结构磁共振影像 (sMRI)作为一种非入侵式的脑成像方式为人们理解阿尔茨海默病 (AD)的患病机制提供了很大的帮助, 目前已有大量研究利用从 sMRI中提取的特征进行 AD的识别。为了充分利用图像信息提取 AD相关的特征, 提出了一种简单易用的基于人类脑网络组图谱的脑区划分进行特征提取的方法。选取美国阿尔茨海默病神经影像组织 (ADNI)数据库中的 226例正常被试 (NC)和227例 AD患者的sMRI数据作为研究对象, 提取每个个体脑网络组图谱中各脑区的平均灰质密度作为特征, 利用支持向量机 (SVM)对NC和AD患者进行分类, 通过 10折交叉验证的方式得到了 85.2%的平均分类准确率。后续的统计分析发现, 海马、杏仁核及梭状回等脑区的平均灰质密度对 NC和AD患者的识别贡献很大, 且这些脑区的萎缩程度与患者的认知能力密切相关。利用最小绝对收缩选择算子 (LASSO)对个体的简易智力状态评分 (MMSE)进行预测, 预测结果与真实值间存在显著的正相关 (r0.65, p0.001)。研究结果表明, 基于脑网络组图谱脑区划分提取的脑区平均灰质密度特征可以有效地对 AD患者进行识别, 并可以用来评估个体的认知水平。

针对当前紫外光通信中气体灯与激光光源调制速率慢、体积较大, 光电倍增管工作电压过高等问题, 提出了一种基于现场可编程门阵列技术的日盲紫外LED实时视频传输系统, 其发射光源采用波长为265 nm的单紫外LED, 探测器采用紫外PIN; 通过电路和逻辑模块设计搭建了视频传输系统, 研究了系统的光功率密度与通信距离之间的关系; 基于二进制开关键控调制方式, 实现日盲紫外LED实时视频传输系统最大传输速率为2.88 Mbit/s, 当通信距离为4 m时, 传输速率达到1.92 Mbit/s。

以SCARA机器人、机械夹爪、摄像机CCD为硬件基础，搭建了基于单目视觉的SCARA机器人自动识别和定位插件系统平台，并利用摄像机参数标定和建立的抓取系统参数化模型，将CCD摄像机获取的工件图像坐标信息转化为机器人坐标系下的抓取位置信息。本系统以Visual studio软件为开发平台，利用OpenCV视觉数据库函数进行颜色识别与定位算法开发，经测试，该视觉算法能够实现工件的颜色识别和获取工件的位置信息，并控制机器人夹爪进行目标工件的精确抓取，满足了一般工业生产中抓取工件实时性的要求。

Micro vapor chamber (MVC) for light emitting diodes (LEDs) can be designed and fabricated to enhance the heat dissipation efficiency and improve the reliability. In this paper, we used photoresist SU-8 and electroforming copper (Cu) to fabricate three kinds of wick structures, which are star, radiation and parallel ones, and the substrate is silicon with thickness of 0.5 mm. Electroforming Cu on silicon to make micro wick structure was a critical step, the ampere-hour factor was used, and accordingly the electroforming time was predicted. The composition of electroforming solution and parameters of electroforming were optimized too. After charging and packaging, thermal behavior tests were carried out to study the heat dissipation performance of MVCs. When the input power was 8 W, the parallel wick structure reached the equivalent temperature of 69.0 °C in 226 s, while the others were higher than that. The experimental results prove that the wick structures have significant influence on the heat transfer capability of MVCs.

Micro heat pipe (MHP) is applied to implement the efficient heat transfer of light emitting diode (LED) device. The fabrication of MHP is based on micro-electro-mechanical-system (MEMS) technique, 15 micro grooves were etched on one side of silicon (Si) substrate, which was then packaged with aluminum heat sink to form an MHP. On the other side of Si substrate, three LED chips were fixed by die bonding. Then experiments were performed to study the thermal performance of this LED device. The results show that the LED device with higher filling ratio is better when the input power is 1.0 W; with the increase of input power, the optimum filling ratio changes from 30% to 48%, and the time reaching stable state is reduced; when the input power is equal to 2.5 W, only the LED device with filling ratio of 48% can work normally. So integrating MHP into high-power LED device can implement the effective control of junction temperature.

根据一种过盈联接精密微小组件的装配要求, 研制了一套精密微小组件自动压装仪器。通过分析压装过程中压装力与系统弹性变形的变化关系, 对组件的压装位移进行了有效补偿。采用具有上下视野的机器视觉装置测量压装组件的相对位置, 并采用XY位移平台对组件进行精确对准。提出了对具有上下视野的机器视觉装置的标定方法; 分析了组件压装力与压装位移的关系, 得出了压装力-位移曲线的预测值。对随机抽取的5套零件进行了装配实验, 结果表明: 对组件进行精密对准后, 组件压装位移-压装力曲线符合预测值; 组件装配后相对位置的距离偏差小于±5 μm, 垂直度偏差小于10 μm。实验表明组件装配精度达到了设计指标要求。

为了实现高精密温度测量,设计了高性能数字温度传感器,该传感器由石英音叉谐振器,数字接口电路和基于现场可编程门阵列的传感器重置控制算法构成。依据石英晶体压电效应原理,对石英音叉谐振器的热敏切型和电极设置进行了研究;基于力学振动原理,导出石英音叉谐振器弯曲振动模式的微分方程;讨论了谐振式温度传感器的工作原理,提取出石英音叉温度传感器的特征参数并进行了非线性误差分析;采用光刻和侵蚀技术加工制作了石英音叉谐振器。该传感器的频率输出信号通过数字接口进入现场可编程门阵列,通过重置控制算法实现传感器的重置和现场自动校验。实验结果表明,在-20 ～140 ℃,该传感器的灵敏度可达65×10^-6/℃,测温分辨率为0.001 ℃,响应时间为1 s ,测温精度为0.01 ℃。