Clothing Parsing Based on Multi-Scale Fusion and Improved Self-Attention Mechanism

Due to the lack of long-range association and spatial location information,fine details and accurate boundaries of complex clothing images cannot always be obtained by using the existing deep learning-based methods.This paper presents a convolutional structure with multi-scale fusion to optimize the step of clothing feature extraction and a self-attention module to capture long-range association information.The structure enables the self-attention mechanism to directly participate in the process of information exchange through the down-scaling projection operation of the multi-scale framework.In addition,the improved self-attention module introduces the extraction of 2-dimensional relative position information to make up for its lack of ability to extract spatial position features from clothing images.The experimental results based on the colorful fashion parsing dataset(CFPD)show that the proposed network structure achieves 53.68%mean intersection over union(mIoU)and has better performance on the clothing parsing task.

Hierarchical multihead self-attention for time-series-based fault diagnosis

Fault diagnosis is important for maintaining the safety and effectiveness of chemical process.Considering the multivariate,nonlinear,and dynamic characteristic of chemical process,many time-series-based data-driven fault diagnosis methods have been developed in recent years.However,the existing methods have the problem of long-term dependency and are difficult to train due to the sequential way of training.To overcome these problems,a novel fault diagnosis method based on time-series and the hierarchical multihead self-attention(HMSAN)is proposed for chemical process.First,a sliding window strategy is adopted to construct the normalized time-series dataset.Second,the HMSAN is developed to extract the time-relevant features from the time-series process data.It improves the basic self-attention model in both width and depth.With the multihead structure,the HMSAN can pay attention to different aspects of the complicated chemical process and obtain the global dynamic features.However,the multiple heads in parallel lead to redundant information,which cannot improve the diagnosis performance.With the hierarchical structure,the redundant information is reduced and the deep local time-related features are further extracted.Besides,a novel many-to-one training strategy is introduced for HMSAN to simplify the training procedure and capture the long-term dependency.Finally,the effectiveness of the proposed method is demonstrated by two chemical cases.The experimental results show that the proposed method achieves a great performance on time-series industrial data and outperforms the state-of-the-art approaches.

Self-Attention Spatio-Temporal Deep Collaborative Network for Robust FDIA Detection in Smart Grids

False data injection attack(FDIA)can affect the state estimation of the power grid by tampering with the measured value of the power grid data,and then destroying the stable operation of the smart grid.Existing work usually trains a detection model by fusing the data-driven features from diverse power data streams.Data-driven features,however,cannot effectively capture the differences between noisy data and attack samples.As a result,slight noise disturbances in the power grid may cause a large number of false detections for FDIA attacks.To address this problem,this paper designs a deep collaborative self-attention network to achieve robust FDIA detection,in which the spatio-temporal features of cascaded FDIA attacks are fully integrated.Firstly,a high-order Chebyshev polynomials-based graph convolution module is designed to effectively aggregate the spatio information between grid nodes,and the spatial self-attention mechanism is involved to dynamically assign attention weights to each node,which guides the network to pay more attention to the node information that is conducive to FDIA detection.Furthermore,the bi-directional Long Short-Term Memory(LSTM)network is introduced to conduct time series modeling and long-term dependence analysis for power grid data and utilizes the temporal selfattention mechanism to describe the time correlation of data and assign different weights to different time steps.Our designed deep collaborative network can effectively mine subtle perturbations from spatiotemporal feature information,efficiently distinguish power grid noise from FDIA attacks,and adapt to diverse attack intensities.Extensive experiments demonstrate that our method can obtain an efficient detection performance over actual load data from New York Independent System Operator(NYISO)in IEEE 14,IEEE 39,and IEEE 118 bus systems,and outperforms state-of-the-art FDIA detection schemes in terms of detection accuracy and robustness.

Synthesis and Coloring Mechanism of Metallic Glaze in R_(2)O-RO-Al_(2)O_(3)-SiO_(2)-P_(2)O_(5)System

A copper-red and silver-white metallic glaze of R_(2)O-RO-Al_(2)O_(3)-SiO_(2)-P_(2)O_(5)system was synthesized by adjusting the firing temperature and glaze components.The coloration mechanism of the metallic glaze was revealed via investigation of the microstructure of the glaze.Our research reveals that the metallic glaze with different colors is mainly due to the amount of Fe_(2)O_(3).The metallic glaze shows a silver-white luster due to a structural color ofα-Fe_(2)O_(3)crystals with a good orientation when the sample contains 0.0939 mol of Fe_(2)O_(3),maintaining temperatures at 1150℃for 0.5 h.The metallic glaze is copper-red which is dominated by the coupling of chemical and structural color ofα-Fe_(2)O_(3)crystals when the sample contains 0.0783 mol of Fe_(2)O_(3).After testing the amount of SiO_(2),we find that 4.0499 mol is the optimal amount to form the ceramic network,and 0.27 mol AlPO_(4)is the best amount to promote phase separation.

