金属切削用量的智能选择与优化

介绍传统选择切削用量的过程和存在的缺点,并提出切削用量优化设计方法.首先确定目标函数,再建立约束条件,然后进行优化设计,因为目标函数为多变量,非线性,约束条件多而复杂,采用直接搜索分层迭代法,对多个复合形计算,并取各复合形中最好与次好点组成新复合形后进行第二层迭代,获取全局最优解.     

基于动态聚类的地面移动目标的低空跟踪

文中提出了一种基于动态聚类的地面移动目标低空跟踪方法。该方法根据目标跟踪的实际情况选择分类特征,确定聚类准则函数,采用动态聚类方法,依次对像平面上的象素进行子窗口聚类,然后对子窗口内的象素进行运动区域聚类,最后对运动区域的象素进行目标聚类,从而比较精确地获得被跟踪目标的重心位置,并将其作为下一帧的聚类中心。由此,目标跟踪的过程变成一个动态聚类的过程。该方法是一个非监督学习方法,简单实用,仿真实验也证明了该方法的有效性。     

Hypothalamic regulatory peptides and their receptors: Cytochemical studies of their role in regulation at the adenohypophyseal level

Hypothalamic regulatory peptides bind to specific receptors on target cells in the pituitary and control secretion. They in turn can be regulated at the pituitary level by steroid and peptide modulators. Affinity cytochemical techniques are important tools for the identification of specific target binding sites for these regulatory peptides. This presentation reviews the work in which potent, biotinylated ligands of gonadotropin releasing hormone (bio-GnRH), corticotropin releasing hormone (bio-CRH), and arginine vasopressin (bio-AVP) were applied to study the target cell responses. Bio-GnRH, bio-CRH, and bio-AVP bind to membrane receptors on specific anterior pituitary cells. Dual labeling for either gonadotropin or adrenocorticotropin (ACTH) antigens further identified the target cells. After 1–3 minutes, the label was in patches or capped on the surface. After 3 minutes, it was internalized in small vesicles and sent to receptosomes and vacuoles in the Golgi complex. Eventually the biotinylated peptides, or a metabolite, was found in the lysosomes (multivesicular bodies) and a subpopulation of secretory granules. The route and rate of uptake was similar to that described for the classical receptor-mediated endocytosis process. In contrast, intermediate lobe corticotropes internalized the bio-CRH in less than 1 minute. The route through the Golgi complex appeared to be bypassed. Instead the labeled peptide was in vesicles, on the membranes of scattered vacuoles, and in multivesicular bodies. Modulation of ligand binding by steroids showed that changes in receptor numbers correlated with changes in the number of cells that bound the ligand. In male rats, dihydrotestosterone reduced the percentage of GnRH-bound cells by 50%. Most of the reduction appeared in cells that stored luteinizing hormone (LH) antigens. In diestrous female rats, estradiol increased the percentage of bio-GnRH-bound cells. However, the steroid decreased the percentage of GnRH-bound cells in cells from proestrous rats. Glucocorticoids decreased the percentage of CRH-bound corticotropes in as little as 10 minutes. Potentiation of secretion by these ligands was correlated with increases in the percentage of ligand-bound cells. AVP pretreatment of corticotropes increased the percentage of cells that bound bio-CRH. It also increased the rate of receptor-mediated endocytosis of CRH and changed the route so that the Golgi complex was bypassed. This effect could be mimicked by activation of its second messengers (calcium and protein kinase C). Similarly, CRH pretreatment increased the percentage of corticotropes that bound AVP. Thyrotropin releasing hormone (TRH) pretreatment also increased the percentage of thyrotropes that bound AVP. Finally, calcium or sodium channel blockers altered CRH binding so that fewer cells were labeled. This binding by CRH was not dependent on extracellular calcium and tests with a calcium channel agonist showed that it was related to activation of calcium channels. To summarize, these affinity cytochemical studies have identified specific target cells in the pituitary for GnRH, CRH, and AVP. They have also identified heterogeneity in the population. They have demonstrated new information about the direct modulatory effects of steroids, ion channels, and neuropeptides on neuropeptide binding by subpopulations of these target cells.     

