-consuming, unsupervised methods are preferred in applications such as fraud detection. Meanwhile, such applications usually require modeling the intrinsic clusters in high-dimensional data, which usually displays heterogeneous statistical patterns as the patterns of different clusters may appear in different dimensions. Existing methods propose to model the data clusters on selected dimensions, yet globally omitting any dimension may damage the pattern of certain clusters. To address the above issues, we propose a novel unsupervised generative framework called FIRD, which utilizes adversarial distributions to fit and disentangle the heterogeneous statistical patterns. When applying to discrete spaces, FIRD effectively distinguishes the synchronized fraudsters from normal users. Besides, FIRD also provides superior performance on anomaly detection datasets compared with SOTA anomaly detection methods (over 5% average AUC improvement). The significant experiment results on various datasets verify that the proposed method can better model the heterogeneous statistical patterns in high-dimensional data and benefit downstream applications.

This paper presents a novel multi-binocular camera based localization method with panoramic view and high precision for amphibious spherical robots. In our previous work, the proposed approach based on RGB-D camera and inertial measurement unit (IMU) realized a cooperative and relatively close-range localization. While, position drifts caused by the rotation movement and short positioning range of RGB-D camera in underwater environments are the drawbacks of the previous approach. This paper designs a multi-binocular camera system on the amphibious spherical robot, which makes the robot observe targets in a nearly 360-degree visual range without rotation movement. Meanwhile, through the stereo vision model, the robot obtains the position information about the target robot. Then, according to the position information and the relative position of the binocular cameras, the cooperative localization model is established. To validate the proposed approach, experiments were performed in gazebo simulator with varying distances and relative degree. Compared with the previous works, the proposed localization method reveals higher accuracy and robustness.

The underwater environment is complex and diverse, and it is difficult to explore the ocean, especially in the deep sea area. The lack of light, high pressure, and lack of oxygen make it difficult for humans to work in such extreme environments. Therefore, people are beginning to hand over the burden of exploring the ocean to underwater robots. The underwater robot's perception of the underwater environment is very important for controlling the motion of the underwater robot. However, in the environment where the light is lacking at night and in the deep sea, and in the narrow space, the robot vision consisting of infrared and camera is traditionally used. The system is difficult to function, so a new sensing system is needed to assist the underwater robot in recognizing the surrounding environment. In nature, real fish use their lateral line system to sense the surrounding environment. Based on the principle of bionics, we have developed an artificial lateral line system on a bionic spherical robot. The basic principle of this artificial lateral line system is to use the pressure sensor to sense the water flow pressure around the robot when the robot swims under water, thus establishing a relationship model between the robot swimming speed and the water flow pressure to realize the relative speed of identifying the underwater robot and the water flow.

In this paper, we propose a scene-aware context reasoning method that exploits context information from visual features for unsupervised abnormal event detection in videos, which bridges the semantic gap between visual context and the meaning of abnormal events. In particular, we build na spatio-temporal context graph to model visual context information including appearances of objects, spatio-temporal relationships among objects and scene types. The context information is encoded into the nodes and edges of the graph, and their states are iteratively updated by using multiple RNNs with message passing for context reasoning. To infer the spatio-temporal context graph in various scenes, we develop a graph-based deep Gaussian mixture model for scene clustering in an unsupervised manner. We then compute frame-level anomaly scores based on the context information to discriminate abnormal events in various scenes. Evaluations on three challenging datasets, including the UCF-Crime, Avenue, and ShanghaiTech datasets, demonstrate the effectiveness of our method.

Visual Grounding is the task of associating entities in a natural language sentence with objects in an image. In this paper, we formulate visual grounding as a graph matching problem to find node correspondences between a visual scene graph and a language scene graph. These two graphs are heterogeneous, representing structure layouts of the sentence and image, respectively. We learn unified contextual node representations of the two graphs by using a cross-modal graph convolutional network to reduce their discrepancy. The graph matching is thus relaxed as a linear assignment problem because the learned node representations characterize both node information and structure information. A permutation loss and a semantic cycle-consistency loss are further introduced to solve the linear assignment problem with or without ground-truth correspondences. Experimental results on two visual grounding tasks, i.e., referring expression comprehension and phrase localization, demonstrate the effectiveness of our method.

