人工神经网络构造经济预测模型方法的探讨

探讨利用人工神经网络建立经济模型的一般原理和方法,并通过构造一个宏观经济模型的实例加以说明.与传统的回归分析方法相比,人工神经网络具有良好的适应性和分析预测能力.     

基于BP神经网络预报的动态矩阵预测控制

提出了一种基于BP神经网络预报的动态矩阵预测控制新算法 ,在该算法中 ,先用BP神经网络辨识对象模型 ,同时预测对象的未来输出 ,然后用动态矩阵控制算法进行滚动优化和反馈校正。该方法解决了非线性、时变对象难以建模及控制的问题 ,仿真结果验证了这一新型算法的可行性     

A multi-layer zone model for predicting temperature distribution in a fire room

Modelingoffireinacompartmentcanbeachievedeitherusingazonemodelingmethodoracomputationalfluiddynamics (CFD )modelingmethod .Themostcommonzonemodelisthetwo zonemodel.Themaincharacterofthismodelisthatitdividestheroom (s)intoahot ,upperlayerandacooler,lowerlayer ,andthatthephysicalpropertiesofeachlayer ,suchasitsgastemperatureandspeciesconcentrationsarealluniform .Somemodelsofthistypehavebeendeveloped[1～3] ,andacomprehensivereviewoftheexistingtwo zonefiremodelscanbefoundinreference [4 ].Thecomp…     

Approaches to prediction of impact of Qinghai-Tibet Railway construction on alpine ecosystems alongside and its recovery

With the aid of the Remote Sensing (RS) and Geographic Information System (GIS) technology, the ecosystem pattern and fragility distribution maps of the 50-km-wide zone along the Qinghai-Tibet Railway were compiled and by using the superimposition method, range, area and indexes of the impact of various engineering activities on the ecosystems alongside the railway were studied. By making reference to the ecosystem recovery process of the Qinghai-Tibet Highway, mechanisms of recovery of the alpine ecosystems alongside the Qinghai-Tibet Railway were studied and extents and rates of the recovery were predicted.The results indicate that the impact of the railway engineering on the Alpine ecosystem depends mainly on how much the original surface soil in the zone has been disturbed and how fragile of the ecosystem per se. Restoration of vegetation coverage and species abundance shows a significantly reverse relationship with disturbance of the original surface soil but an extremely positive one with the length of the restoration period and mean annual precipitation and annual mean relative humidity in the period and no obvious bearings with altitude and temperature. In sections with an annual precipitation over 200mm, as long as a certain percentage of original soil is left in situ, it takes only 30 years or so for biodiversity to get basically restored to the original level after the construction is completed but at least 45-60 years or more for vegetation coverage.     

基于小波变换和ANFIS模型的不规则海浪组合预报

利用小波变换对不规则海浪进行多尺度一维小波分解与重构,得到相对简单、规则的准周期分量信号,然后建立了各信号的多输入、单输出自适应神经模糊推理系统(ANFIS)预报模型,最后对预报结果进行集成。基于Matlab语言的仿真结果表明,该方法不仅降低了预报的难度,而且具有较高的预报精确度。此方法亦可用于船舶横摇、纵摇、艏摇运动的预报。     

一种基于组合RBF网络及混沌理论的弱目标检测算法

针对复杂背景中的弱目标检测问题，提出了一种基于组合RBF网络及混沌理论的检测算法。该算法具有结构简单的优点。文中对算法的可实现性及合理性进行了理论分析，建立了基于预测误差优化的非线性预测模型，并结合闽值化处理实现了对弱目标的检测。在理论分析的基础上，对所提出的算法进行了仿真，结果表明了检测算法的有效性。     

甘肃黄土地区地震动持续时间预测研究

在阐述地震动持时的含义及特征的基础上，分析研究了黄土地区与非黄土地区地震动衰减的差异，建立了甘肃省黄土地区地震动衰减规律，提出一种解决与给定超越概率水平相匹配的地震动持时的概率预测方法，给出了甘肃省黄土主要分布地区50年超越概率10％的地震动持时区划结果。     

分布式虚拟环境中的碰撞一致性

讨论了分布式虚拟环境中的碰撞一致性问题，分析了其表现形式及产生原因。提出了一种基于预测的可以保障分布式虚拟环境中碰撞一致性的控制方法。最后讨论了该方法在分布式虚拟环境DVNE中的应用。     

织物透湿性能测试与MATLAR应用研究

对各种织物透湿性能进行测试与分析．利用MATLAB6．5建立神经网络模型，选择最佳网络参数并对网络进行训练．通过应用实例提出了如何处理实验数据的方法，并通过建立大量的BP网络进行比较，筛选出最具应用价值的网络对透湿性能进行预测．     

基于RBF神经网络的跳频信号频率预测的研究

提出了一种跳频信号频率预测的解决方案——RBF神经网络．比较了BP神经网络和RBF神经网络在跳频信号频率预测中的性能差异．应用MATLAB针对一伴有随机噪声的信号进行了仿真实验，证明了该理论的可行性与优越性。