Lightweight Cross-Modal Multispectral Pedestrian Detection Based on Spatial Reweighted Attention Mechanism

Multispectral pedestrian detection technology leverages infrared images to provide reliable information for visible light images, demonstrating significant advantages in low-light conditions and background occlusion scenarios. However, while continuously improving cross-modal feature extraction and fusion, ensuring the model’s detection speed is also a challenging issue. We have devised a deep learning network model for cross-modal pedestrian detection based on Resnet50, aiming to focus on more reliable features and enhance the model’s detection efficiency. This model employs a spatial attention mechanism to reweight the input visible light and infrared image data, enhancing the model’s focus on different spatial positions and sharing the weighted feature data across different modalities, thereby reducing the interference of multi-modal features. Subsequently, lightweight modules with depthwise separable convolution are incorporated to reduce the model’s parameter count and computational load through channel-wise and point-wise convolutions. The network model algorithm proposed in this paper was experimentally validated on the publicly available KAIST dataset and compared with other existing methods. The experimental results demonstrate that our approach achieves favorable performance in various complex environments, affirming the effectiveness of the multispectral pedestrian detection technology proposed in this paper.

Remaining Useful Life Prediction of Rail Based on Improved Pulse Separable Convolution Enhanced Transformer Encoder

In order to prevent possible casualties and economic loss, it is critical to accurate prediction of the Remaining Useful Life (RUL) in rail prognostics health management. However, the traditional neural networks is difficult to capture the long-term dependency relationship of the time series in the modeling of the long time series of rail damage, due to the coupling relationship of multi-channel data from multiple sensors. Here, in this paper, a novel RUL prediction model with an enhanced pulse separable convolution is used to solve this issue. Firstly, a coding module based on the improved pulse separable convolutional network is established to effectively model the relationship between the data. To enhance the network, an alternate gradient back propagation method is implemented. And an efficient channel attention (ECA) mechanism is developed for better emphasizing the useful pulse characteristics. Secondly, an optimized Transformer encoder was designed to serve as the backbone of the model. It has the ability to efficiently understand relationship between the data itself and each other at each time step of long time series with a full life cycle. More importantly, the Transformer encoder is improved by integrating pulse maximum pooling to retain more pulse timing characteristics. Finally, based on the characteristics of the front layer, the final predicted RUL value was provided and served as the end-to-end solution. The empirical findings validate the efficacy of the suggested approach in forecasting the rail RUL, surpassing various existing data-driven prognostication techniques. Meanwhile, the proposed method also shows good generalization performance on PHM2012 bearing data set.

Sentiment classification model for bullet screen based on self-attention mechanism

With the development of short video industry,video and bullet screen have become important ways to spread public opinions.Public attitudes can be timely obtained through emotional analysis on bullet screen,which can also reduce difficulties in management of online public opinions.A convolutional neural network model based on multi-head attention is proposed to solve the problem of how to effectively model relations among words and identify key words in emotion classification tasks with short text contents and lack of complete context information.Firstly,encode word positions so that order information of input sequences can be used by the model.Secondly,use a multi-head attention mechanism to obtain semantic expressions in different subspaces,effectively capture internal relevance and enhance dependent relationships among words,as well as highlight emotional weights of key emotional words.Then a dilated convolution is used to increase the receptive field and extract more features.On this basis,the above multi-attention mechanism is combined with a convolutional neural network to model and analyze the seven emotional categories of bullet screens.Testing from perspectives of model and dataset,experimental results can validate effectiveness of our approach.Finally,emotions of bullet screens are visualized to provide data supports for hot event controls and other fields.

