基于并行多通道卷积长短时记忆网络的轴承寿命预测方法

在预测轴承剩余使用寿命时，数据间的时序特性是一个可以利用的重要隐藏信息。为了更好地提取具有时序信息的特征用于预测，提出了一种基于并行多通道卷积长短时记忆网络（PMCCNN-LSTM）的剩余使用寿命预测模型。该模型主要由两部分组成：前端为并行多通道卷积网络（PMCCNN），提取信号特征，挖掘数据的时序特性，并采用逐层训练和微调的方式提升参数的收敛性；后端为长短时记忆（LSTM）网络，基于特征进行剩余使用寿命预测，并采用加权平均的方法对预测结果进行平滑处理。在一个轴承加速寿命实验的公开数据集上使用留一法验证了该模型的准确性，实验结果表明：所提模型的平均误差与最大误差分别比传统的卷积神经网络（CNN）低23.38%和15.84%，比传统的LSTM低24.14%和19.01%，比卷积长短时记忆网络（CNN-LSTM）低30.32%和23.09%。     

LNG储罐用06Ni9钢板生产工艺及控制措施

介绍了LNG储罐用06Ni9钢板的技术要求。针对生产中06Ni9钢板性能控制、表面质量控制及超极限规格轧制板形控制的难点问题，经试验及现场实践，提出了相应的控制措施，如优化化学成分设计、优化轧制工艺和热处理工艺，保证了钢板的性能；板坯表面喷涂专用防氧化涂料，并配合合理的除鳞工艺，保证了钢板的表面质量；优化坯料、辊型设计，改进模型参数，保证了极限规格钢板的板形质量。在此基础上，太钢成功开发出LNG工程用06Ni9钢板。     

A triple-shape memory material fabricated by in situ crosslinking of poly(butylene adipate-co-terephthalate)/ε-polycaprolactone via electron-beam irradiation and its triple-shape memory effects

In this work, we fabricated a triple-shape memory polymeric material though in situ crosslinking of poly(butylene adipate-co-terephthalate)/ε-polycaprolactone blends induced by electron beam irradiation. The intrinsic property of irradiation crosslinking, crosslinking rule in this heterogeneous system, formation of interfacial bridge molecular, and the resulted interfacial strength were investigated. No obvious interfacial slips were observed in tensile test, and the elongation at break was above 900%. The high-temperature relaxation and recovery experiments at nonlinear region confirmed the excellent hyperelasticity and recovery ratios (R_r ) can be as high as 95% for samples irradiated above 96 kGy. The mismatch between heating rate and relaxation kinetics resulted in both instable temporary shapes and heating-mode dependent recovery behavior during the recovery process. Self-stability of the temporary shape at middle temperature was also emphasized in view of the phenomenon that the second R_r was above 100% in our experiments. © 2019 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2020 , 137, 48588.     

Stretchable poly(glycerol-sebacate)/β-tricalcium phosphate composites with shape recovery feature by extrusion

There has been a great interest in research towards elastomers and their composites with an attempt to obtain the desired biological and mechanical response to scaffold materials in bone tissue engineering. Composites made of ceramic-thermoplastic mixtures have been shown success to deliver the inorganic component while fail to provide replacement of an elastic protein, that is, collagen, of the target bone tissue. Thus, in order to match up with the inherent elasticity of the native tissue, it is proposed an alternative to well-known thermoplastic-containing matrices by using a poly(glycerol-sebacate) (PGS)–beta-tricalcium phosphate elastomeric composite to offer flexibility and mechanical integrity. This study reports for the first time a successful extrusion of PGS containing biodegradable composites with shape-memory feature. The resulting structures are physically and chemically characterized. In vitro cell culture performance of the obtained materials is investigated by using an MC3T3-E1 mouse preosteoblast cell line. The materials obtained in this study can be shaped into the desired size and various forms via temperature stimuli. Resulting materials have been proposed for craniofacial tissue engineering as a bone filler in which surgeons need to shape biomaterials during the surgical procedure due to the complex geometry of the bones. © 2019 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2019 , 137, 48689.     

异形医用双腔导管挤出离模膨胀变形机理与调控

如何准确预测和精密调控异形医用双腔导管的离模膨胀变形是实现其精密成型的关键技术。通过传统和气辅精密成型的对比分析，研究表明离模膨胀变形是由熔体的第二法向应力差驱动的径向二次流动所诱发，传统挤出成型会产生较大的第二法向应力差，第二法向应力差驱动诱发的径向二次流动是离模膨胀变形的直接驱动力，从而导致传统挤出成型的异形医用双腔导管不仅产生离模膨胀，而且还产生椭圆度误差，其最大离模膨胀比和椭圆度误差分别为1.86和6.3 %。异形医用双腔导管的气辅精密挤出成型基本可以消除熔体的第二法向应力差，必然消除了挤出成型过程的径向二次流动，从而实现了异形医用双腔导管的精密控形。     

Ferroelectric Tunnel Junctions: Modulations on the Potential Barrier

Recently, ferroelectric tunnel junctions (FTJs) have attracted considerable attention for potential applications in next-generation memories, owing to attractive advantages such as high-density of data storage, nondestructive readout, fast write/read access, and low energy consumption. Herein, recent progress regarding FTJ devices is reviewed with an emphasis on the modulation of the potential barrier. Electronic and ionic approaches that modulate the ferroelectric barriers themselves and/or induce extra barriers in electrodes or at ferroelectric/electrode interfaces are discussed with the enhancement of memory performance. Emerging physics, such as nanoscale ferroelectricity, resonant tunneling, and interfacial metallization, and the applications of FTJs in nonvolatile data storage, neuromorphic synapse emulation, and electromagnetic multistate memory are summarized. Finally, challenges and perspectives of FTJ devices are underlined.     

