Search

Safety is one of the important topics in the field of civil aviation. Auxiliary Power Unit (APU) is one of important components in aircraft, which provides electrical power and compressed air for aircraft. The hazards in APU are prone to cause economic losses and even casualties. So, actively identifying the hazards in APU before an accident occurs is necessary. In this paper, a Hybrid Deep Neural Network (HDNN) based on multi-time window convolutional neural network-Bidirectional Long Short-Term Memory (CNN-BiLSTM) neural network is proposed for active hazard identification of APU in civil aircraft. In order to identify the risks caused by different types of failures, the proposed HDNN simultaneously integrates three CNN-BiLSTM basic models with different time window sizes in parallel by using a fully connected neural network. The CNN-BiLSTM basic model can automatically extract features representing the system state from the input data and learn the time information of irregular trends in the time series data. Nine benchmark models are compared with the proposed HDNN. The comparison results show that the proposed HDNN has the highest identification accuracy. The HDNN has the most stable identification performance for data with imbalanced samples.

Ion-exchanged glasses are widely used in optoelectronic devices due to its high strength and optical waveguide effects. The MgF2 films deposited on the ion-exchanged glasses and the ion-exchanged glasses bombarded by argon ions (100 eV) at 220 °C were investigated in the present work. The microstructure of MgF2 films and the structural and physico-chemical properties of the ion-exchanged glasses before and after deposition and bombardment were analyzed by grazing incident X-ray diffraction (GIXRD), field emission scanning electron microscopy (FE-SEM), electron probe micro analysis (EPMA), micro-Raman spectroscopy and surface stress meter. It is shown that MgF2 films exhibit a polycrystalline and microcolumnar structure. The peak of potassium ions concentration distribution in the ion-exchanged glasses, accompanied by outward diffusion of potassium ions, decreases and moves from glass surface to the inner part to different degrees during deposition and bombardment. The potassium ions concentration in the region near the surface of argon ion-bombarded glasses is almost the same as the MgF2-coated glasses. Furthermore, the position and intensity of Raman bands in the subsurface region of the ion-exchanged glasses show obvious changes before and after deposition and bombardment. The values of compressive stress and depth of layer in the MgF2-coated ion-exchanged glasses and the argon ion-bombarded ion-exchanged glasses are far away lower than that of its values obtained in the ion-exchanged glasses. Overall results have shown that variations of chemical composition, Raman bands and surface compressive stress in the exchanged region indicate subsurface modification of the ion-exchanged glasses induced by deposition and bombardment, which provide useful information for the use of the ion-exchanged glasses in further.

In the influencing factors of reading distance of RFID tags, the tags’ geometric spatial structures have an important aspect on the corresponding reading distance. In order to achieve the goal of improving the reading distance of RFID tags, a novel three-dimensional (3D) position measurement and structure prediction method for RFID tag group based on deep belief network (DBN) is proposed. First, a 3D structure prediction system for RFID tags, which is based on stereovision, is designed. Second, an image deblurring scheme which combines the knife-edge with Wiener filtering method is adopted to restore the blur image in the 3D position measurement process. Then, RFID tags’ 3D positions are acquired by using the image matching method. Finally, the relationship model between RFID tags’ 3D positions and reading distance is built by DBN. The experimental results show that the relationship between tag group’s 3D positions and reading distance can be well built by DBN. The established DBN can forecast the reading distance accurately. The MAPE and RMSE of DBN prediction results are 1.64% and 0.168. The DBN can find the best 3D spatial distribution structure of RFID tag group corresponding to the maximum reading distance, which can guide the 3D distribution of the tag group.

Objectives The purpose of this study was to estimate the fatigue life of premolars restored with two dental ceramics, lithium disilicate (LD) and polymer infiltrated ceramic (PIC) using the numerical method and compare it with the published in vitro data. Methods A premolar restored with full-coverage crown was digitized. The volumetric shape of tooth tissues and crowns were created in Mimics®. They were transferred to IA-FEMesh for mesh generation and the model was analyzed with Abaqus. By combining the stress distribution results with fatigue stress–life (S–N) approach, the lifetime of restored premolars was predicted. Results The predicted lifetime was 1,231,318 cycles for LD with fatigue load of 1400 N, while the one for PIC was 475,063 cycles with the load of 870 N. The peak value of maximum principal stress occurred at the contact area (LD: 172 MPa and PIC: 96 MPa) and central fossa (LD: 100 MPa and PIC: 64 MPa) for both ceramics which were the most seen failure areas in the experiment. In the adhesive layer, the maximum shear stress was observed at the shoulder area (LD: 53.6 MPa and PIC: 29 MPa). Significance The fatigue life and failure modes of all-ceramic crown determined by the numerical method seem to correlate well with the previous experimental study.

We used force-field based grand-canonical Monte Carlo simulation method and density functional theory to study adsorption characteristics of carbon dioxide (CO2) molecules in a metal-organic framework (MOF) compound, [Zn(bdc)(dpds)]n. The studied MOF include a metal ion (Zn(II)), an anion organic linker (dianion of benzene dicarboxylicacid, bdc2−) and a neutral organic linker (4,4′-dipyridyldisulfide, dpds). Results from calculated adsorption isotherms and enthalpies of adsorption agree with the experimental data. The interactions between the adsorbed CO2 and the organic linkers were examined in simulations. Calculated results show available absorption sites are surrounded by two dpds ligands in which an S-S bond as an N-N′ spacer connect two pyridines. In contrast, the bdc2− ligand does not give a significant contribution to the substantial adsorption amount even though it contains the carboxylate group that provides available bonding site to CO2.