Convolutional Neural Networks (CNNs) have excellent ability in image recognition, however, the requirement of a large amount of labeled dataset limits its application in the field of synthetic aperture radar (SAR) image processing. In this paper, a kernel rotational network (KR-Net) for SAR target recognition is constructed. When the labeled dataset is small, the KR-net can achieve higher classification rate than standard CNNs benefit from its inherent rotational convolution units. Also, weights sharing strategy is introduced to increase network capacity without multiplying the number of weights parameters. Meanwhile, a simple and feasible multi-branch feature converging method for the KR-Net is proposed to fuse features of rotational convolution units. Experimental results show that our network can achieve state-of-art result in the MSTAR dataset, especially when the training set is small.

Owing to the diminishing resources and growing awareness of environmental issues, significant scientific attention has been paid to the development of physical gel materials using renewable and low-cost natural resources. Inspired by the strengthened mechanism of double-network and nanocomposite (NC) gels, we report a facile and green method to realize a mechanically stiff, fatigue-resistant, and adhesive-debranched waxy corn starch/poly(vinyl alcohol) double-crosslinked NC gel (W-Gel) skeleton material with dynamic noncovalent bonds. The in situ formation of debranched starch nanoparticles leads to self-assembly into three-dimensional waxberry-like microspheres, which act as physical cross-linkers by embedding themselves within network skeleton structures. The resulting hydrogel exhibited an excellent mechanical behavior, including a good stretchability over 1200% strain, a maximum compressive strength of up to 780.7 ± 27.8 kPa, and the ability to sustain as much weight as 4.6 kg (about 2000 times its own weight). Notably, the recovery efficiency exceeded 93% after the 60th compressive successive loading-unloading cycle at 50% strain. The hydrogel successfully adhered onto soft and hard substrates, such as skins, plastics, gauzes, glasses, and metals, manifesting in long-term, stable, sustained release of epigallocatechin gallate (EGCG). The EGCG-loaded W-Gels exhibited predominant antibacterial activity against both Gram-positive bacteria (Staphylococcus aureus) and Gram-negative bacteria (Escherichia coli and Salmonella typhus).

Fourier ptychography microscopy is a powerful tool for wide-field and high-spatial-resolution imaging. It can achieve a large field of view and high-spatial-resolution imaging with a low numerical aperture objective by capturing a series of low-resolution (LR) images that contain the information of different spatial frequencies and then stitching them together in the Fourier domain. Furthermore, the phase information of the object can also be recovered simultaneously. In this Letter, we propose a method to realize adaptive high-dynamic-range (HDR) LR image acquisition with a red, green, and blue camera, which needs only single exposure for each light-emitting diode (LED) illumination. According to the imaging principle of a color camera, the filter of each color channel still allows additional light within a certain wavelength range to pass through, but with a much smaller transmittance. By illuminating the sample with monochromatic light and combining the raw data from three color channels together, an HDR image corresponding to each LED illumination is obtained. The feasibility and good performance of our method are demonstrated by the experimental results.

Stable and efficient ground moving target tracking and refocusing is a hard task in synthetic aperture radar (SAR) data processing. Since shadows in video-SAR indicate the actual positions of moving targets at different moments without any displacement, shadow-based methods provide a new approach for ground moving target processing. This paper constructs a novel framework to refocus ground moving targets by using shadows in video-SAR. To this end, an automatic-registered SAR video is first obtained using the video-SAR back-projection (v-BP) algorithm. The shadows of multiple moving targets are then tracked using a learning-based tracker, and the moving targets are ultimately refocused via a proposed moving target back-projection (m-BP) algorithm. With this framework, we can perform detecting, tracking, imaging for multiple moving targets integratedly, which significantly improves the ability of moving-target surveillance for SAR systems. Furthermore, a detailed explanation of the shadow of a moving target is presented herein. We find that the shadow of ground moving targets is affected by a target’s size, radar pitch angle, carrier frequency, synthetic aperture time, etc. With an elaborate system design, we can obtain a clear shadow of moving targets even in X or C band. By numerical experiments, we find that a deep network, such as SiamFc, can easily track shadows and precisely estimate the trajectories that meet the accuracy requirement of the trajectories for m-BP.

