Search

Emotion recognition commonly relies on single-modal recognition methods, such as voice and video signals, which demonstrate a good practicability and universality in some scenarios. Nevertheless, as emotion-recognition application scenarios continue to expand and the data volume surges, single-modal emotion recognition proves insufficient to meet people’s needs for accuracy and comprehensiveness when the amount of data reaches a certain scale. Thus, this paper proposes the application of multimodal thought to enhance emotion-recognition accuracy and conducts corresponding data preprocessing on the selected dataset. Appropriate models are constructed for both audio and video modalities: for the audio-modality emotion-recognition task, this paper adopts the “time-distributed CNNs + LSTMs” model construction scheme; for the video-modality emotion-recognition task, the “DeepID V3 + Xception architecture” model construction scheme is selected. Furthermore, each model construction scheme undergoes experimental verification and comparison with existing emotion-recognition algorithms. Finally, this paper attempts late fusion and proposes and implements a late-fusion method based on the idea of weight adaptation. The experimental results demonstrate the superiority of the multimodal fusion algorithm proposed in this paper. When compared to the single-modal emotion-recognition algorithm, the accuracy of recognition is increased by almost 4%, reaching 84.33%.

A novel and efficient immunoaffinity column (IAC) based on bispecific monoclonal antibody (BsMAb) recognizing aflatoxin B

In this study, forchlorfenuron (CPPU) was coupled with succinic anhydride to yield a CPPU hapten (CPPU-COOH), and a high-affinity monoclonal antibody (mAb) that can specifically recognize CPPU was produced. Using this mAb as a recognition reagent, a sensitive indirect competitive enzyme-linked immunosorbent assay (icELISA) for CPPU was optimized, which exhibits an IC50 of 1.04 ng/mL, a limit of detection of 0.16 ng/mL, and a linear range of 0.31–3.43 ng/mL for CPPU. Cross-reactivity percentages with six analogues were all below 6%. The average recovery rates for cucumber and orange samples were from 85.23% to 119.14%. The analysis results of this icELISA showed good consistency with those from liquid chromatography mass spectrometry. These results suggest that the proposed icELISA provides a sensitive, specific, and reliable strategy for CPPU detection in food samples.

The electronic and magnetic properties of Ag, Cu or Mg doped NiO are systematically studied in order to reveal the ferromagnetic half-metallic behavior by means of first-principle calculations. In the band structure and density of states (DOS), we find that impurity states crossing Fermi level are induced in the spin down channel, while band gap only appears in the Ni31XO32 (X = Ag, Cu) systems in the spin up channel. The analysis of electron spin density and magnetic show that the dopant atoms (Ag or Cu) and its neighbor O atoms exhibit strong polarization as well as the resultant magnetic moments. However, Mg doping system do not exhibit similar phenomenon. These results indicate that the nonmagnetic dopants of Ag or Cu can introduce the ferromagnetic half-metallic properties in the Ni31AgO32 or Ni31CuO32 system.

In order to accurately measure the three-dimensional transient magnetic field component generated by the discharge current of lightning struke high-speed railway catenary, this paper firstly establishes a simple calculation model of lightning struck high-speed railway catenary, and studies the temporal and spatial distribution characteristics of lightning stroke transient magnetic field in bridge section through calculation. And the design requirements and performance parameters of the measurement system are proposed according to the magnetic field characteristics. Then based on the Tunneling Magnetoresistive Effect (TMR), a three-dimensional measurement system of lightning transient magnetic field is designed and manufactured. In this study, three tunneling magnetoresistive sensors are used as the sensing probe, whose pairwise orthogonal design makes the measurement probe volume smaller and easier to measure. What's more, a signal processing circuit is designed to regulate the output signal, which is then collected through the signal collection module. Finally, the test scheme is designed to calibrate the time domain and verify the performance of the measuring system. The test results show that the measurement system is small in size, high in integration, fast in response to lightning transient magnetic field with the measurement dynamic range reaching $\pm\mathbf{5000A}/\mathbf{m}$ , which can basically meet the measurement requirements of transient magnetic field component in any direction generated by the discharge current of the lightning struck high-speed railway catenary.

