应用ART2人工神经网络自组织形成焊缝图象中的典型空间模式

应用ＡＲＴ２人工神经网络算法，使采集到的焊缝横截面方向上的灰度分布数据自组织形成若干种空间模式，并把它们作为典型空间模式存储在ＡＲＴ２人工神经网络的ＬＴＭ中。对实时采样到的灰度分布进行空间模式匹配程度检验，根据模式分布情况确定出焊缝位置。文中对梯度法检测结果进行了分析和比较，结果表明基于ＡＲＴ２人工神经网络的焊缝位置检测方法具有更强的噪声抑制能力，因而检测结果更准确、可靠。     

Daily impact of climate on the pattern of urban family life : Indigenous courtyard houses of Baghdad regions of the hot-dry climates Part I: Daily shifts or daily movements in summer

During the day, there are considerable variations in the climatic factors prevailing in these regions. This is especially so during the 24-hour daily cycle of the long, hot and dry summer (mid-May to midOctober). Such climatic conditions have had a considerable impact on the daily living pattern of family life in the same season, particularly in the urban areas. During the same 24-hour cycle, they have compelled the inhabitants to shift their living activities both vertically in section and horizontally in plan; they do this in pursuit of more acceptable, if not, desirable internal thermal environmental ambience.     

Comparison of Displacement and Mixing Diffusers

In order to cumpare the peformance of different supply diffuers of ventilation air, the airflow passern, temperature stratifiation and contaminant dispersion in a furnitured office ventilated by three kinds of air diffuer were numerically investigated. The air diffuers studied in this paper are a quarter-cylinder displacement diffuer on the floor and mixing diffuers (linear and vortex diffuers) on the ceiling. The heat sources in the of-fice are considered to be 50% convective and 50% radiative. The k-? two-equatwn model of turbulence is employed to predict the turbulent diffusion. The results show that the displacement diffuser provides a rather uniform flow field with low velocify in most areas, and the vertical temperature difference from floor to ceiling is as high as 6 K. With the linear diffuser, the air velociry is high, and the temperature is uniform both horizontally and vertically. The air velocity generated by the vortex diffuser is moderate. The distributions of the temperature and the contaminant are rather uniform.     

地震数据油气预测中的属性优化方法

本文简单介绍了智能信息处理中新出现的RoughSet（RS）理论及属性选择方法，从双相介质地震波传播理论角度，探讨了地震数据油气预测属性优化原理，提出了基于RS理论的地震数据油气预测属性优化方法。实际应用表明；本方法速度快、易实现，而且在优选属性、最大程度地减少提取地震属性种数、提高分类正确率等方面，明显优于其它方法。本方法将成为地震数据油气预测的一种有效手段。     

从细线化汉字图像生成任意倍率汉字

本文提出了一种对汉字作任意倍率放缩的算法。该算法先对印刷体汉字作细线化处理，抽取笔划特征点。然后，分割和提取笔划。任意倍率汉字生成是通过变换函数。改变特征点位置，再附加原探察到的笔划修饰来实现的。实验表明本算法是有效的。     

我国模压全息技术发展现况述评

模压全息技术作为一种高科技已在近年来得到迅速的发展并趋于成熟,模压全息能以较低的成本和较高的速度进行大批量的工业化生产。本文对我国模压全息技术发展现况进行了述评。     

压电复合材料PT－P（VDF／TeFE）的静水压压电性

用热轧辊机制备了压电复合材料ＰＴ－Ｐ（ＶＤＦ／ＴｅＦＥ），并利用ＬＣＲ电桥和Ｂｅｒｌｉｎｃｏｕｒｔ电路分别测量了其介电常数和压电常数，实验结果表明它具有良好的静水压压电特性。最后讨论了该复合材料在水听器中的应用     

Application of Differential Evolution for Irrigation Planning: An Indian Case Study

The present paper demonstrates the applicability of population based search optimization method, namely, Differential Evolution (DE) to a case study of Mahi Bajaj Sagar Project (MBSP), India. Ten different strategies of DE are employed to assess the ability of DE for solving higher dimensional problems as an alternative methodology for irrigation planning. The parameters considered in DE are population size, crossover constant and weighting factor. Linear Programming (LP) is utilized as a comparative approach to assess the ability of DE. Comparison of results of LP and the 10 DE strategies for the given parameters indicated that both the results are comparable even for high dimensional problems. Extensive sensitivity analysis studies, performed for 3,600 combinations of above parameters for the 10 DE strategies suggested that DE/rand-to-best/1/bin strategy is the best strategy giving maximum benefits taking minimum CPU time. It is concluded that DE can be utilized for efficient planning of any irrigation system with suitable modifications.     

OFA-II型定量荧光仪在油气勘探中的应用

定量荧光仪在现场应用中受到各种因素的影响:如岩屑样品浸泡时的干、湿;图谱是否出现了淬灭;环境温度的影响等等。文章通过具体事例给出解决这些问题的办法。     

RSFM网络及其在信号滤波中的应用

介绍了一种适用于模式识别的新型神经网络模型--局部有监督特征映射（Reglonal Supervised Feature Mapping,RSFM）网络,描述了该网络的拓扑结构和学习算法,研究了网络的基本性能,最后将其应用到了信号滤波中.理论研究和仿真实验表明,该网络结构简单、算法简洁,收敛速度快、识别精度高,适用于需要大样本训练、随机干扰严重的复杂模式的分类与识别.