In conventional multi-label learning framework, each example is assumed to be represented by a single feature vector and associated with multiple valid labels simultaneously. Nonetheless, real-world objects usually exhibit complicated properties which can have multi-view feature representation as well as false positive labeling. Accordingly, the problem of multi-view partial multi-label learning (MVPML) is studied in this paper, where each example is assumed to be presented by multiple feature vectors while associated with multiple candidate labels which are only partially valid. To learn from MVPML examples, a novel approach named FIMAN is proposed which makes use of multi-view feature representation to tackle the noisy labeling information. Firstly, an aggregate manifold structure over training examples is generated by adaptively fusing affinity information conveyed by feature vectors of different views. Then, candidate labels of each training example are disambiguated by preserving the feature-induced manifold structure in label space. Finally, the resulting predictive models are learned by fitting modeling outputs with the disambiguated labels. Extensive experiments on a number of real-world data sets show that FIMAN achieves highly competitive performance against state-of-the-art approaches in solving the MVPML problem.

Considering the lack of information on simultaneously removing multiple pharmaceuticals from water or wastewater by electrochemical methods, this study aimed to investigate the removal of multiple pharmaceuticals by electro-coagulation and electro-oxidation based on two types of electrodes (aluminum and graphite). The synthetic wastewater contained a nonsteroidal anti-inflammatory drug (diclofenac), a sulfonamide antibiotic (sulfamethoxazole) and a &beta, blocker (atenolol). The pharmaceutical removal with electro-oxidation was much higher than those with the electro-coagulation process, which was obtained from a five-cell graphite electrode system, while the removal of pharmaceuticals with aluminum electrodes was about 20% (20 &micro, M). In the electro-coagulation system, pharmaceutical removal was mainly influenced by the solubility or hydrophilicity of the compound. In the electro-oxidation system, the removal mechanism was influenced by the dissociation status of the compounds, which are attracted to the anode due to electrostatic forces and have a higher mass transformation rate with the electro-oxidation process. Therefore, atenolol, which was undissociated, cannot adequately be eliminated by electro-oxidation, unless the electrode&rsquo, s surface is large enough to increase the mass diffusion rate.

Perfluorooctanoic acid (PFOA) was separated and recovered using a foam flotation process aided by cationic surfactants. The PFOA removal efficiency was in the following decreasing order: OTAB (C8TAB) > DTAB (C10TAB) > CTAB (C16TAB) > TBAB, which indicates that cationic surfactants with an alkyl chain that had a similar length to that of PFOA had higher affinities to PFOA. PFOA removal slightly decreased with increasing ionic strength of the surfactant but did not change with the pH. PFOA could be completely removed in 20 min with 1.25 mM of OTAB in actual wastewater. The energy yield value of foam flotation with a cationic surfactant was much higher than those of other methods, which means that using foam flotation with a cationic surfactant as the collector is a simple, fast, and energy-efficient method to separate and recover PFOA from dilute water solutions.