Based on the Stillinger–Weber potential, molecular dynamics simulations of the solidification processes of multicrystalline silicon were carried out. Every stage of the whole solidification process, including homogeneous nucleation, grain growth, grain coarsening and defect characterization, was investigated. In the nucleation stage, it showed two typical nucleation models (spontaneous nucleation and sporadic nucleation) at high and low temperatures. Local heterogeneity was occasionally observed during homogeneous nucleation. The simulated nucleation rates at different temperatures were measured, whose trends were in agreement with the results of theoretical calculations, and both of them reached the maximum nucleation rate at a critical temperature of ∼0.65Tm. In the growth stage, the nuclei showed the maximum growth exponent at ∼0.65Tm. The evolution of the grain number at different temperatures exhibited three different patterns. Furthermore, the grains showed modest anisotropic growth before the growth was influenced by other grains. In the grain coarsening stage, the grain size distribution could be described suitably by the log-normal distribution. The grain coarsening exponent was approximately one order of magnitude lower than the nuclei growth exponent. The analyses of crystal defects showed that the dislocation density is about 105 m−2, twin percentages are above 40%, and the percentage of CSL grain boundaries is about 30.30% in mc-Si of the simulations. The statistics of crystal defects are similar to experimental results.

As-cast Cu-1.5Fe-0.5Co (wt%) alloy displays both high tensile strength of 307 MPa and elongation of 33%. In situ transmission electron microscopy was used to investigate crack propagation in the alloy, to analyze the origin of the good properties. At different deformation stages in thin Cu foils, the interactions of a propagating crack with iron-rich nanoparticles and growth twins are investigated. Crack-bridging processes via near-tip twinned bridges were identified. The multiple deformation mechanisms act synergistically to contribute to high strength and high ductility in the alloy.

Magnesium-based bulk metallic glass matrix composites (BMGMCs) have better plasticity than the corresponding bulk metallic glasses (BMGs); however, their strength and density are often compromised due to the fact that the effective reinforcement phase is mostly plastic heavy metal. For lightweight SiC-particle reinforced BMGMCs, interface wettability and the sharpness of the particles often reduce the strengthening effect. In this work, SiC particles were coated with a thin Cu coating by electroless plating, and added to Mg54Cu26.5Ag8.5Gd11 melt in an amount of 5 wt % to prepare a BMGMC. The microstructure of the interface, mechanical behavior and fracture morphology of the BMGMC were studied by scanning electron microscopy and quasi-static compression testing. The results showed that the Cu coating improved the wettability between SiC and the matrix alloy without obvious interfacial reactions, leading to the dispersion of SiC particles in the matrix. The addition of Cu-coated SiC particles improved the plastic deformation ability of Mg54Cu26.5Ag8.5Gd11 BMG, proving that electroless plating was an effective method for controlling the interface microstructure and mechanical behavior of BMGMCs.

The pigments on the painted pottery figurines from two tombs of Northern Wei Dynasty (AD 386-534) in Luoyang were analyzed by Raman microscopy. All the pigments were identified compared with the Raman spectra of standard pigments. The red pigments were identified as haematite, the blue pigment as lapis lazuli, the green pigment as malachite, the black pigment as carbon black and the white pigment as calcite. Similar pigments were used in the two tombs despite the pottery figurines were very different in artistic style. The use of lapis lazuli as blue pigment on Chinese painted pottery figurines was found for the first time. This pigment and the painted pottery figurine of Sogdians are of great archaeological significance because it demonstrated that the trade and cultural exchanges via the Silk Road had extended to Luoyang city in the Northern Wei Dynasty. The result also confirms that micro-Raman spectroscopy is a powerful analytical method for the identification of pigments on ancient artworks.

The charging behavior of PS (polystyrene) particles dispersed in nonpolar solvent containing surfactant AOT (sodium di-2-ethylhexylsulfosuccinate) was researched by phase angle light scattering (PALS). The effects of the AOT concentration, the particle concentration and the particle size on the zeta potential of the particles were analyzed systemically. The results showed, at different particle concentrations (expressed in the volume fraction of 10 − 5–10 − 3), that the zeta potential could be adjusted by surfactant AOT over a wide concentration range of 0.001–100 mM. An obvious difference of zeta potential on particle concentration was revealed between the high AOT concentrations (beyond 10 mM) and the low ones (below 1 mM). Meantime, it was found that the relationship of particle size to zeta potential showed a great discrepancy between the dilute particle concentrations (below 10 − 4) and the concentrated ones (beyond 6 × 10 − 4). These findings were consistent with the mechanism of preferential adsorption of the charged micelles in nonpolar solvent.