RESUMEN
The precise recognition of entire classroom meta-actions is a crucial challenge for the tailored adaptive interpretation of student behavior, given the intricacy of these actions. This paper proposes a Dynamic Position Embedding-based Model for Student Classroom Complete Meta-Action Recognition (DPE-SAR) based on the Video Swin Transformer. The model utilizes a dynamic positional embedding technique to perform conditional positional encoding. Additionally, it incorporates a deep convolutional network to improve the parsing ability of the spatial structure of meta-actions. The full attention mechanism of ViT3D is used to extract the potential spatial features of actions and capture the global spatial-temporal information of meta-actions. The proposed model exhibits exceptional performance compared to baseline models in action recognition as observed in evaluations on public datasets and smart classroom meta-action recognition datasets. The experimental results confirm the superiority of the model in meta-action recognition.
RESUMEN
Oil/water separation is an essential process in the petrochemical industry, environmental remediation, and water treatment. Alkanes are the major components of crude oil and are difficult to separate once they form emulsions in water. Much less attention has been focused on the feature of liquid alkanes that could, in turn, influence the separation process. The role of chain length is systematically studied herein by separating the alkane-in-water emulsions with superwetting titanium microchannels of 14-55 µm. The chain length covers the entire liquid alkane spectrum with carbon numbers ranging from 6 to 16. The separation efficiency decreases while the TOC content increases with the chain length of liquid alkanes for a given channel. This is attributed to the small Ostwald ripening rate with the long chains, which stabilize the oil droplets of small sizes that could pass through the zigzag channels. Accordingly, a high separation efficiency of >99.97% and a low TOC content of <5 ppm are achieved with superhydrophilic channels of 14 µm for alkanes with less than 12 carbons. The metallic microchannels surpass the conventional organic membranes and inorganic frameworks over the entire liquid n-alkane spectrum, paving the way for the future development of oil/water separation using porous metals.
RESUMEN
The degradation mechanisms for environmental barrier coatings (EBCs) under high-temperature water vapour conditions are vital for the service of aero-engine blades. This study proposes a theoretical model of high-temperature water vapour corrosion coupled with deformation, mass diffusion and chemical reaction based on the continuum thermodynamics and the actual water vapour corrosion mechanisms of an EBC system. The theoretical model is suitable for solving the stress and strain fields, water vapour concentration distribution and coating corrosion degree of an EBC system during the water vapour corrosion process. The results show that the thickness of the corrosion zone on the top of the EBC system depended on water vapour diffusion, which had the greatest influence on the corrosion process. The top corroded area of the rare-earth silicate EBC system was significantly evident, and there was a clear dividing line between the un-corroded and corroded regions.