The confocal Raman microscope (CRM) is a powerful tool in analytical science. Image quality is the most important performance indicator of CRM systems. The point spread function (PSF) is one of the most useful tools to evaluate the image quality of microscopic systems. A method based on a point-like object is proposed to measure the PSF of CRM, and the size effect of spherical objects is discussed. A series of phantoms are fabricated by embedding different sizes of polystyrene microspheres into polydimethylsiloxane matrix. The diameters of microspheres are from 0.2 µm to 5 µm. The phantoms are tested by measuring the PSF of a commercial CRM whose nominal lateral resolution is about 1 µm. Results of the PSF are obtained and the accuracy of resolution is used to evaluate the size effect of the microspheres. Experimental results are well consistent with theoretical analysis. The error of the PSF can be decreased by reducing the diameter of the microsphere but meanwhile the signal-to-noise ratio (S/N) will be lowered as well. The proper diameter of microspheres is proposed in consideration of the trade-off between the S/N and the measurement error of the PSF. Results indicate that the method provides a useful approach to measurement of the PSF and the resolution of the CRM.

The deformation measurement of a specular surface is of great importance during the quality inspection and installation of optical elements or wafers, especially those with large apertures. We propose a deflectometry method with speckle digital image correlation (DeSDIC) to realize the dynamic and high-accuracy measurement of the deformation on specular surfaces, with a simple system structure and robustness to noises and environmental vibrations. Random speckle pattern displayed on liquid crystal display is reflected by the original surface under test (SUT), and the distorted pattern is recorded by a camera. This originally distorted pattern is taken as the reference image, and the patterns captured afterwards are digitally correlated with the reference image to calculate the gradient change and deformation of the SUT. The theoretical relationships and an experimental one-step calibration scheme are proposed. Both static and dynamic deformations of a deformable mirror were experimentally measured to demonstrate the feasibility and accuracy of DeSDIC, which is comparable to phase-measuring deflectometry and interferometry.

We report a phase recovery method utilizing the chromatic aberration of telecentric objective and alternating projection algorithm. The axial defocus distance of each color channel and complex function of sample can be calculated simultaneously.

Click-through rate (CTR) prediction is an essential task in industrial applications such as video recommendation. Recently, deep learning models have been proposed to learn the representation of users' overall interests, while ignoring the fact that interests may dynamically change over time. We argue that it is necessary to consider the continuous-time information in CTR models to track user interest trend from rich historical behaviors. In this paper, we propose a novel Deep Time-Stream framework (DTS) which introduces the time information by an ordinary differential equations (ODE). DTS continuously models the evolution of interests using a neural network, and thus is able to tackle the challenge of dynamically representing users' interests based on their historical behaviors. In addition, our framework can be seamlessly applied to any existing deep CTR models by leveraging the additional Time-Stream Module, while no changes are made to the original CTR models. Experiments on public dataset as well as real industry dataset with billions of samples demonstrate the effectiveness of proposed approaches, which achieve superior performance compared with existing methods., Comment: 8 pages. arXiv admin note: text overlap with arXiv:1809.03672 by other authors

The behavior of slender ultra-high performance concrete (UHPC) columns under eccentric loading was investigated in this study. Eight slender UHPC columns were tested, with the major experimental variables being the amounts of confinement steel and steel fibers. The behavior of the columns was evaluated using various performance measures, including the failure pattern, load–deflection relationship, ductility, steel reinforcement strain, and axial load–moment interaction diagram. Results showed that the inclusion of steel fibers with a volume fraction of 0.75% or more effectively restrained spalling and crushing of the slender UHPC columns. In particular, the inclusion of a 1.5% volume fraction of steel fibers was able to compensate a 70% reduction in the confinement steel while still promoting the ductility of the slender columns under eccentric loading. Furthermore, analytical studies were conducted to estimate the crushing strain of the UHPC material in the outermost fiber of the column section under a combined flexural and axial action. The applicability of an analytical procedure and the current design codes for evaluating the total moment demand of the slender UHPC columns under the influence of the P-Delta effect were also evaluated.

