Sift‐proofness of Dangerous Goods Packagings – An Incompletely Defined Requirement

Sift‐proofness is a requirement for different types of dangerous goods packagings for solid substances according to the international Dangerous Goods Regulations. In these regulations, a sift‐proof packaging is defined as a packaging that is completely impermeable to dry contents. This means indirectly that absolutely no mass transport of solid substances is allowed. Moreover, this requirement applies both to the original filling substance and to fine solid material generated during transport. Further specifications, test conditions or tolerable limit values are not given. This is in contrast to physical principles and the usual practice in other fields of technology in which sift‐proofness is relevant. This paper shows the necessary steps for how the requirements for sift‐proofness of dangerous goods packagings can be defined more precisely. Physical basics of the term ‘sift‐proofness’ are explained. A qualitative as well as a quantitative approach is possible. In any case, it is essential to carry out appropriate vibration tests to assess the sift‐proofness. There is a need for systematical investigations of the sift‐proofness of dangerous goods packagings. Copyright © 2017 John Wiley & Sons, Ltd.     

基于频率筛分的无监督人体姿态特征提取与识别研究

基于视频序列的人体行为分析需要检测和判别人体姿态,已有人体姿态检测与判别方法往往达不到实用性要求.从两个方面探讨应用BEMD(bidimensional empirical mode decomposition)算法提升特征分离度与判别性,以进行人体姿态检测和判别:BEMD分解源图像得到的多层固有模态图BIMF具有判别特征,可形成具有强边缘的对比度高的区域,其中包括人体轮廓区域;从低分辨率尺度BIMF图像到高分辨率尺度BIMF图像递归计算,建立基于BEMD的多尺度树(BEMD muhiscale-trees tructured)模型,快速提取目标区域并获取人体形状轮廓特征.实验证明,利用该方法进行人体姿态轮廓特征提取,并建立人体姿态的简化模型,可快速检测并判别人体姿态,以达到实时识别.     

基于图匹配的人脸识别算法

为了识别有较大表情变化的人脸,利用sift算法找到人脸图像中的DoG关键点,用这些极值点所包含的灰度信息和位置信息建立最小生成树,再使用最小生成树估计联合Rényi熵,结合图论中的图匹配知识,进行人脸识别.实验结果表明,该算法在人脸出现较大表情变化的情形下仍能够得到较为准确的识别结果.     

基于PCA-SIFT特征的目标识别算法

针对尺度不变特征变换(SIFT)算法在匹配时特征向量过多,从而导致耗时过长的问题,提出PCA-SIF]算法,对目标进行匹配与识别.首先,利用SIFT算法提取出原图像中稳健的特征点以及特征向量;其次,利用PCA算法对SIFT特征向量的维数进行约减;最后利用降维后的图像与原始图像进行匹配.实验证明,与原始SIFT算法相比,该算法不仅保持了SIFT算法的鲁棒性和稳定性,同时提高了匹配效率,增强了实时性.     

离心甩浆机筛桶几何角度的设计研究

根据离心甩浆机工作原理和实际存在问题,从几何角度对甩浆机筛桶进行设计,从而大大提高甩浆机的效率,有一定的实用价值。     

对ZD型直线等厚筛筛面分段和增大筛面倾角的探讨

对提高ZD型直线等厚筛处理能力提出了将筛面分成多段,可降低物料层高度,使筛孔透筛能力增强;合理分配参数,强化筛分过程;增大筛面倾角,提高物料运动速度等问题进行了探讨,并在实际中进行了验证。     

Influence of the length of the internal sleeve valve on the release of powdery substances from the closures of valved bags for dangerous goods

In practice, checks on dangerous goods transports often detect leaks of powdered dangerous goods from valved bags. In this work, the influence factors of a sudden release of powdery substances from the valves of valved bags were investigated. Drop tests were performed on paper bags of UN design type 5M2 with internal sleeve valve using 2 different powdery substances (Esplas H130 and zinc oxide “Rotsiegel”). The internal sleeve valves of all test samples were not sift‐proof with respect to both filling substances. For almost all test samples, the Esplas H130 powder already leaked out of pasted joints during manual filling. This is a contradiction to the requirement in UN 6.1.4.18.1, according to which closures and joints of paper bags 5M2 should be sift‐proof. In the drop tests, longer valve lengths had a greater sealing effect for both filling substances (for filling degrees of at least 95% and for test samples which had already been mechanically loaded). As an extreme example, at the drop height of 1.20 m and a filling degree of 100%, the released amount of zinc oxide powder from a 10‐cm‐long valve was about 16 times higher than from a valve length of 12.5 cm. The valve length is therefore a safety‐relevant parameter and should be specified by the manufacturer. To ensure that only filling goods with similar physical properties in comparison with the test substance are used for valved bags, the user must be informed of the particle size of the test substance.     

一种基于OUS的最大频繁项集挖掘算法

分析实际应用中有效访问序列的特点,提出了一种采用自底向上策略快速挖掘最大频繁项集的OUS算法。该算法首先对用户项集进行重叠操作统计浏览次数,然后合并,依据用户给出的最小支持度删除原项集中的非频繁页面元素,并对两两用户项集筛选生成候选频繁项集,最后扫描数据库,统计各个候选频繁项集的支持度计数。实验结果表明,该算法能有效地发现用户最大频繁项集。     

快速脑白质损伤分割

本文提出了一种快速鲁棒的脑白质损伤分割算法。该算法能可靠地分割白质损伤区域。与大多数基于学习的算法不同,他们大多以专家标记的映射图为基础来学习分类器。本文方法基于如下事实：白质损伤区域的体素相对于正常体素具有较高的亮度。该方法只需要用户标出一些包含白质损伤的兴趣区域（ROI）,然后算法首先执行一个K-means聚类,做初步的白质损伤区域标记。对标记的区域做进一步的学习获得一幅损伤区域的概率图,根据概率图产生一张置信图。最后引入SIFT算法中的多尺度处理的方法。在初始的标签的基础上,将依据邻近的多层切片的概率分布做进一步的标签的确认,最后得到较好的分割结果。     

双质体自同步直线振动筛分机动力学分析

进行了双质体自同步直线振动筛分机的动力学分析 ,确定了其参振质量、支承弹簧刚度、振动器偏心块的质量以及振动圆频率之间的关系 ,为支承弹簧设计和筛分处理量精确计算提供了条件 ,并为受力分析和强度计算提供了基础