局内同类机器加工任务排序问题的一种算法及仿真试验

对于局内同类机器加工任务排序问题设计了一种Space Preparation算法(简称SP算法)，改进了经典LIST算法的平均执行效率．文中证明SP算法的竞争比为2，并通过仿真试验进行验证SP算法的执行效率。     

改进的混合高斯与YOLOv2融合烟雾检测算法

提出一种融合了改进的混合高斯和YOLOv2的烟雾检测算法。首先,针对烟雾的早期特征对混合高斯算法进行改进,有效框定动态目标感兴趣区域,提取出烟雾前景;在此基础上将烟雾检测转换为回归问题,利用端对端目标检测算法YOLOv2训练烟雾数据集,进行二次检测和筛选,最终框定出烟雾发生区域的具体位置和范围,满足对不同场景火灾烟雾的有效检测。实验结果表明,融合算法改善了烟雾区域的检测效果,提高准确性并有效降低烟雾误检率。     

对Sutherland-Cohen算法的改进

提出了对Sutherland-Cohen算法的改进.通过引进辅助直线进行直线裁剪,只需两次求交运算就即求出线段的可见部分,还减少了对线段的重编码运算,具有和NLN算法相同高的效率.通过引入辅助平面进行三维裁剪,克服了NLN算法不能扩展到三维的缺点,而且其三维裁剪效率高于Sutherland-Cohen算法和梁友栋-Barsky算法.     

基于改进PP-YOLOv2的IC引脚焊接缺陷检测算法研究

针对IC引脚焊接缺陷因目标尺寸小、引脚密集导致检测精度低等问题,提出一种基于改进PP-YOLOv2的IC引脚焊接缺陷检测算法。通过在骨干网络后引入SE注意力机制,区分特征图中不同通道的重要性,强化目标区域的关键特征,提升网络的特征提取能力。使用k-means++聚类算法产生9个聚类中心,以降低因初始聚类中心随机选择不当对检测结果所造成的误差影响。实验结果表明：改进算法对IC引脚焊接短路、缺脚、翘脚、少锡缺陷检测的平均精度分别为97.9%, 96.1%, 96.7%, 95.8%;在阈值为0.5的情况下,平均精度均值达到了96.6%,与YOLOv3、PP-YOLOv2相比,分别提高了14.9%, 5.1%。改进算法对单幅图片的检测时间为0.151 s,满足IC质检的速度要求,为IC引脚焊接缺陷检测提供了参考价值。     

改进的射频识别系统的防冲突算法

在射频识别系统中，须采取有效的防冲突算法解决多个标签与阅读器数据交换时引起的数据冲突问题。新的防冲突算法采取动态互补的二进制树形搜索法，充分利用已得到的冲突信息，有效减小了判决过程中数据的传输量，提高了标签的识别效率。     

基于TSP的改进差分进化算法

针对TSP问题,提出一种改进的差分进化算法:利用贪心算法产生初始种群,定义特有的编码匹配函数进行变异操作,排序法修复变异个体,并采用顺序交叉,在变异操作之后,加入新的选择机制,防止交叉操作破坏变异出的优良个体,实验结果表明改进后的差分进化算法能够高效地解决TSP问题,体现良好的优化性能。     

入侵检测模式匹配算法的研究与改进

模式匹配算法是实现基于规则检测的核心技术,其效率直接影响到入侵检测系统的准确性和实时性。通过分析传统的模式匹配算法BM算法和BMH算法等,提出一种基于BM跳跃思想的模式匹配改进算法,简化了初始化过程,加大了匹配失败后向后跳跃的幅度。经过算法测试,与原算法相比新算法可以有效的减少比较次数,提高模式匹配效率。     

基于改进A*算法的移动机器人全局最优路径规划

在栅格地图环境下,传统A*算法搜索路径时选取的路径点受限于栅格中心,并且路径的转折角度固定为特定的离散值,因此存在长度非最优以及冗余转折较多的问题.为此,提出一种新启发搜索策略下的改进A*算法.在探索当前节点的每个邻域节点时,将邻域点父节点的选取范围扩大到从当前点至起始点的整个支路,采用邻域点与支路上的点直接相连的方式...     

基于BMN算法的几点改进

交点算法是计算几何的一个基本算法,也是我们实现空间关系的一个基础.对BMN算法从两方面做改进,一方面单独解决BMN算法的5种特殊情况;另一方面是利用原有的数据结构而不是重新创建新的结构,这样带来效率优势和提高了可移植性.     

人工神经网络BP算法的改进与分析

针对标准BP算法存在的缺陷,讨论一种改进算法LMBP算法,阐述该算法的优化技术原理,并利用MATLAB语言对该算法进行仿真,对训练速度和内存消耗情况做比较.仿真结果证明采用LMBP算法的BP神经网络可以使实际输出与期望输出更接近,误差更小,效率更高.     

基于二进制树的RFID系统自适应多分支防碰撞算法

为提高射频识别(RFID)系统的识别效率,研究了系统的标签防碰撞问题。考虑到对标签进行多分支处理能够有效地提高RFID系统标签识别效率,而传统的多分支防碰撞算法都是在标签估计的基础上对标签进行最优分组,标签估计产生的时延和误差都会影响整体的识别效率,提出了一种基于二进制树的自适应多分支(AMB)防碰撞算法。该算法根据二进制树结构特点,利用识别的标签数目对树结构中右节点标签进行估计并进行分组识别,经过多次调整的自适应多分支防碰撞算法,可以规避一次标签估计所引入的估计误差,从而提高系统的识别效率。仿真结果表明自适应多分支算法可以大大提高标签识别效率,在标签数量较大时系统效率可达43%左右。同时该算法实现简单,只需在阅读器中增加若干计数器,不需要改变任何空中接口,很容易与现有协议兼容。     

Solution of the inverse problem in solidification of binary alloy by applying the ACO algorithm

The paper presents a solution of the inverse problem consisting in reconstruction of the heat flux and the distribution of temperature in the process of binary alloy solidification when the temperature measurements in the selected points of the alloy are known. The considered task is mathematically modelled by means of the heat conduction equation with the substitute thermal capacity and with the liquidus and solidus temperatures varying in dependence on the concentration of the alloy component, whereas for describing the concentration the lever arm model is applied. An important part of the procedure consists in minimization of some functional executed with the aid of ACO algorithm.     

