Highly stretchable,sensitive and wide linear responsive fabric-based strain sensors with a self-segregated carbon nanotube (CNT)/Polydimethylsiloxane (PDMS) coating

Flexible strain sensors have received increasing attention with the development of wearable electronic devices. However, integrating wide strain detection range, high sensitivity while maintaining relatively wide linear response range for such sensor system still remain a challenge. A fabric based flexible sensor (S-CNT/PDMS-F) was designed and fabricated, and the sensor can simultaneously achieve high sensitivity, wide linear response and strain detection range by combining self-segregated carbon nanotube (CNT)/polydimethylsiloxane (PDMS) composites and elastic medical bandage. It has been observed that this new sensor system can achieve a high sensitivity with a gauge factor of 615, a large linear responsive range of 0–100% strain (R² ?= ?0.993) and a wide strain detection range of ～ 200%, which is superior to almost all the reported CNT/PDMS flexible strain sensors. Compared to the similar fabric based strain sensor system deploying non self-segregated structure (C-CNT/PDMS-F), our S-CNT/PDMS-F demonstrates higher electrical conductivity and lower electrical percolation threshold and response time of 55 ?ms, as well as more stable and repeatable performance under cyclic loading conditions. The capability of the sensors in monitoring physiological activities and weight distribution has also been demonstrated.     

电解液浓度和沉积时间对Zn涂层表面形貌影响的研究

采用阳极氧化法在纯钛基体表面制备一层TiO₂纳米管（TiO₂ nanotubes，TNs），随后采用电化学沉积法在ZnCl₂溶液中于TNs上制备了Zn涂层。采用扫描电子显微镜（scanning electron microscopy，SEM）、能量色散X射线光谱仪（energy dispersive X-ray spectrometer，EDS）和X射线衍射仪（X-ray diffraction，XRD）对涂层的表面和截面形貌、元素含量和相组成进行表征。结果表明：TNs长度为1.5 μm，Zn涂层厚度为2 μm；在沉积的过程中，Zn涂层首先沿着TNs的管壁生长，逐渐覆盖TNs，然后形成球状晶体，最后形成白色颗粒堆叠在TNs上；电解液浓度的增加提高了涂层生长速度，沉积时间的延长使晶体有充足的时间长大；涂层中Zn是以单质形式存在，白色颗粒也主要为Zn单质。     

固溶和时效处理对碳纳米管/2024铝合金复合材料组织和性能的影响

对碳纳米管/2024铝合金复合材料进行固溶和时效处理，通过维氏硬度和室温拉伸实验测试了复合材料的性能，对固溶和时效处理后复合材料的微观组织和析出相进行了表征。研究发现，复合材料的强塑性与固溶和时效处理密切相关，碳纳米管/2024铝合金复合材料经530 ℃×4 h固溶处理后维氏硬度达到最高，为179.45，较原始复合材料维氏硬度提高约31%。时效处理后加速了时效硬化行为，经130 ℃×4 h时效处理后复合材料的抗拉强度、屈服强度和伸长率最高，分别为430.4 MPa，606.1 MPa和9.5%。结果表明，碳纳米管/2024铝合金复合材料适宜的固溶和时效制度为：固溶处理530 ℃×4 h，时效处理130 ℃×4 h。     

一种量化因子自适应学习量化训练算法

深度神经网络因参数量过多而影响嵌入式部署，解决的办法之一是模型小型化（如模型量化，知识蒸馏等）。针对这一问题，提出了一种基于BN（batch normg lization）折叠的量化因子自适应学习的量化训练算法（简称为LSQ-BN算法）。采用单个CNN（convolutional neural）构造BN折叠以实现BN与CNN融合；在量化训练过程中，将量化因子设置成模型参数；提出了一种自适应量化因子初始化方案以解决量化因子难以初始化的问题。实验结果表明：8bit的权重和激活量化，量化模型的精度与FP32预制模型几乎一致；4bit的权重量化和8 bit的激活量化，量化模型的精度损失在3%以内。因此，LSQ-BN是一种优异的模型量化算法。     

考虑概率型共因失效的多阶段任务系统可靠性分析模型

针对多阶段任务系统(phased-mission system, PMS)任务可靠性受概率型共因失效(probabilistic common cause failure, PCCF)影响的问题, 提出一种基于贝叶斯网络(Bayesian network, BN)的PCCF-PMS分析模型。首先, 研究基于BN的PMS表征方法, 建立不考虑共因失效的PMS基础BN模型, 即PMS-BN。其次, 构建共因空间节点, 并研究在共因空间节点影响下系统模型参数的修正方法。最后, 引入共因节点对PMS-BN模型进行扩展, 实现考虑共因失效影响的PMS可靠性量化分析。以地球同步轨道卫星的首次变轨任务为例说明所提模型的正确性, 分析结果表明, 共因失效问题对于PMS的可靠性存在显著影响。PCCF-PMS模型能够综合处理受概率型与确定型共因失效影响的PMS可靠性分析问题。所提模型适用于共因事件间呈独立、互斥、统计相关等统计关系的情况, 且网络模型规模可控。