利用VB实现WinCC归档数据处理

WinCC是一款功能强大的HMI系统,它可以组态界面并可以对数据进行归档,但是对于一些复杂的数据处理如数据查询,计算和统计,就显得力不从心．文章利用VB编程工具建立了WinCC的归档数据库与Access数据库和EXCEL的连接,并生成了．exe文件,通过WinCC的直接调用,实验人员可以根据自己的喜好去选择数据存储方法和路径,实现数据的有条件查询和复杂处理．     

基于TinyOS的无线热舒适度测量系统

单一的室内环境温度作为被控变量的控制系统,难以满足人们对室内环境舒适性以及节能的要求。开发了基于TinyOS操作系统的无线热舒适度测量系统,无线传感网络节点组成多跳网络,用以采集温湿度等室内环境参数,并实时计算热舒适度PMV（Predicted Mean Vote）指标值。分析了测量误差产生的主要原因,并利用超闭球CMAC（Hyperball CMAC,HCMAC）神经网络进行了误差补偿,实验结果表明,补偿后的PMV精度得到了明显的改善。该系统可为热环境舒适度实时控制提供便捷的无线数据采集和有效的PMV指标测量方法。     

A four-point-bending-test for the stability assessment of glass frit bonded molded microsensors

For increasingly miniaturized micromechanical sensors the silicon package is thinned and therefore more sensitive to thermomechanical stresses caused by the production stages. We present a four-point-bending-test that enables the investigation of reliability relevant loading like warping. The initiation of crack growth at the glass frit bonding frame is observed by an infrared camera and critical fracture mechanical parameters are determined for different tensile and shear mode contributions. Based on this crucial fracture mechanical data the stability of silicon packages can be assessed.     

Collaborative design: a process model for refurbishment

Collaborative design related to refurbishment projects is a complex process. The number of parties involved compounds this complexity. So too the iterative nature of design whereby information flow and decision rules are typically nonlinear. Hypermedia is therefore suggested as an appropriate medium as well as a tool to model this process. By reviewing the refurbishment of mechanical and electrical (m and e) equipment within a shopping center, the appropriateness of a hypermedia paradigm is examined. In this way, primary information elements and relations between such elements are identified. These form the basis for the development of the hypermedia model described in this paper.     

结合Conformer与N-gram的中文语音识别

Transformer模型对输入序列中重要的信息进行学习, 相比传统的ASR (automatic speech recognition)模型提升了准确性. Conformer模型在Transformer的编码器中加入卷积模块, 增加了获取细微局部信息的能力, 进一步提高了模型性能. 本文结合使用Conformer模型和N-gram语言模型(language model , LM)用于中文语音识别, 获得了良好的识别效果. 在数据集AISHELL-1和aidatatang_200zh上的实验表明, 使用Conformer模型字错率分别可降低到5.79%和5.60%, 较Transformer模型降低了5.82%和2.71%. 结合N-gram语言模型后字错率分别可降低到4.86%和5.10%达到最佳性能, 实时率(real time factor , RTF)达到0.14566. 测试信噪比降低为20 dB时模型字错率才明显下降到8.58%, 表明该模型具有一定的抗噪能力.     

基于背景分类的运动目标检测算法

针对光线暗、对比度和分辨率低的监控视频,提出了一种基于背景分类的运动目标检测算法。 首先用视频第一帧图像HSV空间的色度H和亮度V作为背景特征进行初始化,建立两种包含色度和亮度特征的背景模型类,即初始化得到的原始背景类和受光照或者其他因素影响得到的在原始背景周围波动的背景波动类,利用这两个背景模型进行前景检测和背景更新。为提高前景检测的准确率,背景模型的更正加入背景更正机制和权重机制,使得背景中样本的数量根据背景的实际情况处在一种动态的变化中,提高前景分割的效率。用不同场景下的监控视频进行算法对比实验,结果证明,该算法获得的前景完整清晰,视频处理的速度较快。提出的算法简单实用,对噪声干扰表现出良好的鲁棒性。     

