Application of artificial neural network method to predict the breakage properties of PGE bearing chromite ore

In the present investigation, systematic grinding experiments were conducted in a laboratory ball mill to determine the breakage properties of low-grade PGE bearing chromite ore. The population balance modeling technique was used to study the breakage parameters such as primary breakage distribution (B_{i, j}) and the specific rates of breakage (S_i). The breakage and selection function values were determined for six feed sizes. The results stated that the breakage follows the first-order grinding kinetics for all the feed sizes. It was observed that the coarser feed sizes exhibit higher selection function values than the finer feed size. Further, an artificial neural network was used to predict breakage characteristics of low-grade PGE bearing chromite ore. The predicted results obtained from the neural network modeling were close to the experimental results with a correlation of determination R² = 0.99 for both product size and selection function.     

High tunability in (1 1 0)-oriented BaZr0.2Ti0.8O3 (BTZ) lead-free thin films fabricated by pulsed laser deposition

Using a pulsed laser deposition method the BaZr_0.2Ti_0.8 O₃ (BTZ) lead–free thin films with a thickness of ~250 nm were grown on FTO, ITO and Pt–Si substrates, respectively. The analysis results of microstructural, dielectric properties and leakage current reveal that the thin films deposited on Pt–Si substrates are oriented growth along the (1 1 0) direction and exhibit the optimal performance characteristics. Calculations of figure of merit (FoM) and dielectric tunability display a maximum value of ~42.8 and ~68.5% at E = 400 kV/cm at room temperature, respectively. The excellent tunable properties, high dielectric constant (~635@ 100 kHz) and low leakage current density of (9.3 × 10^–8 A/cm² at 400 kV/cm) make the (1 1 0)–oriented BaZr_0.2Ti_0.8 O₃ thin film to be an attractive material for applications of tunable devices.     

基于深度学习的秀丽隐杆线虫显微图像分割方法

利用计算机实现自动、准确的秀丽隐杆线虫(C.elegans)的各项形态学参数分析,至关重要的是从显微图像上分割出线虫体态,但由于显微镜下的图像噪声较多,线虫边缘像素与周围环境相似,而且线虫的体态具有鞭毛和其他附着物需要分离,多方面因素导致设计一个鲁棒性的C.elegans分割算法仍然面临着挑战。针对这些问题,提出了一种基于深度学习的线虫分割方法,通过训练掩模区域卷积神经网络(Mask R-CNN)学习线虫形态特征实现自动分割。首先,通过改进多级特征池化将高级语义特征与低级边缘特征融合,结合大幅度软最大损失(LMSL)损失算法改进损失计算;然后,改进非极大值抑制;最后,引入全连接融合分支等方法对分割结果进行进一步优化。实验结果表明,相比原始的Mask R-CNN,该方法平均精确率(AP)提升了4.3个百分点,平均交并比(mIOU)提升了4个百分点。表明所提出的深度学习分割方法能够有效提高分割准确率,在显微图像中更加精确地分割出线虫体。     

Evaluation of hydrodynamic behavior of the perforated gas distributor of industrial gas phase polymerization reactor using CFD-PBM coupled model

A 2D computational fluid dynamics (Eulerian–Eulerian) multiphase flow model coupled with a population balance model (CFD-PBM) was implemented to investigate the fluidization structure in terms of entrance region in an industrial-scale gas phase fluidized bed reactor. The simulation results were compared with the industrial data, and good agreement was observed. Two cases including perforated distributor and complete sparger were applied to examine the flow structure through the bed. The parametric sensitivity analysis of time step, number of node, drag coefficient, and specularity coefficient was carried out. It was found that the results were more sensitive to the drag model. The results showed that the entrance configuration has significant effect on the flow structure. While the dead zones are created in both corners of the distributors, the perforated distributor generates more startup bubbles, heterogeneous flow field, and better gas–solid interaction above the entrance region due to jet formation.     

Modeling and simulation of mixing in water-in-oil emulsion flow through a valve-like element using a population balance model

Emulsion flows are very common in natural processes as well as in several engineering areas, such as in the process of desalting crude oil that occurs in refineries. This kind of flow is described as a polydispersed multiphase flow. In this work, we evaluated the behavior of water-in-oil emulsion flowing through a duct with an element used to mimic the effect of a globe valve. An Eulerian multi-fluid approach was employed by solving the population balance equation coupled with computational fluid dynamics. Coalescence and breakage models recently developed were extended to this inhomogeneous model. A bivariate population balance problem was also solved to demonstrate the mixing caused by the valve-like element. The simulated results showed good agreement with the available experimental data for the Sauter and DeBroukere mean diameters.     

