基于多元公共品演化博弈的无人集群合作演化机制

针对控制规则失效下的无人集群自主协同问题,提出了一种基于多元公共品演化博弈的无人集群合作演化机制。首先分析无人集群自主协同需求,在愿景驱动规则下,构建多元公共品演化博弈模型。在此基础上,理论推导出模型的平均丰度函数,并仿真计算成本、收益系数、愿景水平三类参数分别对平均丰度的影响。最后,结合案例研究参数变化对集群合作行为演化的作用机理,并给出合理化建议,为相关理论向实际应用转化提供有意义的探索。     

基于LMKL和OC-ELM的航空电子部件故障检测方法

针对航空电子部件故障样本获取困难以及检测准确率不高的问题,提出基于局部多核学习(localized multiple kernel learning, LMKL)和一类超限学习机(one-class extreme learning machine, OC-ELM)的故障检测方法。仅运用正常状态的小样本数据,给出了LMK-OC-ELM的数学表达形式,并在不同的门模型下推导了LMK-OC-ELM中局部核权重的优化方法;在获取局部核权重的基础上,定义了离线故障检测所需的统计检验量与阈值,以便工程实现。将所提方法应用于某型接收机,结果表明,在训练时间可控的前提下,与4种常见的一类分类(one-class classification, OCC)算法相比,所提方法可均衡地提高召回率、查准率和特异度,以LMK-OC-ELM-sig为代表,其在F1、曲线下方面积(area under curve, AUC)、G-mean和准确率4个指标上,比最近提出的局部多核异常检测(localized multiple kernel anomaly detection, LMKAD)方法分别提高了1.60%、1.57%、1.53%和2.23%。     

基于自适应背景模型的动态场景目标检测

背景建模一直是运动目标检测中的一个重要课题。该文提出一个适用于动态背景的基于非参数估计的前景背景对比模型。模型通过核函数估计的方法模拟了像素点五维特征向量的概率分布,并在图像序列中滚动更新。实验证明,上述算法能够在一般目标检测,特别是动态场景(摇动树枝等)的检测中取得较好的效果。     

一种基于聚类分析的数值积分新算法

通过聚类分析找出一般数值求积公式的被积函数出现数值信息波动较大的区域,自适应地采用相应的高精度公式,构造出高精度的自适应数值求积公式.数值试验结果表明,这种新数值求积算法较一般数值求积公式,能显著地减少数据存储量和计算量,提高数值求积公式的稳定性和数值积分的精度.     

Breaking biological symmetry in membrane proteins: The asymmetrical orientation of PsaC on the pseudo-C2 symmetric Photosystem I core

The elucidation of assembly pathways of multi-subunit membrane proteins is of growing interest in structural biology. In this study, we provide an analysis of the assembly of the asymmetrically oriented PsaC subunit on the pseudo C₂-symmetric Photosystem I core. Based on a comparison of the differences in the NMR solution structure of unbound PsaC with that of the X-ray crystal structure of bound PsaC, and on a detailed analysis of the PsaC binding site surrounding the F_X iron-sulfur cluster, two models can be envisioned for what are likely the last steps in the assembly of Photosystem I. Here, we dissect both models and attempt to address heretofore unrecognized issues by proposing a mechanism that includes a thermodynamic perspective. Experimental strategies to verify the models are proposed. In closing, the evolutionary aspects of the assembly process will be considered, with special reference to the structural arrangement of the PsaC binding surface. Received 22 October 2008; received after revision 17 November 2008; accepted 05 December 2008     

Dark matter,the Equivalence Principle and modified gravity

Dark matter (DM) is an essential ingredient of the present Standard Cosmological Model, according to which only 5% of the mass/energy content of our universe is made of ordinary matter. In recent times, it has been argued that certain cases of gravitational lensing represent a new type of evidence for the existence of DM. In a recent paper, Peter Kosso attempts to substantiate that claim. His argument is that, although in such cases DM is only detected by its gravitational effects, gravitational lensing is a direct consequence of Einstein's Equivalence Principle (EEP) and therefore the complete gravitational theory is not needed in order to derive such lensing effects. In this paper I critically examine Kosso's argument: I confront the notion of empirical evidence involved in the discussion and argue that EEP does not have enough power by itself to sustain the claim that gravitational lensing in the Bullet Cluster constitutes evidence for the DM Hypothesis. As a consequence of this, it is necessary to examine the details of alternative theories of gravity to decide whether certain empirical situations are indeed evidence for the existence of DM. It may well be correct that gravitational lensing does constitute evidence for the DM Hypothesis—at present it is controversial whether the proposed modifications of gravitation all need DM to account for the phenomenon of gravitational lensing and if so, of which kind—but this will not be a direct consequence of EEP.     

基于能力分析的空间对抗装备体系效能评估方法

首先研究了空间对抗的核心能力要素;然后研究了空间对抗装备体系的构成,即将核心能力要素细化为基本能力要素并映射到空间对抗装备子系统;最后构建由作战能力指标→核心能力指标及权重赋值→基本能力指标及权重赋值→单项效能指标及权重赋值组成的指标体系,并提出基于该指标体系的单项效能评价→系统综合评价→体系综合评价→作战效能评价→结果验证的综合评估途径.     

末端修正子母弹对面目标的射击效能评估

作为射击精度评估的常用指标,无论是圆概率偏差,还是命中概率都没有涉及面目标的功能结构的重要度,而命中重要程度不同的区域,其射击效能是不同的。针对末修子母弹打击面目标时射击效能的评价问题,在通过数值模拟法产生末修子母弹的落点散布,以及利用毁伤树分析方法对面目标进行功能区域划分的基础上,利用Monte-Carlo仿真方法计算了子母弹对各功能区域的命中概率Pi,并定义了基于区域结构及重要度的射击命中度和毁伤概率,从而综合评估了整体的射击效能。最后通过仿真算例说明了该方法的合理性,并利用均匀设计法定量分析了末修子母弹的性能参数对射击效能的影响。     

基于遗传算法的改进径向基支持向量机及其应用

通过对径向基核函数进行分析后发现:根据样本各个特征的识别能力赋予其不同大小的核参数,可以提高支持向量机的推广能力。此结论基础上,提出了一种基于遗传算法的多核参数径向基支持向量机算法,通过遗传算法最小化验证误差,实现了根据各个特征的识别能力赋予其不同大小的核参数。将该算法用于轴承故障诊断,实验结果表明,与传统支持向量机相比,多核参数径向基支持向量机具有更好的推广能力,同时,核参数的大小反映了对应特征识别能力的大小。     

基于一种最优核鉴别矢量集的雷达目标识别

基于高分辨距离像的雷达目标识别的关键是如何提取距离像的有效鉴别特征.通过分析核Fisher准则函数下核类内散布矩阵的零空间信息与非零空间信息对识别的贡献,基于同构映射思想,求解出一组最优鉴别矢量集,用于提取距离像的鉴别特征.同时,给出一种快速算法,用较少的训练样本表达投影方向,加快了鉴别特征的提取速度.基于外场实测数据的实验证明了方法的有效性.