多时相Radarsat数据在广东肇庆地区稻田分类中的应用

将1996年获取的4个时相的Radarsat图像用于广东肇庆地区的稻田分类试验,结果表明,多时相Radarsat数据对水稻类型的识别精度较高,而且稻田的轮作规律容易推测出来。本文系统地介绍了这一试验研究的最新进展,探讨了神经网络分类方法在SAR图像处理中的应用潜力和Radarsat数据在中国南方水稻监测中的最佳时相选择和有效分辨率问题。     

一种基于浑沌理论的联想记忆神经网络模型

根据联想记忆神经网络的基本原理，提出了一种基于浑沌理论的联想记忆神经网络模型及相应的求解方案，并对该种模型进行了预报试验。结果表明，该种模型的历史预报准确率和实际预报准确率均超过一般随机预报，值得进一步研究和应用。     

神经网络在判别煤矿突水水源中的应用

矿井突水水源的识别是矿井防治水工作的基础，快速准确地判别突水的来源对整个矿井的安全生产起着十分重要的作用。以峰峰矿区梧桐庄矿为例，应用神经网络的方法，对矿井突水水源进行了系统研究。结果表明，运用该方法取得了较好的效果，并可以用此模型来解决类似的评价和判别问题。     

PNN网络在预测MCS移动中的应用

神经网络是空间数据挖掘的一种重要手段。本文运用概率神经网络(PNN网络)对1998年夏季影响青藏高原上中尺度对流系统(MCS)东移的环境物理量场的空间分布特征进行了研究，得到了高原上MCS移动方向与环境物理量场空间分布之间的关系。研究表明，使用概率神经网络预测中尺度对流系统的移动具有较好的效果，从而为研究高原上MCS东移与环境场之间的关系提供了一种新的方法和思路。     

基于神经网络的建筑物沉降原因分析

在我们应用回归方法对建筑物沉降原因分析及沉降趋势的预测中，由于实际情况的复杂性及主观认识的局限性，这样所得的结果含有较多的人为因素，可能会与实际情况有所差异。针对这种情况，讨论了应用神经网络方法来发现和验证引起建筑物沉降的因素及对沉降趋势的预测。实例表明，该方法能取得较好的效果。     

Computation of response spectra from mining tremors using neural networks

Underground coal and copper ore exploitation in two Polish mining regions causes mining tremors and a series of other negative phenomena in the environment. Although these tremors are strictly connected with human activity, they differ considerably from other paraseismic vibrations. The moment of their occurrence is not to be foreseen likewise for earthquakes. The main problem discussed in the paper was formulated as the neural network evaluation of a relation between mining tremor energies, epicentral distances and acceleration response spectra. Back-propagation neural networks with Resilient back-propagation learning method were used. Each input vector included information about the mining tremor energy and the epicentral distance. Values of acceleration response spectrum were expected as the outputs of neural networks. Neurally evaluated spectra were compared with spectra computed on the basis of experimental data. After the network is trained and tested, it can be used for mapping of new data of mining tremor energies and epicentral distances into the spectra. Then, what is the substantial advantage of neural approach, the prediction of acceleration response spectra can be performed without recording of surface vibrations. In the light of the results, it is visible that the presented way of computation of acceleration response spectra can be peculiarly applied to prognosis of mining tremors influences on structures.     

Slope stability evaluation using Back Propagation Neural Networks

The Yudonghe landslide, located in western Hubei Province of China, consists of eastern and western subunits as well as a main landslide mass with upper and lower slip surfaces. As an important landslide close to Shuibuya Dam on the Qing River, its stability is crucial, as the slide might reactivate because of a change in ground-water level caused by filling of the Shuibuya Reservoir. Existing weakness zones, growth of ruptures, the downslope attitude of geologic strata, and water infiltration, which reduced the strength of rocks and soils, have been found to be the most important factors contributing to the Yudonghe landslide. With regard to the landslide processes, it can be noted that the original large-scale slide activity was due to erosion by the Qing River, the second sliding resulted from the fall of blocks from the head scarp, and the final activity was the growth of the eastern and western secondary slides. A base failure was the main type of slope movement, however, it was obvious that more than one sliding event occurred, as inferred from striations and fractures detected by microstructure analysis of soils along the failure surfaces. Slope instability was evaluated by the method of Back Propagation Neural Networks (BPNN), in which a four-layer BPNN model with five input nodes, two hidden layers, and two output nodes was constructed using a training data set of landslide samples throughout the Qing River area. The predicted results of this analysis showed that the factor of safety was 1.10, which indicates that the Yudonghe landslide is currently in a marginally stable condition.     

火山岩储层建模初探

火山岩储层建模研究是一个世界性难题。近年来 ,国内外在含油气盆地的火山岩中发现了大量的油气藏。为了探索火山岩储层物性的空间分布特征 ,作者选取了松辽盆地徐家围子地区火山岩发育的营城组为研究目的层 ,以火山岩相分析为基础 ,以基于相分析的神经网络技术为手段 ,通过多种参数的选取和计算 ,尝试性地建立了该区火山岩储层地质模型。该模型表明 :层状火山机构比块状火山机构的物性好 ;多相火山机构比单相火山机构物性好 ;裂缝发育的火山机构比裂缝不发育的火山机构物性好。基底涌流相、火山空落相、火山沉积相是有利的储集相带。     

人工神经网络方法在鹤壁汛期降水预报中的应用

在用经验统计方法和降水判别函数进行24h和12h晴雨预报的基础上，再用BP人工神经网络建立降水量级预报模型。经2003年汛期试用，预报准确率高于上级指导预报准确率，12h预报准确率高于24h预报准确率。     

极区电离层对流速度的浅层神经网络建模与分析

电离层对流是太阳风与地球磁场相互作用下驱动的磁层大尺度对流循环与对流电场在极区电离层的映射, 与行星际磁场-地球磁场耦合系统息息相关.本文基于SuperDARN(Super Dual Aurora Radar Network)分布在北半球的23部高频相干散射雷达获取到的二维电离层对流速度对其进行建模研究.模型输入为行星际磁场三分量、太阳风速度、太阳风密度和地磁指数六个空间物理参数, 模型输出为二维对流速度.模型选择两种广泛应用于空间物理建模的浅层神经网络即广义回归神经网络(General Regression Neural Network, GRNN)和误差反向传播(Back Propagation, BP)神经网络.实验结果显示, GRNN模型和BP模型的速度幅值均方根误差分别为174.96 m·s^-1和234.21 m·s^-1, 速度方向角均方根误差分别达到62.30°和88.07°, 相比于对流速度最大值2000 m·s^-1和360°的方向角范围来说, 其误差是可以接受的.外推性实验结果显示, 在第24个太阳周期时, GRNN模型和BP模型的速度幅值均方根误差分别为305.35 m·s^-1和738.15 m·s^-1, 速度方向角均方根误差分别为82.01°和90.56°.实验结果表明, GRNN在时间外推性上的效果优于BP神经网络, 更适用于预测对流速度.我们发现在四种典型空间环境条件下, 利用GRNN模型预测的瞬时对流速度来构建的全域对流模式与现有统计模型构建的对流模式相似, 从而验证预测的对流速度可以用于分析瞬时极区电离层对流.