1.5 mm厚热轧高强耐候SPA-H带钢的开发

薄规格带钢生产中面临温降大、板形控制难、穿带速度快、轧制稳定性差等难题，通过对SPA-H带钢化学成分、连铸工艺、温度和板形控制的分析研究，确定了合理的工艺参数，开发出1.5 mm极限规格高强耐候SPA-H带钢。结合连铸拉速和结晶器液面控制，过程温度控制和设备精度控制，实现了1.5 mm热轧集装箱用SPA-H带钢的稳定轧制和批量供应，其板形和尺寸精度高，力学性能稳定，满足了用户的要求。     

采用图像分析技术对球形Ti-6Al-4V粉末粒形的定量分析

采用图像分析技术对4种不同方法制备的球形Ti-6Al-4V粉末进行粒形的定量分析,分别测量了粉末的球形度、椭圆率、赘生物指数及粗糙度。结果表明:4种方法制备的粉末平均球形度均在90%以上。等离子旋转电极雾化法(PREP)、等离子火炬雾化法(PA)、等离子惰性气体雾化法(PIGA)、电极感应熔炼气雾化法(EIGA)制得粉末球形度依次降低,粗糙度依次增加。PREP、PA、EIGA、PIGA法制得粉末的表面卫星球粘附依次增加。对于PREP法制得粉末,粉末粒径范围越细,球形度越高,平均粗糙度越小。粉末粒形指标的差异与其制备方法的原理有关。采用图像分析技术可以实现对金属粉末粒形指标的科学定量分析。     

Comparison of Cu–Al–Ni–Mn–Zr shape memory alloy prepared by selective laser melting and conventional powder metallurgy

This work aimed to investigate and critically analyze the differences in microstructural features and thermal stability of Cu−11.3Al−3.2Ni−3.0Mn−0.5Zr shape memory alloy processed by selective laser melting (SLM) and conventional powder metallurgy. PM specimens were produced by sintering 106−180 µm pre-alloyed powders under an argon atmosphere at 1060 °C without secondary operations. SLM specimens were consolidated through melting 32−106 µm pre-alloyed powders on a Cu−10Sn substrate. Mechanical properties were measured through Vickers hardness testing. Differential scanning calorimetry was conducted to assess the martensitic transformation temperatures. X-ray diffraction patterns were collected to identify the metallurgical phases. Optical and scanning electron microscopy was used to analyze the microstructural features. β′₁ martensite was found, irrespective of the processing route, although coarser martensitic variants were present in PM-specimens. In conventional powder metallurgy samples, intergranular eutectoid constituents and stabilized austenite also formed at room temperature. PM-specimens showed similar average hardness values to the SLM-specimens, albeit with high standard deviation linked to the porosity. The specimens processed by SLM showed reversible martensitic transformation (T₀=171 °C). PM-processed specimens did not show shape memory effects.     

Design,fabrication of honeycomb-shaped 1–3 connectivity piezoelectric micropillar arrays for 2D ultrasound transducer application

As a core component of 2D ultrasound transducers, honeycomb-shaped 1–3 connectivity piezoelectric micropillar arrays have attracted enormous attention due to their unique performance and functionality. In this paper, honeycomb-shaped 1–3 connectivity piezoelectric micropillar arrays with a high aspect-ratio were designed and fabricated by means of deep X-ray lithography and powder injection molding in six steps: preparation of lost mold, powder-binder mixing, injection molding and demolding, removal of binders, and densification of powder. A polymer-based lost mold insert was generated by a synchrotron X-ray exposure and development process. The optimal volumetric ratio between the piezoelectric powder and binders was determined by torque rheology behavior, then they were homogeneously mixed with a twin extruder mixer. To fully fill in the micro-cavities of the lost mold, rheological properties of the mixture were analyzed with a capillary rheometer using different shear rates (50–5000 s⁻¹) and temperatures (140 °C, 150 °C, and 160 °C). After the mixture was completely injected, the lost mold was chemically dissolved in acetone and rinsed in methanol without bending or clustering of the micropillar arrays during evaporation. The binders in the injection molded portion were thermally decomposed using a continuous heating schedule of 200 °C, 390 °C, and 600 °C in argon gas under atmospheric conditions. Finally, the particles in the sample were densified into a coherent, solid mass by eliminating pores at 1300 °C. Based on the proposed micro-manufacturing process, defect-free honeycomb-shaped 1–3 connectivity piezoelectric micropillar arrays with a pattern dimension of 42 μm and aspect-ratio of 5 were successfully produced.