An efficient approach for the siting and sizing problem of distributed generation

This paper deals with the subject of finding the site and size of distributed generation in distribution systems. This problem corresponds to a mixed-integer nonlinear problem which is difficult and hard to solve with classical optimization techniques. Many approaches and with different objective functions have been applied to solve it. In this paper, the problem to solve comprises multiple distributed generation sources and the objective of minimizing power losses and generation costs, both for the distributed and conventional generators. Since it is highly combinatorial, a search space reduction is needed. So, an approximate model is used to reduce the search space of the possible buses where the distributed generation has to be located. Then, for each combination of the reduced space search, a nonlinear equations system is solved by a numerical method to get the size of the distributed generation, checking the voltage limits and the lines’ capacities. To test the method, a comparison between the proposed algorithm and a force brute algorithm is performed on a 69 and 118 bus test systems. The obtained results indicate that this method find the optimal or near optimal solution in a reasonable computation time.     

How platinum oxide affects the degradation analysis of PEM fuel cell cathodes

In this work, proton exchange membrane fuel cell cathodes are degraded with accelerated-stress-tests.These PtCo containing cathodes are analyzed at begin-of-life and end-of-test with a dedicated diagnostic procedure. For every individual load point, the oxygen transport resistance and voltage losses due to the formation of platinum oxides were obtained in addition to commonly measured electrochemical surface area, high frequency resistance, as well as cathode ionomer resistance. These data were used to break down the voltage losses into six different contributors. With this break down, performance gains and performance losses were determined at end-of-test. At low current densities, it was found that voltage losses due to degradation are dominated by the loss of specific activity and catalyst surface area - in line with the state-of-the-art knowledge. But by quantifying the losses from platinum oxide formation explicitly, we show that end-of-test an unassigned voltage loss is not only present at highest current densities, but already at low current density. More precisely, the unassigned voltage loss shows a linear increase with decreasing half cell voltage and is independent from the chosen accelerated stress test. As this unassigned loss depends on half cell voltage, it might arise from ionomer adsorption.     

Multi-label,multi-task CNN approach for context-based emotion recognition

This paper proposes a new deep learning architecture for context-based multi-label multi-task emotion recognition. The architecture is built from three main modules: (1) a body features extraction module, which is a pre-trained Xception network, (2) a scene features extraction module, based on a modified VGG16 network, and (3) a fusion-decision module. Moreover, three categorical and three continuous loss functions are compared in order to point out the importance of the synergy between loss functions when it comes to multi-task learning. Then, we propose a new loss function, the multi-label focal loss (MFL), based on the focal loss to deal with imbalanced data. Experimental results on EMOTIC dataset show that MFL with the Huber loss gave better results than any other combination and outperformed the current state of art on the less frequent labels.     

RXDNFuse: A aggregated residual dense network for infrared and visible image fusion

This study proposes a novel unsupervised network for IR/VIS fusion task, termed as RXDNFuse, which is based on the aggregated residual dense network. In contrast to conventional fusion networks, RXDNFuse is designed as an end-to-end model that combines the structural advantages of ResNeXt and DenseNet. Hence, it overcomes the limitations of the manual and complicated design of activity-level measurement and fusion rules. Our method establishes the image fusion problem into the structure and intensity proportional maintenance problem of the IR/VIS images. Using comprehensive feature extraction and combination, RXDNFuse automatically estimates the information preservation degrees of corresponding source images, and extracts hierarchical features to achieve effective fusion. Moreover, we design two loss function strategies to optimize the similarity constraint and the network parameter training, thus further improving the quality of detailed information. We also generalize RXDNFuse to fuse images with different resolutions and RGB scale images. Extensive qualitative and quantitative evaluations reveal that our results can effectively preserve the abundant textural details and the highlighted thermal radiation information. In particular, our results form a comprehensive representation of scene information, which is more in line with the human visual perception system.     

云南铜厂沟铜钼矿低贫损分段崩落与浅孔留矿耦合采矿法探索

铜厂沟矿区多为破碎、中厚、急倾斜矽卡岩型铜钼矿体，原设计采用上向水平分层充填采矿法，在采矿实验中存在作业安全性差、劳动生产效率低、损失贫化大、采矿成本高等缺陷，严重制约了试生产组织。矿山急迫需要一种安全、高效、成本能够控制的采矿方法满足试生产矿石持续供应。通过充分研究，提出了低贫损分段崩落与浅孔留矿耦合采矿方法的设想，进一步降低了安全风险、简化工艺、控制损失贫化等。通过采矿方法的优化设计，解决了企业生产存在的实际问题，提高了采矿效率，对其它同类矿山具有一定的参考借鉴意义。