Self-supervised learning-based underwater acoustical signal classification via mask modeling.

The classification of underwater acoustic signals has garnered a great deal of attention in recent years due to its potential applications in military and civilian contexts. While deep neural networks have emerged as the preferred method for this task, the representation of the signals plays a crucial role in determining the performance of the classification. However, the representation of underwater acoustic signals remains an under-explored area. In addition, the annotation of large-scale datasets for the training of deep networks is a challenging and expensive task. To tackle these challenges, we propose a novel self-supervised representation learning method for underwater acoustic signal classification. Our approach consists of two stages: a pretext learning stage using unlabeled data and a downstream fine-tuning stage using a small amount of labeled data. The pretext learning stage involves randomly masking the log Mel spectrogram and reconstructing the masked part using the Swin Transformer architecture. This allows us to learn a general representation of the acoustic signal. Our method achieves a classification accuracy of 80.22% on the DeepShip dataset, outperforming or matching previous competitive methods. Furthermore, our classification method demonstrates good performance in low signal-to-noise ratio or few-shot settings.

Role and Effect of Meso-Structuring Surfactants on Properties and Formation Mechanism of Microfluidic-Enabled Mesoporous Silica Microspheres.

We have shown in a previous work that the combination of the emulsion solvent evaporation technique and droplet-based microfluidics allows for the synthesis of well-defined monodisperse mesoporous silica microcapsules (hollow microspheres), whose size, shape and composition may be finely and easily controlled. In this study, we focus on the crucial role played by the popular Pluronic® P123 surfactant, used for controlling the mesoporosity of synthesised silica microparticles. We show in particular, that although both types of initial precursor droplets, prepared with and without P123 meso-structuring agent, namely P123+ and P123- droplets, have a similar diameter (≃30 µm) and a similar TEOS silica precursor concentration (0.34 M), the resulting microparticles exhibit two noticeably different sizes and mass densities. Namely, 10 µm and 0.55 g/cm3 for P123+ microparticles, and 5.2 µm and 1.4 g/cm3 for P123- microparticles. To explain such differences, we used optical and scanning electron microscopies, small-angle X-ray diffraction and BET measurements to analyse structural properties of both types of microparticles and show that in the absence of Pluronic molecules, P123- microdroplets divide during their condensation process, on average, into three smaller droplets before condensing into silica solid microspheres with a smaller size and a higher mass density than those obtained in the presence of P123 surfactant molecules. Based on these results and on condensation kinetics analysis, we also propose an original mechanism for the formation of silica microspheres in the presence and in the absence of the meso-structuring and pore-forming P123 molecules.

Source apportionment and potential ecological risk assessment of heavy metals in soils on a large scale in China.

The properties and sources of soil heavy metals (Pb, Zn, Cu, Cd, As, Hg, Cr, and Ni) need to be comprehensively analyzed to take effective steps to control and reduce soil pollutants. In this research, 416 soil samples were collected on a large scale in China. Two receptor models (PCA/MLR and PMF) were utilized to identify pollutant sources and quantify the contributions. The means of soil heavy metals (Zn, Cu, As, Hg, Cr, and Ni) were lower than the corresponding screening values and intervention values. Cd was greater than the intervention value, while Pb was between the screening value and the intervention value. Source apportionments suggested that mine sources were the most polluted (64.28%), followed by traffic sources (38.98%), natural sources (11.41-39.58%), industrial sources (9.8-18.65%), and agricultural sources (2.79-14.51%). Compared to the PCA/MLR model, the PMF model had a better effect in evaluating soil heavy metal pollution. It gave corresponding weights according to the data concentration and its uncertainty, which made the result reasonable. The ecological risk assessment indicated that Cd posed a significant risk, while Hg caused a mild risk and the other six heavy metals posed a low risk. The spatial distribution of ecological risk suggested that severe risk points were mainly distributed in the central area, while high-risk points were distributed in the southern region. The SRI method was developed to link pollution sources and their potential ecological risks and indicated better applicability to the PMF model. The study findings could provide guidelines for monitoring the main sources and reducing the pollution of soil heavy metals.

[Influence of laser energy on measurement of unburned carbon in fly Ash particle flow].

The fly ash particle flow was produced by a screw feeder and then ablated by a pulse laser to create plasma. The emission spectra of fly ash were detected by laser-induced breakdown spectroscopy. The present paper focused on the influence of laser energy on the measurement of unburned carbon. Seven groups of pulse laser in the range of 40 to 130 mJ were used to ablate the fly ash particle flow. The results show that the carbon line intensity is increased linearly with the increases in laser energy, but the SNR of carbon line increases in the range of 40 to 90 mJ and then trends to saturation, while the elimination rate of false data decreases. In this experiment, laser energy ranging from 90 to 100 mJ can enhance the plasma emission signal and improve the true spectral data of fly ash particle flow. So laser energy has close correlations with the ablation of the particle flow and the carbon line intensity. Reasonable laser energy is good for the effective ablation of the fly ash particle flow to get plasma spectra signals with good SNR.