Neuro-optimal operation of a variable air volume HVAC&R system

Low operational efficiency especially under partial load conditions and poor control are some reasons for high energy consumption of heating, ventilation, air conditioning and refrigeration (HVAC&R) systems. To improve energy efficiency, HVAC&R systems should be efficiently operated to maintain a desired indoor environment under dynamic ambient and indoor conditions. This study proposes a neural network based optimal supervisory operation strategy to find the optimal set points for chilled water supply temperature, discharge air temperature and VAV system fan static pressure such that the indoor environment is maintained with the least chiller and fan energy consumption. To achieve this objective, a dynamic system model is developed first to simulate the system behavior under different control schemes and operating conditions. A multi-layer feed forward neural network is constructed and trained in unsupervised mode to minimize the cost function which is comprised of overall energy cost and penalty cost when one or more constraints are violated. After training, the network is implemented as a supervisory controller to compute the optimal settings for the system. Simulation results show that compared to the conventional night reset operation scheme, the optimal operation scheme saves around 10% energy under full load condition and 19% energy under partial load conditions.     

Enhanced visible light photocatalytic activity of Zn2SnO4 via sulfur anion-doping

Sulfur anion doped Zn₂SnO₄ was prepared by calcining the mixture of thiourea and spinel Zn₂SnO₄ at 300 °C under argon atmosphere and characterized by XRD, XPS and DRS. It was found that S^2? was incorporated interstitially into the bulk phase of Zn₂SnO₄. After the doping of S^2?, the band gap of Zn₂SnO₄ was sharply decreased to 2.7 eV compared with that of undoped Zn₂SnO₄ (~ 3.6 eV). The photocatalytic activity of S-doped Zn₂SnO₄ was enhanced for photodegradation of Rhodamine B (RhB) in aqueous solution under visible light irradiation.     

基于图像特征的彩色图像防拷贝水印算法

数字信号处理和计算机网络技术的迅猛发展,使得数字产品的版权保护问题面临极为严峻的挑战。数字水印技术主要是通过在数字产品中隐藏某些数字信息而达到数据的版权保护和内容认证。本文提出一种新的彩色图像防拷贝水印算法,算法选择彩色图像YUV空间,对色度分量进行离散余弦变换与二值化获得唯一描述图像的特征信息,经过混沌加密后嵌入到亮度分量小波变换系数的中频部分,同时,该算法能够实现盲提取。实验结果表明,该算法对噪声攻击、JPEG压缩、滤波和剪切等攻击具有较好的鲁棒性,同时能够有效地抵抗拷贝攻击。     

融合运动特征和深度学习的跌倒检测算法

为了利用计算机视觉技术准确检测老年人的跌倒状况,针对现有跌倒检测算法中人为设计特征造成的不完备性以及跌倒检测过程中前后景分离困难、目标混淆、运动目标丢失、跌倒检测准确率低等问题,提出了一种融合人体运动信息的深度学习跌倒检测算法对人体跌倒状态进行检测。首先,通过改进YOLOv3网络进行前景与背景的分离,并根据YOLOv3网络的检测结果对前景人体目标进行最小外接矩形标记;其次,分析人体跌倒过程中的运动特征,将人体运动特征向量化并通过Sigmoid激活函数转化为0到1之间的运动权重信息;最后,通过全连接层将将运动特征与卷积神经网络（CNN）提取的特征进行拼接和融合从而实现人体跌倒分类判别。将所提跌倒检测算法与背景差分、高斯混合、VIBE、方向梯度直方图（HOG）等人体目标检测算法及阈值法、分级法、支持向量机（SVM）分类和CNN分类等人体跌倒判断方案进行了对比实验,并将所提跌倒检测算法在不同光照条件下和混合日常噪声运动干扰下进行了实验,结果表明所提算法在环境适应性和跌倒检测准确率上都优于传统的人体跌倒检测方法。该算法能有效检测出视频中的人体并对人体跌倒状态进行准确检测,进一步验证了融合运动信息的深度学习识别方法在视频跌倒行为分析上的可行性与高效性。