A multi-objective metamodel-assisted memetic algorithm with strength-based local refinement

Metamodel-assisted evolutionary algorithms are low-cost optimization methods for CPU-demanding problems. Memetic algorithms combine global and local search methods, aiming at improving the quality of promising solutions. This article proposes a metamodel-assisted memetic algorithm which combines and extends the capabilities of the aforementioned techniques. Herein, metamodels undertake a dual role: they perform a low-cost pre-evaluation of population members during the global search and the gradient-based refinement of promising solutions. This reduces significantly the number of calls to the evaluation tool and overcomes the need for computing the objective function gradients. In multi-objective problems, the selection of individuals for refinement is based on domination and distance criteria. During refinement, a scalar strength function is maximized and this proves to be beneficial in constrained optimization. The proposed metamodel-assisted memetic algorithm employs principles of Lamarckian learning and is demonstrated on mathematical and engineering applications.     

Binary TLBO algorithm assisted for designing plasmonic nano bi-pyramids-based absorption coefficient

A new efficient binary optimization method based on Teaching–Learning-Based Optimization (TLBO) algorithm is proposed to design an array of plasmonic nano bi-pyramids in order to achieve maximum absorption coefficient spectrum. In binary TLBO, a group of learners consisting of a matrix with binary entries controls the presence (‘1’) or the absence (‘0’) of nanoparticles in the array. Simulation results show that absorption coefficient strongly depends on the localized position of plasmonic nanoparticles. Non-periodic structures have more appropriate response in term of absorption coefficient. This approach is useful in optical applications such as solar cells and plasmonic nano antenna.     

基于免疫遗传算法的二元光学元件的位相设计

在激光诱导扩散中,需要利用二元光学元件对激光器输出的高斯光束进行整形,以实现曝光区的温度分布均匀化。为了得到二元光学元件的位相分布,采用免疫遗传对相位分布进行设计。免疫遗传算法中采取变频率的交叉操作、变异操作,克服了遗传算法在局部搜索解空间上效率差的缺点,并使算法跳出局部极大值的能力得到了增强。采取由正向记忆细胞库提取的免疫疫苗对抗体群进行接种,使群体的进化方向得到引导,提高了算法的进化效率;采取由反向记忆细胞库提取的劣化疫苗对抗体群进行反向接种,减少算法的重复运算,极大地抑制了群体退化;采用B、T细胞的作用机制,保持群体在进化过程中的多样性,很大程度上抑制了算法未成熟收敛。运算结果表明,免疫遗传算法较遗传算法具有更高的算法效率和更强的寻优能力。最后考虑到实际加工,对最优解做适当调整得到了更适合于实际加工的二元光学元件的位相分布。     

A secure authentication watermarking for halftone and binary images

An authentication watermark is a hidden data inserted into an image that can be used to detect any accidental or malicious alteration in the image. Many authentication‐watermarking techniques for continuous‐tone images are described in the literature, but only a quite small number of secure authentication watermarking techniques are available for binary/halftone images. This article proposes a simple solution for inserting a secure authentication watermark in binary/halftone images. It consists of choosing a set of pseudo‐random pixels in the image, clearing them, computing the message authentication code (or the digital signature) of the now‐cleared image, and inserting the resulting code into the selected random pixels. Dispersed‐dot halftone images watermarked by the proposed technique present better visual quality than do watermarked generic binary images. However, in practice, the visual degradation is hardly noticeable in either case. The proposed technique seems to be the only binary/halftone watermarking scheme that can detect even a single pixel alteration in the host image. It can be used with secret‐key or public‐key ciphers. © 2004 Wiley Periodicals, Inc. Int J Imaging Syst Technol 14, 147–152, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ima.20018     

五点二重逼近细分法

提出了一种新的构造曲线的算法——五点二重逼近细分法。利用细分格式的生成多项式讨论了该细分格式的一致收敛性及Ck连续性。该细分格式带有一个张力参数μ,通过选取不同的μ值,可以分别生成C1~C5连续的极限曲线。特别是当μ=9/256时,细分格式生成的极限曲线可以达到C7连续。最后给出了五点二重逼近曲线细分的实例,表明了这种细分格式是有效的。     

A first principles technique for the analysis of alloy phase stability in random binary alloys

In this communication we introduce the augmented-space recursion method coupled with the orbital peeling technique, as a powerful tool for the calculation of effective cluster interactions, useful for the study of alloy phase stability. An application to the well studied PdV system has been carried out. This work was done in collaboration with Tanusri Saha and Indra Dasgupta, S N Bose National Centre for Basic Sciences, Calcutta     

Suspensions of Bernoulli shifts

We show that for a given finitely generated group, its Bernoulli shift space can be equivariantly embedded as a subset of a space of pointed trees with Gromov–Hausdorff metric and natural partial action of a free group. Since the latter can be realized as a transverse space of a foliated space with leaves Riemannian manifolds, this embedding allows us to obtain a suspension of such Bernoulli shift. By a similar argument, we show that the space of pointed trees is universal for compactly generated expansive pseudogroups of transformations.