Keyphrase Generation Based on Self-Attention Mechanism

Keyphrase greatly provides summarized and valuable information.This information can help us not only understand text semantics,but also organize and retrieve text content effectively.The task of automatically generating it has received considerable attention in recent decades.From the previous studies,we can see many workable solutions for obtaining keyphrases.One method is to divide the content to be summarized into multiple blocks of text,then we rank and select the most important content.The disadvantage of this method is that it cannot identify keyphrase that does not include in the text,let alone get the real semantic meaning hidden in the text.Another approach uses recurrent neural networks to generate keyphrases from the semantic aspects of the text,but the inherently sequential nature precludes parallelization within training examples,and distances have limitations on context dependencies.Previous works have demonstrated the benefits of the self-attention mechanism,which can learn global text dependency features and can be parallelized.Inspired by the above observation,we propose a keyphrase generation model,which is based entirely on the self-attention mechanism.It is an encoder-decoder model that can make up the above disadvantage effectively.In addition,we also consider the semantic similarity between keyphrases,and add semantic similarity processing module into the model.This proposed model,which is demonstrated by empirical analysis on five datasets,can achieve competitive performance compared to baseline methods.

Kinetics and mechanisms of solvent extraction and separation of La(Ⅲ) and Ni(Ⅱ) with DEHPA in petrofin

The kinetics of extractive separation of La(Ⅲ) and Ni(Ⅱ) from nitrate medium in the presence of lactic acid (HLac) using di-2-ethylhexyl phosphoric acid (DEHPA) diluted in petrofin was investigated using a cell with constant interfacial area and continuous stirring. The effects of stirring speed, interfacial area, pH, HLac concentration, extractant concentration, concentrations of metal ions and temperature on the extraction rate were examined. Results suggested that the extraction regime is diffusion-controlled. The reaction which occurred at the interface was found to be the rate-determining step. The extraction rates of both metal ions are found to be independent of pH. The extraction rates of La(Ⅲ) and Ni(Ⅱ) are first-order dependent with respect to lactic acid and metal ions (La(Ⅲ) and Ni(Ⅱ)) concentrations. The extraction rate of La(Ⅲ) is first-order dependent on DEHPA concentration and for Ni(Ⅱ), it varies to the power of 1.5. The separation of La(Ⅲ) and Ni(Ⅱ) from nitrate solution is possible at low interfacial area and low stirring speed.

SrTiO3/BiOI heterostructure: Interfacial charge separation, enhanced photocatalytic activity, and reaction mechanism

Heterostructured photocatalysts provide an effective way to achieve enhanced photocatalytic performances through efficient charge separation.Although both wide-and narrow-band-gap photocatalysts have been widely investigated,the charge separation and transfer mechanism at the contacting interface of the two has not been fully revealed.Here,a novel SrTiO3/BiOI(STB)heterostructured photocatalyst was successfully fabricated by using a facile method.The heterostructure in the photocatalyst extends the photoabsorption to the visible light range,and thus,high photocatalytic NO removal performance can be achieved under visible light irradiation.A combination of experimental and theoretical evidences indicated that the photogenerated electrons from the BiOI semiconductor can directly transfer to the SrTiO3 surface through a preformed electron delivery channel.Enhanced electron transfer was expected between the SrTiO3 and BiOI surfaces under light irradiation,and leads to efficient ROS generation and thus a high NO conversion rate.Moreover,in situ diffused reflectance infrared Fourier transform spectroscopy revealed that STB can better inhibit the accumulation of the toxic intermediate NO2 and catalyze the NO oxidation more effectively.This work presents a new insight into the mechanism of the interfacial charge separation in heterostructures and provides a simple strategy to promote the photocatalytic technology for efficient and safe air purification.

Separation of atropisomers by chiral liquid chromatography and thermodynamic analysis of separation mechanism

In the pharmaceutical industry, the analysis of atropisomers is of considerable interest from both scientific and regulatory perspectives. The compound of interest contains two stereogenic axes due to the hindered rotation around the single bonds connecting the aryl groups, which results in four potential configurational isomers(atropisomers). The separation of the four atropisomers was achieved on a derivatized β-cyclodextrin bonded stationary phase. Further investigation showed that low temperature conditions, including sample preparation(-70 °C), sample storage(-70 °C), and chromatographic separation(6 °C), were critical to preventing interconversion. LC-UV-laser polarimetric analysis identified peaks 1 and 2 as a pair of enantiomers and peaks3 and 4 as another. Thermodynamic analysis of the retention data indicated that the separation of the pairs of enantiomers is primarily enthalpy controlled as indicated by the positive slope of the van't Huff plot. The difference in absolute Δ(Δ H), ranged from 2.20 k J/mol to 2.42 k J/mol.

