基于小波变换和概率神经网络的焊接接头类型识别

为了实现机器人自动焊接过程中快速、精确地提取焊缝特征信息,提出了一种基于小波变换和概率神经网络的焊接接头类型识别方法。先采用小波变换对由激光视觉传感器采集的焊接接头图像进行降噪和增强,对重构后的图像进行二值化,然后提取图像的特征信息,组成图像特征向量,最后构建概率神经网络分类器并进行测试。结合视觉传感器中激光器与摄像机的位置关系,最终识别出4种焊接接头。实验结果表明,所提出的方法特征提取简单,识别率高,并具有较好的实时性。     

国际次声监测研讨会

全面核禁试条约第三届全球次声监测工作研讨会于1997年8月25至28日在美国新墨西哥州西班牙式的古城圣菲举行．会议云集了中国、美国、法国、俄罗斯、阿根廷、澳大利亚等5大洲12个国家61位活跃在次声学科领域的研究人员、政府官员以及联合国临时技术秘书等要员，围绕着全球次声监测中的7个专题进行报告和讨论，它们依次为：欢声阵的设计和信号处理，次声阵性能和减噪设备；法国欢声监测系统，国际次声监测系统60个站网的能力模型，对流层风对长距离次声传播的影响，高空风对次声同性能的影响，渗透管的特性以及对次声监测的减噪作用，欢声减噪器；爆炸检测，声遥感技术对爆炸源能量的估计，小当量地下，地面和近地面爆炸远距离声传     

An Anti-Control Scheme for Spiral under Lorenz Chaotic Signals

The Fitzhugh-Nagumo (FHN) equation is used to generate spiral and spatiotemporal chaos. The weak Lorenz chaotic signal is imposed on the system locally and globally. It is found that for the right chaotic driving signal,spiral and spatiotemporal chaos can be suppressed. The simulation results also show that this anti-control scheme is effective so that the system emerges into the stable states quickly after a short duration of chaotic driving (about 50 time units) and the continuous driving keeps the system in a homogeneous state.     

Multi-path propagation of acoustical wave and time reversal field in a solid plate

The multi-path effect of the acoustical wave in a solid plate is studied. The multireflection and wave conversion of the cylindrical compressional and shear waves, which are excited by an infinite strip on a free surface of the solid plate, are analyzed thoroughly by the far-field approximation method. The concise analytical representations of the cylindrical waves are obtained. The time reversal processing is then applied to the propagation of the cylindrical waves and analyzed theoretically and experimentally. It is shown that the waves coming from different array elements and different paths all arrive at the original place after the time reversal operation. It indicates that the time reversal can compensate automatically the wave aberration caused by the multi-path effect. The self-adaptive focusing of the time reversal field is also analyzed quantificationally by the focusing gain and the ratio of the principal to the second lobe. The effects of the focus position and the aperture of the transducer array on the focused field are also investigated. It shows that theoretical and experimental results are consistent to each other very well.     

基因组重排问题的一个近似算法

分子生物学中基因无方向的反向基因组重排问题在数学上已被证明是一个NP困难问题．基于断点图的概念，给出一个时间复杂性为O(max{b^(π)，nb(π)})，空间复杂性为0(n)的求其近似最优解的算法．其中n为基因组中基因个数，π=(π1,π2，…，πn)表示n个基因的一种排列，b(π)表示排列π中的断点数．数据实验的结果表明，该近似算法可以求得较好的结果．     

Temporal Skewness of Electromagnetic Pulsed Waves Propagating Through Random Media with Ebedded Irregularity slab

Electromagnetic pulsed waves can be distorted in the propagation through random media,and their energy distributions change along the leading and trailing edge of the waveform,which can be presented by the temporal skewness.The skewness presents asymmetry and is treated by the third-order temporal moment,in which an analytic solution for the two-frequency mutual coherence function is obtained recently.Then,transionospheric pulses are discussed in details,Both theoretical analysis and numerical computation indicate that the contributions from scattering and dispersion of irregularities dominate over those of background,so the latter can be neglected in most cases.Also,the temporal skewness of a transionopheric pulse is negative and energy is shifted to the leading edge.     

带惩罚项与随机输入的BP神经网络在线梯度学习算法的收敛性

本文对三层BP神经网络中带有惩罚项的在线梯度学习算法的收敛性问题进行了研究,在网络训练每一轮开始执行之前,对训练样本随机进行重排,以使网络学习更容易跳出局部极小,文中给出了误差函数的单调性定理以及该算法的弱收敛和强收敛性定理。     

农户信用评估系统的设计与运用研究

农户信用评估的研究对推动农村消费信用的发展，促进农村经济良好运行十分重要。本在构造农户信用评估的指标体系的基础上，提出了农户信用评估神经网络模型的算法，利用实际搜集到的农户资料进一步建立了农户信用评估模型，继而构造了农户信用评估系统，并举例说明了该系统的实际运用，以期能为农村经济发展中的农户信用评价及相关研究提供一丝基础性启发。     

BP神经网络在上海住宅市场需求预测中的应用

人工神经网络是近期发展最快的人工智能领域研究成果之一 ,本文在介绍 BP神经网络的有关原理的基础上 ,建立了一个上海住宅市场的 BP神经网络模型 ,并通过该模型对上海住宅市场的需求进行了预测和分析 .分析结果表明人工神经网络方法在住宅市场需求预测中的应用是可行的并且是有效的 .