结合深度学习与支持向量机的金属零件识别

目的 在视觉引导的工业机器人自动拾取研究中，关键技术难点之一是机器人抓取目标区域的识别问题。特别是金属零件，其表面的反光、随意摆放时相互遮挡等非结构化因素都给抓取区域的识别带来巨大的挑战。因此，本文提出一种结合深度学习和支持向量机的抓取区域识别方法。方法 分别提取抓取区域的方向梯度直方图（HOG）和局部二进制模式（LBP）特征，利用主成分分析法（PCA）对融合后的特征进行降维，以此来训练支持向量机（SVM）分类器。通过训练Mask R-CNN（regions with convolutional neural network）神经网络完成抓取区域的初步分割。然后利用SVM对Mask R-CNN识别的抓取区域进行二次分类，完成对干扰区域的剔除。最后计算掩码完成实例分割，以此达到对抓取区域的精确识别。结果 对于随机摆放的铜质金属零件，本文算法与单一的Mask R-CNN及多特征融合的SVM算法就识别准确率、错检率、漏检率3个指标进行了比较，结果表明本文算法在识别准确率上较Mask R-CNN和SVM算法分别提高了7%和25%，同时有效降低了错检率与漏检率。结论 本文算法结合了Mask R-CNN与SVM两种方法，对于反光和遮挡情况具有一定的鲁棒性，同时有效地提升了目标识别的准确率。     

Reaction synthesis of spark plasma sintered MoSi2-B4C coatings for oxidation protection of Nb alloy

MoSi₂-B₄C coatings with different B₄C contents were prepared on Nb alloy by spark plasma sintering (SPS) process. Powder mixtures of Mo, Si and B₄C were used as the coating starting materials. Besides MoSi₂ and B₄C phases, small amounts of SiC and MoB are also found in the coatings because of the reactions of Mo, Si and B₄C powders during sintering. Compared with single MoSi₂ coating, the MoSi₂-B₄C coatings show better oxidation resistance at 1450?℃, and dense B₂O₃-SiO₂ oxide scales form after 100?h oxidation. The B₄C or MoB in the MoSi₂-B₄C coatings can serve as the B donor for the formation of B₂O₃. A slight degradation in the microstructure of the MoSi₂-B₄C coatings after oxidation is observed, which can be attributed to the presence of an NbB layer in the inter-diffusion zone of the coatings that retards the inward diffusion of Si from the coating into the substrate alloy. The microstructure development and oxidation behavior of the MoSi₂-B₄C coatings have been discussed.     

A Novel Strategy for Preparing Stretchable and Reliable Biphasic Liquid Metal

Low‐melting liquid metal is a hugely promising material for flexible conductive patterns due to its excellent conductivity and supercompliance, especially low‐cost and environmental liquid processing technology. However, the ever‐present fluidity characteristic greatly limits the stable shape and reliability of prepared liquid metal conductive electronics. Herein, a novel solidification strategy of liquid GaIn alloys by Ni doping and heat treatment is first reported, which can efficiently create a solid phase in the liquid metal and provide an effective solution for practical applications. Particularly, the liquid characteristic is preserved for conveniently fabricating different flexible electronic circuits, and then the solidification is carried out on prepared conductive patterns by heat treatment. The solidification mechanism is revealed by the interface chemical reaction between Ni and GaIn, creating the solid phase of intermetallic compound (Ga₄Ni₃ and InNi₃) during heat treatment. Moreover, a biphasic GaInNi can be obtained by regulating the atomic ratio of gallium, indium, and nickel. As a result, the obtained GaInNi possesses extremely low sheet resistance (15 ± 4.5 to 135 ± 2.5 mΩ sq^?1) and the variation of ΔR/R₀ exhibits low level (0–2) when strained up to 100%, which offers a promising strategy to prepare stretchable and reliable liquid metal electronics.     