We have successfully synthesized a multilayer-structured composite based on a combination of polydopamine (PDA)-coated silver nanoparticles (Ag@PDA) and silver nanowires (AgNWs@PDA) using a flexible and rapid method. We incorporated the structure into polyvinyl alcohol (PVA) polymer to enhance thermal conductivity and dielectric properties of nanocomposite films. Compared with Ag@PDA/PVA and AgNWs@PDA/PVA films, the use of multilayer-structure facilitated increased thermal conductivity and reduced permittivity of the nanocomposite film. The maximum thermal conductivity of composite with 10 wt% filler loading was increased to 3.63 W/mK, which is an enhancement by nearly 20 times compared with the use of neat PVA matrix. The relative permittivity and dielectric loss were about 6.65 and 0.024 at 1 MHz, respectively. We discussed the underlying mechanisms in an effort to optimize the structure for polymer matrix nanocomposites.

Objective Numerous bond strength tests have been performed on dental adhesion experiments. Yet, the validity of these bond strength tests is controversial due to the name ( e.g. , “shear” or “tensile”) may not reflect to the true and complete stress situation, i.e. , assumed uniform shear or uniaxial tensile conditions. Thus, the aim of this study was to simulate and compare the stress distribution of and between shear bond strength (SBS), tensile bond strength (TBS), mold-enclosed shear bond strength (ME-SBS) and de novo lever-induced mold-enclosed shear bond strength (LIME-SBS) tests. Methods 3-Dimensional finite element method (FEM) was used on the dental resin-bonded surfaces ( i.e. , titanium alloy, dentine and porcelain) interphased with adhesive layer (thickness 5 μm) to simulate the mechanical tests. For ME-SBS, both polycarbonate and stainless steel molds were used. For LIME-SBS, stainless steel levers and molds with lengths of 3 mm, 6 mm, 12 mm, 15 mm and 18 mm were used. The applied loads on these models were 50 N, 100 N and 200 N. Results De novo LIME-SBS test was the most optimal configuration to evaluate “shear” bond strength of adhesive in regards to providing significantly high and uniform shear stress as well as eliminating tensile stress at the interface. The conventional SBS test created very high tensile stress at the load area, whereas the TBS created optimal tensile stress but shear stress indeed co-exist. The ME-SBS test could also eliminate some of the tensile stress. Similar stress distributions pattern appeared on the Ti-adhesive models, the dentine-adhesive models and porcelain-adhesive models. Significance None of the bond strength tests could give purely “shear” or “tensile” bond strength, but LIME-SBS seems to be the best model to evaluate the bond strength under true “shear” mode.

Background Social differences for individuals with autism spectrum disorder (ASD) could be related to attentional bias toward threatening emotional faces. However, studies investigating this attentional bias in individuals with ASD vary in paradigms and results. Methods We conducted a meta-analysis across different reaction time tasks, including dot-probe, spatial cuing, visual search and emotional Stroop tasks, to examine the attentional bias toward threatening emotional faces in individuals with ASD. Narrative synthesis further summarized possible moderators affecting attentional bias. Results We used systematic searches to identify 21 empirical studies with 1,805 participants in total (ASD n = 909). Individuals with ASD showed a small but significant attentional bias toward threatening emotional faces over other faces (g = 0.162), which was similar to individuals with typical development (g = 0.203). Moderator analysis showed that adults with ASD showed a larger attentional bias than children with ASD. For the dot-probe and spatial cuing paradigms, attentional bias toward threatening emotional faces was observed when stimuli were presented for less than 500 ms, and bias away from threatening emotional faces was shown when stimuli were presented for over 500 ms. For the visual search paradigm, attentional bias was significant in button-press tasks, but not in touch-screen tasks. Furthermore, attentional bias was only observed when the stimuli were schematic faces and reference materials were happy faces. Conclusions Individuals with ASD showed a small but significant attentional bias toward threatening emotional faces. Furthermore, significant moderators included stimulus presentation, response type, reference face type, stimulus type, and age. Thus, more research is needed to further explore attentional bias toward threatening emotional faces in individuals with ASD.

Recently, the discriminant learning denoising model has deserved special attention due to its outstanding denoising property. This paper designs the RFID multi-label dynamic localization system based on visual measurement to avoid electromagnetic interference. Firstly, the potential sharp images are estimated via denoising images acquired by dual CCD. Then the labels are matched by template. Finally, Shaft Alignment Method (SAM) models the multi-label 3D coordinates in world space and image pixel. Flexible Feed-forward Denoising Convolutional Neural Network (FDnCNN) is proposed to reach the equilibrium between denoising effect and image particulars without producing artifacts in GPU effectively and flexibly. FDnCNN increases the image quality by at least 0.5 dB than WNNM and DnCNN. The 3D coordinate measurement system is evaluated by error, which proves the feasibility and effectiveness of the positioning system.