Highly Flexible Graphene Oxide Nanosuspension Liquid‐Based Microfluidic Tactile Sensor

A novel graphene oxide (GO) nanosuspension liquid‐based microfluidic tactile sensor is developed. It comprises a UV ozone‐bonded Ecoflex–polydimethylsiloxane microfluidic assembly filled with GO nanosuspension, which serves as the working fluid of the tactile sensor. This device is highly flexible and able to withstand numerous modes of deformation as well as distinguish various user‐applied mechanical forces it is subjected to, including pressing, stretching, and bending. This tactile sensor is also highly deformable and wearable, and capable of recognizing and differentiating distinct hand muscle‐induced motions, such as finger flexing and fist clenching. Moreover, subtle differences in the handgrip strength derived from the first clenching gesture can be identified based on the electrical response of our device. This work highlights the potential application of the GO nanosuspension liquid‐based flexible microfluidic tactile sensing platform as a wearable diagnostic and prognostic device for real‐time health monitoring. Also importantly, this work can further facilitate the exploration and potential realization of a functional liquid‐state device technology with superior mechanical flexibility and conformability.     

基于块稀疏贝叶斯学习的体域网心电压缩采样

为有效提高体域网的实时性和降低体域网的功耗,提出一种基于块稀疏贝叶斯学习的体域网心电压缩采样方法。该方法在体域网框架下,利用压缩采样理论,在体域网的传感节点利用二进制随机观测矩阵对心电信号进行压缩采样,远程监护中心获得采样值之后,利用块稀疏贝叶斯学习重构算法和离散余弦稀疏变换矩阵对心电信号进行重构。实验结果表明,当心电信号压缩率在70%~90%时,基于块稀疏贝叶斯学习的重构算法要比其他重构算法的重构信噪比高出3 dB~21 dB。该方法能有效减少数据采样,减轻后续的数据存储、数据传输压力,提高体域网的实时性。同时该方法具有功耗低,易于硬件实现的优点。     

时间序列到网络新方法应用于心电信号分析卢青,陈晓光

提出了一种新的时间序列转换成网络的方法,从而可以通过网络拓扑统特征来刻画时间序列。将该方法应用于心室纤颤的检测中,发现健康个体与室颤病人相应网络非常明显的不同,将网络的聚类系数与同配系数作为联合指标可以区分两者。此外,还将该算法应用于HRV分析,结果表明,算法能够区分不同病态下的心跳间隔信号。     

面向数值模拟数据的HDF5性能优化

大规模数值模拟数据对可视化分析提出了挑战,I/O是影响可视化交互性能的重要因素.HDF5是科学计算领域广泛采用的存储格式,介绍了HDF5的抽象数据模型、数据读写流程,并使用典型数值模拟数据测试了HDF5的读性能.测试发现HDF5的数据集定位开销较大.根据数值模拟数据的数据块以整数有规律编号的特点,通过在HDF5中增加数据块视图对象来提高读性能.测试表明,该方法可显著加速数据的读取性能.     

数千上万核上模拟结果的并行可视化

在科学与工程计算中,在数千上万核上,模拟包含上亿网格单元的模型通常输出TB量级的时变数据集。这些数据集包含多个时刻的数据,每个时刻的数据分布存储在数千个文件,单时刻的数据量达到几GB甚至几十GB。为了并行地可视化这种时变数据集,本文设计了基于网格片的层次化数据结构,基于该数据结构改进和优化了并行可视化流程和数据通信算法。在集成到可视化软件后,实现了两个应用的模拟结果可视化。实测数据表明,对于单时刻5GB和32GB的数据,在数十上百个处理器核上,从数据读入到获得体绘制结果的时间分别为19秒和80秒,交互操作为2秒到10秒;三维面绘制在数秒钟内,切片分析在2秒以内。     

Degradable and Fully Recyclable Dynamic Thermoset Elastomer for 3D-Printed Wearable Electronics

Wearable electronics have become an important part of daily lives. However, its rapid development results in the problem of electronic waste (e-waste). Consequently, recyclable materials suitable for wearable electronics are highly sought after. In this study, a conductive recyclable composite (PFBC) is designed based on a dynamic covalently cross-linked elastomer and hierarchical hybrid nanofillers. The PFBC shows excellent wide-ranging properties including processability, elasticity, conductivity, and stability, which are superior to previous materials used for recyclable electronics, and exhibits outstanding mechanical properties and environmental tolerance including high temperature, high humidity, brine, and ethanol owing to its covalent cross-linking. Reversible dissociation of Diels–Alder networks allows for convenient processing and recycling. After three recycles, the toughness of the PFBC remained at 10.1 MJ m⁻³, which is conspicuous among the reported recyclable electronic materials. Three types of PFBC-based wearable electronics including a triboelectric nanogenerator, a capacitive pressure sensor, and a flexible keyboard, are successfully 3D printed with excellent performance. The PFBC possessed both recyclability and degradability, the combination of which provides a new way to reduce e-waste. This is the first work to recycle electronics using direct 3D printing and presents promising new design principles and materials for wearable electronics.     

脯氨酸肽键数据集的构建

由分辨率<0.25nm,同一性(identity)<30％的2401条肽链中计算提取了全部顺式与反式脯氨酸肽键的位置,数目分别为1221个与26401个,从而建立了一个较大规模的脯氨酸肽键数据集。统计分析了该数据集的基本特征:肽键N端残基的分布、N端残基的二面角统计、在二级结构中的分布情况、顺式肽键在脯氨酸肽键中所占比例。此数据集对于进一步研究顺反X-Pro肽键的结构、与氨基酸序列之间的关系,以及肽链折叠动力学具有重要作用。     

基于反卷积和特征融合的SSD小目标检测算法

由于小目标的低分辨率和噪声等影响,大多数目标检测算法不能有效利用特征图中小目标的边缘信息和语义信息,导致其特征与背景难以区分,检测效果差。为解决SSD(single shot multibox detector)模型中小目标特征信息不足的缺陷,提出反卷积和特征融合的方法。先采用反卷积作用于浅层特征层,增大特征图分辨率,然后将SSD模型中卷积层conv11_2的特征图上采样,拼接得到新的特征层,最后将新的特征层与SSD模型中固有的4个尺度的特征层进行融合。通过将改进后的方法与VOC2007数据集和KITTI车辆检测数据集上的SSD和DSSD方法进行比较,结果表明:该方法降低了小目标的漏检率,并提升整体目标的平均检测准确率。     

基于USB接口的心电信号数据采集系统

介绍了基于USB接口的心电信号数据采集系统，给出了该系统的硬件组成原理及软件设计方法。由于该系统使用了高性能的USB专用芯片CY7C64613，因而具有使用方便、即插即用等特点。与笔记本电脑相连即可构成移动式心电信号检测分析仪器，具有较高的实用推广价值。