Zernike矩的不变性分析及其改进

Zernike矩作为一种基于区域的形状描述子,具有良好的旋转不变性,能够很好地提取图像内部的形状信息。介绍了Zernike矩的定义,对其不变性进行了深入分析,并对Zernike矩算法进行了改进,使其在具有良好的旋转不变性的基础上,也具有较好的平移不变性和比例不变性。结果表明,改进后的Zernike矩具有更好的不变性,可更好地应用于图像识别中。     

基于深度森林的网络流量分类方法

随着网络应用的迅猛发展,流量分类在网络资源分配、流量调度和网络安全等诸多研究领域受到广泛关注。现有的机器学习流量分类方法对流量数据特征的选取和分布要求苛刻,导致在实际应用中的复杂流量场景下分类精确度和稳定度难以提高。为了解决样本特征属性的复杂性给分类性能带来的不利影响,引入了基于深度森林的流量分类方法。该算法通过级联森林和多粒度扫描机制,能够在样本数量规模和特征属性选取规模有限的情况下,有效地提高流量整体分类性能。通过网络流量公开数据集Moore对支持向量机、随机森林和深度森林机器学习算法进行训练和测试,结果表明基于深度森林的网络流量分类器的分类准确率能够达到96. 36%,性能优于其他机器学习模型。     

Study on dynamic response of multi-degree-of-freedom explosion vessel system under impact load

In order to study the dynamic response and calculate the axial dynamic coefficient of the monolayer cylindrical explosion vessel, the wall of vessel is simplified as a multi-degree-of-freedom (MDoF) undamped elastic foundation beam. Decoupling the coupled motion equation and using Duhamel's integrals, the solutions in generalized coordinates of the equations under exponentially decaying loads, square wave loads and triangular wave loads are calculated. These solutions are consistent in form with the solutions of single-degree-of-freedom (SDoF) undamped forced vibration simplified model. Based on the model, equivalent MDoF design method (also called MDoF dynamic coefficient method) of cylindrical explosion vessel is proposed. The traditional method can only predict the dynamic coefficient of torus portion around the explosion center, but this method can predict that of the vessel wall at any axial n dividing point position. It is verified that the prediction accuracy of this model is greatly improved compared with the SDoF model by comparing the results of this model with SDoF model and numerical simulation in different working conditions. However, the prediction accuracy decreases as the scaled distance decreases and approaches the end of the vessel, which is related to the accuracy of the empirical formula of the implosion load, the simplification of the explosion load direction, the boundary conditions, and the loading time difference.     

适应低时延业务需求的分布式可穿戴单兵作战信息系统

针对云计算应用在单兵作战系统场景下业务处理时延高、服务质量无法保障的问题,提出一种基于可穿戴计算的分布式单兵作战信息系统。利用士兵身上的可穿戴智能设备构建本地计算层,在作战地点就近处理计算数据,提供给士兵本地的信息处理与融合能力,并采用广义扩散负载均衡算法平衡各设备负载,降低业务处理时延;同时利用分布式计算的容错能力增强系统的可靠性。仿真结果表明,基于可穿戴设备的分布式本地网络架构能有效地降低作战任务的处理时延,同时增强系统的可靠性。     

Reforming the bras tendus? The Congolese National Police reforms, 2003–2016

The security sector reform (SSR) programme in the Democratic Republic of the Congo (DRC) has failed, according to a former high-level member of the United Nations (UN) mission in the DRC, as a large section of the country remains outside government control, and the security institutions of the state continue to constitute one of the predominant sources of insecurity for the local population in a number of ways. Based on several field studies, this article critically scrutinises the SSR of the Congolese National Police (Police nationale congolaise; PNC) and the efforts to reform it between 2004 and 2016. It further attempts to explain why so little progress was made in the SSR of the PNC, despite extensive involvement from donors. The article shows that the instrumental and traditional approach to SSR is partly to blame, because in this case it failed to address the root problems and initiate the needed fundamental reform and reconstruction of the police force. It also shows that reforming local security institutions becomes even more difficult when the local authorities do not support the effort.     

基于核主成分分析的铁谱磨粒特征提取方法研究

针对铁谱分析的磨粒识别过程中存在原始磨粒特征描述指标参数多、非线性突出的问题,提出基于核主成分分析的磨粒特征提取方法,介绍该方法的原理与算法。结合某柴油发动机故障检测与分析系统中铁谱磨粒自动识别的应用实例,并与传统主成份分析方法进行对比分析,结果表明该方法在进行样本非线性特征参数指标综合以及特征维数压缩方面具有可行性和有效性。     

试论公安现役高等院校办学特色

办学特色是高等学校生存发展的前提和办学优势的集中体现,也是形成核心竞争力的关键。公安现役高等院校突出办学特色,要着力创新办学理念,强化学科专业特色和人才培养特色。     

数据中心火灾危险性分析与防治对策

数据中心极其重要,一旦发生火灾,后果严重。为保证数据中心免受火灾危害,在介绍数据中心的特点的基础上,分析了其火灾特点及火灾危险性,提出了合理选择消防设施、加强电气防火、建立长效管理机制等防火对策。     

基于多元高斯概率分布的无线局域网定位方法研究

针对无线局域网环境及室内目标定位需求,研究了基于接收信号强度指示的定位技术,提出了一种基于多元高斯概率分布的信号强度模式匹配方法,讨论了定位系统框架。试验结果表明,该方法具有较好的室内定位效果。