基于云推理的年降水量预测

预测未来年的降水量是对旱情做出预警并采取合理抗旱措施的重要前提,但如何获得准确的预测值一直是研究的热点。降水随时间的变化通常具有随机性和模糊性,而云模型是在传统模糊数学和概率统计的基础上建立起来的,能够实现不确定概念与定量数值之间的自然转化,通过年降水量历史数据及当前趋势挖掘并制定出相应规则,从而推理获得未来年份的降水量。在此基础上,提出了基于小波消噪预处理和理论频率曲线修正的云推理预测模型,并将其运用到了西安市年降水量的预测当中。从所得结果来看,降水量预测值能较好的反映其年际变化规律,模型预测精度得到了很大提高,基本可将误差控制在30%以内。     

An evidential reasoning-based sustainability index for water resources management

ABSTRACT

Evidential reasoning (ER) is introduced as a new basis for developing and implementing sustainability indices (SI). The ER-based evaluations keep the merits of previous well-known sustainability indices and provide added values such as more accurate grade-based assessment and aggregation of performance criteria. The proposed modifications significantly improve the capability of the SI to scrutinize and differentiate various water-supply conditions. Available data from Gorganrud-Qaresou basin, northern Iran, are used to evaluate the applicability and efficiency of the sustainability index. The results of the proposed method are compared with two water resources sustainability indices with similar concepts but different bases. It is shown that the ER-based SI can better distinguish undesirable water-supply scenarios.     

空间目标的方向关系定性推理

由于空间问题固有的复杂性和不确定性,空间关系的描述和推理普遍采用定性的方法来符合人们的认知。方向关系是一类重要的空间关系,以往对方向关系的定性推理大多在2个参照系下进行,将目标对象抽象为点或者认为目标的大小近似相等。为了更细致地推理方向关系,提出四种不同情况下基于1个参照系的定性方向关系推理方法,并就本方法存在的问题和今后的研究方向提出了一些看法和思路。     

中文文本中时间信息解析方法

探讨了基于触发词汇和规则模型相结合的中文文本中时间信息解析方法。通过分析、归纳中文文本中时间信息描述特点,构建时间词汇词典和时间信息描述模式库,设计时间信息抽取、规范化表达和语义推理算法,实现了中文文本中时间信息的解析。实验结果表明,中文文本中时间信息抽取的准确率、召回率和F1值分别为75.00%、88.24%和40.54%,为泛在时空信息动态关联更新和实时挖掘分析提供数据源,且通过与空间维数据有机地、交互地组织,能够实时展现地理现象和事物的时空演化过程、时空分布特征,从而推动地理信息检索、LBS等地理信息服务向动态化、多维化方向发展。     

双投影矩阵模型的方向关系组合推理研究

将三维空间中的参照物体分别向xoy和xoz平面投影,得到的最小外接矩形将两个平面都划分成9个区域,用两个三行三列矩阵联合表示目标物体投影与两个参考区域的相交情况,据此提出了双投影方向关系矩阵模型。基于该模型,利用矩阵的计算性能,提出了一种三维空间方向关系的组合推理方法,解决了方向关系的组合推理问题。     

NEW PUBLICATIONS

Ten Years o f Physicogeographic Research in Hungary

Antarctic Snow and Ice Studies     

A case-based reasoning approach for task-driven spatial–temporally aware geospatial data discovery through geoportals

ABSTRACT

There is a critical need to develop a means for fast, task-driven discovery of geospatial data found in geoportals. Existing geoportals, however, only provide metadata-based means for discovery, with little support for task-driven discovery, especially when considering spatial–temporal awareness. To address this gap, this paper presents a Case-Based Reasoning-supported Geospatial Data Discovery (CBR-GDD) method and implementation that accesses geospatial data by tasks. The advantages of the CBR-GDD approach is that it builds an analogue reasoning process that provides an internal mechanism bridging tasks and geospatial data with spatial–temporal awareness, thus providing solutions based on past tasks. The CBR-GDD approach includes a set of algorithms that were successfully implemented via three components as an extension of geoportals: ontology-enhanced knowledge base, similarity assessment model, and case retrieval nets. A set of experiments and case studies validate the CBR-GDD approach and application, and demonstrate its efficiency.     

An algebraic approach to automated geospatial information fusion

This paper presents a new technique for information fusion. Unlike most previous work on information fusion, this paper explores the use of instance‐level (extensional) information within the fusion process. This paper proposes an algorithm that can be used automatically to infer the schema‐level structure necessary for information fusion from instance‐level information. The approach is illustrated using the example of geospatial land‐cover data. The method is then extended to operate under uncertainty, such as in cases where the data are inaccurate or imprecise. The paper describes the implementation of the fusion method within a software prototype. Finally, the paper discusses several key topics for future research, including applications of this work to spatial‐data mining and the semantic web.     

基于范例推理的公路隧道拱顶变形时序支持向量机外延预测

针对公路隧道拱顶变形预测模型的普适性与外推预测的准确性,提出了基于人工智能推理的隧道工程属性（地理位置、监测位置、隧道高宽比、围岩级别和埋深）与拱顶变形时序曲线原子矩阵的相似范例检索方法,并在深入分析了获取的相似范例特征的基础上,进一步以LPG新核函数支持向量机建立先验知识的预测模型。应用该方法对通渝隧道工程K19+994断面拱顶下沉进行了预测与评估。结果表明,对于不同隧道间或同一隧道不同区段预判拱顶变形或收敛,基于范例推理能够获知良好的先验背景知识,且以此进行的支持向量机预测模型学习的回归内插（1～14步序）的平均相对误差为1.36%,而一次性外推预测15 d内的8个变形值（16～30步序）的平均相对精度为97.28%,证实了方法的可靠性