Powerbuilder在开发森林资源管理数据库系统中的图形处理

讨论了利用Powerbuilder开发森林资源数据库的好处,及如何保存有关森林资源图形数据,并给出了具体方法.     

Network thinking in riverscape conservation – A graph-based approach

Graph theoretic approaches have received increased interest recently in landscape planning and conservation in the terrestrial realm, because these approaches facilitate the effective modelling of connectivity among habitats. We examined whether basic principles of graph theory can be extended to other ecosystems. Specifically, we demonstrate how a network-based context can be used for enhancing the more effective conservation of riverine systems. We first show how to use graph theoretic techniques to model riverscapes at the segment level. Then we use a real stream network (Zagyva river basin, Hungary) to examine the topological importance of segments in maintaining riverscape connectivity, using betweenness centrality, a commonly used network measure. Using the undirected graph model of this riverscape, we then prioritize segments for conservation purpose. We examine the value of each of the 93 segments present in the Zagyva river basin by considering the conservation value of local fish assemblages, connectivity and the size of the habitat patches. For this purpose we use the ‘integral index of connectivity’, a recently advocated habitat availability index. Based on the results the selection of the most valuable habitat segments can be optimized depending on conservation resources. Because of their inherent advantage in the consideration of connectivity relationships, we suggest that network analyses offer a simple, yet effective tool for searching for key segments (or junctions) in riverscapes for conservation and environmental management. Further, although the joint consideration of aquatic and terrestrial networks is challenging, the extension of network analyses to freshwater systems may facilitate the more effective selection of priority areas for conservation in continental areas.     

基于神经网络的电控发动机怠速不稳定的故障识别方法

以电控发动机怠速不稳定这种十分常见的故障为研究对象。系统地介绍了应用神经网络进行故障原因识别的方法。详细地分析了不同网络结构的训练效率和以不同样本形式训练的网络对相同故障症兆的故障原因的识别能力。     

人工神经网络研究概述

论述了人工神经网络的概念,特点,重要性及学习训练方法,并指出了神经网络的发展现状。     

基于伽玛能谱数据土壤肥力因子空间分布预测初探

目的为了探索伽玛能谱数据实现土壤表层（0 ~ 30 cm）肥力空间分布预测。方法使用偏最小二乘回归（PLSR）、支持向量机（SVM）和反向传播神经网络（BPNN）分别建立土壤伽玛能谱肥力因子模型。结果BPNN模型土壤肥力因子预测精度整体要优于PLSR模型和SVM模型;土壤全氮、pH、黏粒和砂粒含量BP神经网络模型预测精度较高,决定系数R²分别为0.564、0.556、0.612和0.626,全钾和全磷含量预测精度较低;研究区土壤全氮、pH、黏粒和砂粒预测空间分布与样本点实际空间分布相比,数值统计特征和趋势均基本一致。结论研究区伽玛能谱数据预测土壤全氮、pH、黏粒和砂粒含量空间分布具有一定的可行性,全磷、全钾和粉粒含量无法实现有效预测。     

基于粗糙集与神经网络的农机化发展水平评估方法

在分析现有农业机械化发展水平评价方法的基础上 ,提出了粗糙集理论与人工神经网络技术相结合的农业机械化发展水平评价指标简化体系和评价方法。实例表明 ,约简后的指标体系可以获得原有指标体系的分类结果 ,但比原有模型少了 6 0 %的指标 ,大大减少了信息的需求量 ,为农机化水平评价提供了一种新的途径。     

肉种鸡神经型大肠杆菌病病原的分离,鉴定及防治研究

本研究从病原学角度证实了神经型大肠杆菌（血清型Ｏ７８、Ｏ１３３）在河北省的存在。１９９７年春河北省石家庄某大型肉鸡饲养场的肉种鸡发生一种传染病，经充行病学调查，病理剖检、病原分离及鉴定等，初步诊断为大肠杆菌病。同时从病死鸡的脑膜分离到典型的大肠杆菌，人工回归雉鸡体试验，典型发病。最后确诊为肉种鸡神经型大肠菌病。经血清学凝集试验鉴定分离菌为Ｏ７８血清型大肠杆菌和Ｏ１３３血清型大肠杆菌，将分离菌灭活后