The interface oxidation induced microstructural evolution of air plasma spayed (APS) thermal barrier coatings (TBCs) is investigated. As the oxidation progresses, the thermally grown oxide (TGO) formed at the YSZ/NiCrAlY interface develops from a bi-layered structure, to a three-layered structure, then to a four-layered structure of Al2O3, Y3Al5O12 (YAG), NiCr2O4 and NiO, respectively. An oxidation mechanism is proposed to account for the growth of the layered TGO. The YAG formation consumes the underlying alumina layer and facilitates the outward diffusion of Cr and Ni, leading to a detrimental effect on the TBC life.

Automated valet parking technology allows the vehicle to automatically drive into the parking lot and park itself without a human in the vehicle, relieving humans from parking completely. However, the existing automated valet parking system relies on infrastructural intelligence or pre-acquisition of parking lot maps. Given the disadvantages of these researches, this paper proposes a general method for automatic valet parking system based on autonomous exploration. This method depends on no prior knowledge of the parking lot. Our method extracts road structure from perception result using the Voronoi diagram. A multi-factor exploration strategy we proposed is used to generate exploration candidates for autonomous exploration from the road structure. Also, a motion planning method based on the lateral priority of the road guides the vehicle to the candidates while obeys the rules as far as possible. The autonomous exploration will continue until it finds free parking slots or all the spaces in the parking lot are explored. Related experiments have verified the effectiveness of the method presented in this paper.

Due to recent artificial intelligence (AI) technology progress, more and more applications present all-to-all, irregular or unpredictable communication patterns among compute nodes in high-performance computing (HPC) systems. Traditional communication infrastructures, e.g., torus or fat-tree interconnection networks, may not handle well their matchmaking problems with these newly emerging applications. For these typical non-random network topologies, there are already many communication-efficient application mapping algorithms. However, for the above unpredictable communication patterns, it is difficult to efficiently map their applications onto the non-random network topologies. In this case, a simple optimization is to map their applications with small diameter or average shortest path length (ASPL) among the assigned compute nodes. In this context, we recommend to use random network topologies as the communication infrastructures, which have drawn increasing attention for the use of HPC interconnects. In this study, we make a comparative study to analyze the performance impact of application mapping on non-random and random network topologies. We list several application mapping policies, and compare their job scheduling performances assuming that the communication patterns are unpredictable to the computing system. Evaluations with a large compound application workload show that, when compared to non-random topologies, random topologies can reduce the average turnaround time up to 39.3% by a random connected mapping method and up to 72.1% by a diameter/ASPL-based mapping method.

The steadily increasing number of compute nodes in high-performance computing (HPC) systems calls for more efficient interconnection networks and application mapping strategies. Traditionally, the user applications are mapped onto a regular network with nearby compute nodes to reduce the network distances. However, it is likely that the rigid mapping limitation on regular topologies may largely delay the dispatch time when the workload is heavy at a short period. In this work, we target the applications with unknown communication patterns over random interconnection networks, which have drawn increasing attention in the HPC world due to their low diameter and low average shortest path length (ASPL). We propose to use topology embedding metrics, i.e., diameter and ASPL, for application mapping to overcome the rigid limitation on regular topologies. We investigate the time-space tradeoff among several application mapping algorithms using a compound application workload. Evaluation results show that, when compared to the baseline random connected topology mapping, the proposed diameter/ASPL-based topology mapping algorithms reduce up to 48.0% makespan and up to 78.1% average turnaround time and improve up to 1.9X system utilization when the instantaneous workload is heavy over random interconnection networks.

In this study, the effects of integrating chloride addition and ultrasonication with electrocoagulation (EC) to remove passive layers were investigated. Experimental results were evaluated by conducting an impedance modulus analysis and by considering the amounts of aluminum released. The effect of different operating parameters, such as the phosphate concentration, ultrasonic power, chloride concentration, and the number of cycles on the efficiency of phosphate removal was examined. The results indicated that the integrated ultrasonic and chloride addition process effectively enhanced the energy efficiency and lifespan of plates due to the reduction of the electric impedance. The concentration of released aluminum increased from 150 to 440 mg/L with the increase of ultrasound from 0 to 100 W. After three test cycles, the use of the ultrasonic process and chloride addition with EC enhanced the phosphate removal efficiency. The removal efficiency of the EC process decreased to 30%; however, the removal efficiency of the integrated process increased to 90%. Scanning electron microscope images showed that using ultrasound could successfully remove the passivation layer on the electrode to enhance the phosphate removal efficiency. However, excessive ultrasonic intensity would result in a thicker passivation layer and thus decrease the removal efficiency. It suggests that the coupled use of chloride addition and ultrasonication in the EC process effectively reduced the passivation layer, thereby increasing the energy efficiency while decreasing the impedance modulus and enhancing the concentration of aluminum ions.