目标驱动的基于形状线探索布线算法

对于实现焊盘外形尺寸各异、线宽及线间距离多变的印刷电路板布线,本文提出了目标驱动的基于形状线探索布线算法,该算法是在对障碍物做出包容矩形的基础上,又结合无网格线探索布线算法的特点来完成布线,使得布线具有目标性,同时,对于障碍物采用了灵活的绕障探索,大大提高了布线的成功率和布线速度。     

光电经纬仪测量信息的无损压缩技术

在光电经纬仪测量信息转换成数字图像过程中,如果直接将光测胶片转换成位图图像,形成的数据量极大,不利于传输和存储,而进行压缩又影响图像质量,降低判读精度. 本文采用图像处理技术,根据光测胶片的信息记录原理及数据处理的具体要求,设计了一种将胶片图像划分成不同区域进行处理的方法,对影响判读精度的区域进行直接存储,对其他区域进行压缩处理,并保存各区域与十字丝中心的相对位置以保证图像恢复时各区域精确地合成,此种方法降低了图像信息的数据量,确保了数据处理精度,实现了光测胶片图像的数字化,解决了光电经纬仪测量信息存储和传输中数据量巨大的难题.     

临近空间高超声速跳跃式滑翔目标跟踪模型

针对临近空间高超声速跳跃式滑翔目标跟踪问题,在分析目标运动特性的基础上提出了一种二阶时间相关的新型机动跟踪模型,该模型的核心是将加速度作为具有衰减振荡自相关的零均值随机过程,并据此构建了跟踪临近空间高超声速跳跃式滑翔目标的状态方程;为进一步分析模型的运动适应性,结合卡尔曼滤波算法推导了模型的系统动态误差稳态值,从模型参数取值的角度探讨了模型的适应性问题;为提高模型参数设计的合理性,分析了模型中两个参数的对应关系,并给出了参数设计的大致参考区间.理论分析表明,模型具备周期性与衰减性的统一,在短时间内主要表现为周期性,在长时间内主要表现为指数衰减性,这一特性提高了对临近空间高超声速跳跃式滑翔运动描述的合理性,此外,该模型跟踪临近空间高超声速跳跃式滑翔目标时的参数理论取值具有较低的系统动态误差稳态值.仿真实验表明与已有的临近空间高超声速目标单噪声机动模型相比,该模型具有较高的跟踪精度;并通过比较不同参数下的仿真结果一定程度说明了参数取值方法的合理性.     

软件定义的内网动态防御系统设计与实现

当前,自带设备（BYOD）的兴起对传统基于边界的内网防护观念提出了新的挑战——内部不设防导致堡垒易从内部攻破.从扰乱攻击链的角度,本文提出了"隔离+动态"的防护方法,设计并实现了一种基于软件定义的内网动态防御系统.通过为内网终端分配虚拟IP地址空间,以隐藏各自的真实信息;并且将IP跳变和路径跳变结合起来,实现了更全方面的防护.结果表明,在正常网络应用不受影响的情况下,该系统能大幅降低网络侦察扫描的可用性,阻断网络窃听,提高攻击者实时攻击难度.     

融合灰度与显著性特征的空中红外目标跟踪

针对红外目标特征简单且信息量少导致跟踪精度不高,提出一种基于灰度和显著性特征融合的核相关滤波算法用于空中红外目标跟踪。首先,在保证目标足够特征信息量的前提下对较大的目标进行不同等级压缩。然后将提取的二维灰度特征与显著性特征按页方式进行拼接扩展成三维特征,再将融合的特征用于核相关滤波。实验表明所提算法能够适应多种环境下的空中红外目标跟踪,跟踪精度和成功率典型值分别达到84.8%和63.9%,较大部分算法有很大提高,平均跟踪速度高达125 f/s,体现出了良好的实时性。因此,本文提出的算法在保证实时性的同时提高了跟踪的可靠性,具有一定的实用意义。     

基于数字射频存储的假目标高度控制干扰技术

针对基于数字射频存储（DRFM）的雷达高度欺骗干扰方法存在假目标逼真度低、易被识别和易被滤除的问题,为提高假目标干扰效果,根据雷达顺序波瓣法测仰角和DRFM干扰机的工作原理,提出功率等比转发与比例延时相结合的假目标高度控制算法,阐述了无人机载DRFM干扰机实施干扰的工作流程。经仿真计算,对比了高度控制算法、非均匀重复转发和功率颠倒转发所产生的假目标在航迹、高度和速率方面的差异,证明了高度控制算法产生的假目标运动状态更接近于真实目标,从而验证了该方法的有效性。     

采用改进型AlexNet的辐射源目标个体识别方法

针对辐射源目标精确识别需求,结合以深度学习为代表的机器学习理论技术,提出将改进型AlexNet作为特征提取器,实现目标细微特征提取固化,形成智能化识别网络模型。以广播式自动相关监视（ADS-B）信号为实验对象,在机场实地采集了13个目标的ADS-B脉冲信号数据作为辐射源目标个体识别的训练和测试样本,利用AlexNet和改进的AlexNet验证了算法的有效性。结果表明,改进的AlexNet网络训练时间更快,综合识别率达到98.32%.