3D inner-outer asymmetric sponge for enormous-volume emulsion wastewater treatment based on a new“demulsification-transport”mechanism

Although oily wastewater treatment realized by superwetting materials has attracted heightened attention in recent years,how to treat enormous-volume emulsion wastewater is still a tough problem,which is ascribed to the emulsion accumulation.Herein,to address this problem,a material is presented by subtly integrating chemical demulsification and 3D inner-outer asymmetric wettability to a sponge substrate,and thus wettability gradient-driven oil directional transport for achieving unprecedented enormous-volume emulsion wastewater treatment is realized based on a“demulsification-transport”mechanism.The maximum treatment volume realized by the sponge is as large as 3 L(2.08×10^(4) L per cubic meter of the sponge)in one cycle,which is about 100 times of the reported materials.Besides,owing to the large pore size of the sponge,9000 L m^(2)h^(-1)(LMH)separation flux and 99.5%separation efficiency are realized simultaneously,which overcomes the trade-off dilemma.Such a 3D inner-outer asymmetric sponge displaying unprecedented advantage in the treatment volume can promote the development of the oily wastewater treatment field,as well as expand the application prospects of superwetting materials,especially in continuous water treatment.

Advances in depressants for flotation separation of Cu–Fe sulfide minerals at low alkalinity:A critical review

The flotation separation of Cu–Fe sulfide minerals at low alkalinity can be achieved using selective depressants.In the flotation system of Cu–Fe sulfide minerals,depressants usually preferentially interact with the pyrite surface to render the mineral surface hydrophilic and hinder the adsorption of the collector.This review summarizes the advances in depressants for the flotation separation of Cu–Fe sulfide minerals at low alkalinity.These advances include use of inorganic depressants (oxidants and sulfur–oxygen compounds),natural polysaccharides (starch,dextrin,konjac glucomannan,and galactomannan),modified polymers (carboxymethyl cellulose,polyacrylamide,lignosulfonate,and tricarboxylate sodium starch),organic acids (polyglutamic acid,sodium humate,tannic acid,pyrogallic acid,salicylic acid,and lactic acid),sodium dimethyl dithiocarbamate,and diethylenetriamine.The potential application of specific inorganic and organic depressants in the flotation separation of Cu–Fe sulfide minerals at low alkalinity is reviewed.The advances in the use of organic depressants with respect to the flotation separation of Cu–Fe sulfide minerals are comprehensively detailed.Additionally,the depression performances and mechanisms of different types of organic depressants on mineral surfaces are summarized.Finally,several perspectives on depressants vis-à-vis flotation separation of Cu–Fe sulfide minerals at low alkalinity are proposed.

Mechanism of pool separation and stratification in laser remelting and cladding

The laser remelting with a two-layer material system (upper material was Al-30 % Ti-20 % Ni alloy,substrate was commercial aluminum alloy) and the laser cladding of a commercial 45 steel with copper Powder (including 25%SiC) were carried out using a 2kW continuous CO2 laser. For the case of laser remelting, a upper Pool in the alloying layer and a lower Pool in the substrate separated by the unmelted Al-Ti-Ni alloy were observed. For laser cladding, a stratified Pool was observed, whose top layer was Cu alloy liquid and bottom was Fe alloy liquid. The mechanism of laser Pool separation and stratification is illustrated by numerical calculation of heat transter process of the two-layer system, combining with material physical properties (especially mixed enthalpy). A classification criterion for laser Pool with the two-layer material system has been presented and four types of the laser Pool are divided into unique Pool, separated Pool, mixed Pool and stratified pool,which provides a theoretical basis for obtaining a excellent surface coating.

Actional Mechanism of Trifluoroacetic Acid for the Separation of Biopolymers by Reversed-phase Liquid Chromatography

The present paper covers the actional mechanism of trifluoroacetic acid for the separation of biopolymers investigated by using the parameters of stoichiometric displacement model for retention(SDM-R) in reversed-phase liquid chromatography. It was found that the trifluoroacetic acid(TFA) may participate in, or stimulate the association among displacing agent molecules in mobile phase, and decrease the affinity of both the associate molecules of the displacing agent and the TFA-protein ion-pairing. The former dominates over the separation selectivity of biopolymers as the concentration of TFA is lower than a given value, and the two contrary functions partly offset to each other and the latter dominates as its concentration is greater than the given value.