This paper presents the experimental results of small-scale model tests of an instrumented floating energy pile group in which the piles were embedded in dry medium-dense sand and subjected to the seasonal temperature pattern of the city of Nanjing in China. The study also included a model test to assess the effect of including nonthermal piles on the thermo-mechanical behaviour of the floating energy pile group. For comparison, a model test of a single floating energy pile embedded in the same soil and subjected to a similar temperature pattern was also conducted. The results show that the thermo-mechanical behaviour of an energy pile group is different from that of a single energy pile in terms of the thermally induced change in axial pile stress and the displacement of the pile top and tip. This difference in behaviour could be explained by the higher lateral confining pressure expected on a single pile than on a pile in a group due to pile interaction effects, which could lead to different end restraint boundary conditions. We conclude that the thermo-mechanical behaviour of an energy pile is controlled mainly by the end restraint boundary conditions.

In this paper, the authors use Climate Research Unit (CRU) land surface air temperature (SAT) and empirical orthogonal function (EOF)-based spectral optimal averaging (SOA) method to estimate the regional average annual mean SAT series for East Asia since 1901 with a comparison of arithmetic averaging (AA). The SOA method minimizes the first-M-mode mean square error (MSE) between the estimates and true SAT anomalies series. In the optimization process, the errors of CRU SAT dataset are used to test their contribution to the uncertainties of regional average annual mean SAT series estimated by SOA method. Furthermore, the MSE of SOA estimates is also calculated. Our results show that (i) the EOF-based SOA method can accurately estimate the regional average SAT anomalies series than AA method with smaller error, particularly when the sampling is poor, (ii) the SAT of East Asia has also experienced pronounced interdecadal cycles during last century, with the linear trend of being 0.91 °C(100a)−1, (iii) the errors of SOA estimates are large and close to 0.2 °C at the beginning of the twentieth century and then declining to 0.1 °C by 1960s, and (iv) the SOA estimates are non-sensitive to the values of error variance in the range of 0.001 to 1.0, but tends to zero as error variance tends infinity or extremely small values near or equal to zero.

The deflection demand of slender reinforced concrete columns could be magnified substantially due to its interaction with the axial load, i.e., the P-Delta effect. This study experimentally investigated the behavior of ten slender high-strength concrete columns under concentric axial loads. The experimental variables were the amounts of confinement steel and steel fibers. The adopted high-strength concrete material had an average compressive strength of about 100 MPa. When it was reinforced with steel fibers at a volume fraction of 0.75% or more, it showed the characteristics of high performance fiber reinforced concrete, i.e., tensile strain hardening behavior and multiple narrow cracks. The behavior of the slender high-strength columns was evaluated using various performance measures, such as the failure type, peak strength, dilatation, ductility, steel reinforcement strain, and the P-Delta effect. The results showed that all the columns failed significantly due to the adverse P-Delta effect. The inclusion of steel fibers with a volume fraction of 0.75% or more effectively restrained concrete spalling and crushing and also promoted multiple narrow cracking patterns in the column. In particular, the inclusion of a 1.5% volume fraction of steel fibers enabled the column without steel confinement to exhibit comparable behavior to that of the control specimen that was detailed according to current codes of practice. Furthermore, reasonably accurate equations were suggested for predicting the axial strength of the tested slender, high-strength concrete columns.