15CrMoG钢棒材表面缺陷机理分析和工艺改进

分析得出,棒材表面细小纵裂纹和表面裂口缺陷产生于铸坯加热之前,且与结晶器弯月面保护渣有关。利用Thermo-Calc热力学软件计算15CrMoG钢凝固相变过程,结合亚包晶钢连铸凝固特点综合分析15CrMoG钢棒材表面缺陷的产生原因和产生机理。结果表明:15CrMoG钢在固相线温度附近发生包晶反应L+δ→γ和包晶转变δ→γ,不仅导致初生坯壳生长不均匀,而且加剧P、S元素在凝固前沿的偏析。而初生坯壳不均匀是导致棒材表面缺陷根本原因。棒材表面细小纵裂纹产生于结晶器内坯壳薄弱处,经过二冷和轧制工序在夹杂物和硫偏聚处扩展长大。棒材表面裂口缺陷是初生坯壳不均匀导致结晶器内液面波动大,造成铸坯夹渣所致。通过控制[C]0.16%～0.17%、[S]≤0.005%、保护渣碱度1.2、熔点≥1200℃、粘度≥1.0Pa·s,260 mm×30mm铸坯水量150 m³/h,拉速0.5 m/min等措施,裂纹合格探伤合格率由原45%提高至98%。     

In situ characterization of LiY(WO4)2 nanotubes for electrochemical energy storage

Here, LiY(WO₄)₂ nanotubes are prepared via a feasible electrospinning technique. This new anode material shows excellent electrochemical properties. The capacity loss of LiY(WO₄)₂ nanotubes is as low as 6.9% after 156 cycles, while bulk LiY(WO₄)₂ presents the capacity loss higher than 55.0%. Even after 600 long-life cycles, the capacity loss of the nanotubes is only 9%. It can be seen that the hollow structure with a rough surface and a porous morphology contributes to the improvement of electrochemical performance. Furthermore, online X-ray diffraction (XRD) method is firstly applied to understand the lithium ions insertion/extraction mechanism of LiY(WO₄)₂ nanotubes. It can be concluded that it is an asymmetrical two-phase reaction. A phase transformation from LiY(WO₄)₂ to Li₃Y(WO₄)₂ can be obviously seen from the in situ XRD during discharge process. While Li₂Y(WO₄)₂ appears as an intermediate phase with a reverse charge reaction. In addition, in situ XRD also demonstrates that LiY(WO₄)₂ nanotubes have surprised electrochemical reversibility. All the above results indicate that LiY(WO₄)₂ nanotubes can be expected to be anode candidate for rechargeable lithium ion batteries (LIBs).     

Optimal and Elastic Energy Trading for Green Microgrids: a two-Layer Game Approach

Mobile Networks and Applications - Energy trading mechanism for microgrids has an inherent two-layer architecture, in which the energy trading at the first layer is between a microgrid aggregator...     

Low complexity carrier phase estimation for m-QAM optical communication systems

Photonic Network Communications - A low complexity carrier phase estimation (CPE) algorithm for M-ary quadrature amplitude modulation (m-QAM) optical communication systems is investigated in this...     

非调质钢50MnSiV新能源汽车电机轴的开发

非调质钢50MnSiV(/%:0. 50C,0.52S,1.20Mn,0.010P,0.025S,0.15Cr,0.10V,0.015Ti,0.015N)制造的新能源汽车电机轴,省去传统调质钢20CrMnTiH渗碳热处理和矫正精整工序,不仅可提高材料95%利用率和2天交货周期,还可以降低25%成本。50MnSiV钢抗拉强度1100 MPa、屈服强度858 MPa、冲击功48J。其静扭扭矩(4697 Nm)、疲劳寿命(±1600 Nm双向扭转12万次、±1400 Nm双向扭转27万次)较渗碳20CrMnTiH钢提高了27%和93%～140%,并且其性能指标和8万公里的路试结果均满足电机轴的技术要求。