STUDY ON THE SEPARATION AND EXTRACTION MECHANISM OF RARE EARTH ELEMENTS BY MEANS OF REVERSED-PHASE PAPER CHROMATOGRAPHY

Reversed-phase paper chromatography technique is used for study on the extraction mechanism and sep- aration of rare earth elements.As the stationary phase,chromatographic paper strips are impregnated with a solution of monomyristyl phosphoric acid (MPA) in chloroform.Mineral acids are used as developers. The effect of concentration of acids and/or salts upon R_f has been investigated.According to the re- sults of R_f values for a given rare earth element in various acids,the order of extraction ability is HCl>HNO_3>H_2SO_4.A tetrad effect is clearly observed.for the R_f value of rare earth elements.The effects of other parameters on the R_f value,such as the quantities of extractant retained by the paper and the temperature are also examined.Based on the determination of the molar ratio of MPA to rare earth elements and the number of H^+ ions released in extraction reaction,a reasonable mechanism is proposed.The mutual separation of heavy rare earth elements will be better than that of the light rare earth group because of the larger separation coefficient of the former.A mixture of Ho-Er-Tm-Lu is successfully separated by the present method.

Study on the separation and mechanism of the enzymolysis of Dioscorea Zingiberensis

Based on a through study of the chemical compositions of Dioscorea Zingiberensis, a new separation technology is presented in this paper. This technology can effectively separate the starch and cellulose from Dioscorea Zingiberensis, thus the output of diosgenin is raised by 20%. Meanwhile using the suspension after separation, the synergy of enzymes, which are under the best condition for enzymolysis is studied, and an enzymolysis model, which uses two enzymes at the same time, is established. This model provides a theoretical basis for finding new enzyme resources. What＇s more orthogonal experiments show that by using this model, either the output of diosgenin can be increased by 50.58% or the amount of acid used can be reduced by 83% which means less pollution.

Liquid phase separating mechanism and preparation techniques of immiscible alloys

Immiscible alloys have attracted growing interest for their valuable physical and mechanical properties. However, their production is difficult because of metallurgical problems in which there is a serious tendency for gravity separation in the region of the miscibility gap. So far the study on the liquid separation mechanism is still one of the important projects in the spatial materials science and the spatial fluid science. The studied results about the liquid phase separating mechanism of immiscible alloys are presented, at the same time the preparation techniques of homogeneous immiscible alloys are summarized, and the existing problems and the related researching areas in the future are also pointed out.

COF-based membranes for liquid phase separation:Preparation,mechanism and perspective

Covalent organic frameworks(COFs)are a new kind of crystalline porous materials composed of organic molecules connected by covalent bonds,processes the characteristics of low density,large specific surface area,adjustable pore size and structure,and easy to functionalize,which have been widely used in the field of membrane separation technology.Recently,there are more and more researches focusing on the preparation methods,separation application,and mechanism of COF membranes,which need to be further summarized and compared.In this review,we primarily summarized several conventional preparation methods,such as two-phase interfacial polymerization,in-situ growth on substrate,unidirectional diffusion method,layer-by-layer assembly method,mixed matrix membranes,and so on.The advantages and disadvantages of each method are briefly summarized.The application potential of COF membrane in liquid separation are introduced from four aspects:dyeing wastewater treatment,heavy metal removal,seawater desalination and oil-water separation.Then,the mechanisms including pore structure,hydrophilic/hydrophobic,electrostatic repulsion/attraction and Donnan effect are introduced.For the efficient removal of different kind of pollutions,researchers can select different ligands to construct membranes with specific pore size,hydrophily,salt or organic rejection ability and functional group.The ideas for the design and preparation of COF membranes are introduced.Finally,the future direction and challenges of the next generation of COF membranes in the field of separation are prospected.

Mechanical mechanism of overlying strata breaking and development of fractured zone during close-distance coal seam group mining

This study mainly investigates the mechanical mechanism of overlying strata breaking and the development of fractured zones during close-distance coal seam group mining in the Gaojialiang coal mine.First,a mechanical model for the second"activation"of broken overlying strata is established,and the related mechanical"activation"conditions are obtained.A recursive formula for calculating the separation distance of overlying strata is deduced.Second,a height determining method for predicting the height of fractured zones during close-distance coal seam group mining is proposed based on two values,namely,the separation distance and ultimate subsidence value of overlying strata.This method is applied to calculate the fractured zone heights in nos.20107 and 20307 mining faces.The calculated results are almost equal to the field observation results.Third,a modified formula for calculating the height of a waterflowing fractured zone is proposed.A comparison of the calculated and observed results shows that the errors are small.The height determining method and modified formula not only build a theoretical foundation for water conservation mining at the Gaojialiang coal mine,but also provide a reference for estimating the height of water-flowing fractured zones in other coal mines with similar conditions.