Fuzzy model identification based on cluster estimation for reservoir inflow forecasting

Fuzzy theory appears to be extremely effective at handling dynamic, non‐linear and noisy data, especially when the underlying physical relationships are not fully understood. Since hydrologists are still uncertain about many of the aspects of the physical processes in the watershed, fuzzy theory has proved to be a very attractive tool enabling them to investigate such problems. The effectiveness of the fuzzy model lies in the identification of the antecedent membership function (MF), which is generally addressed through a fuzzy clustering approach. Most of the applications of fuzzy computing in hydrology seem to have selected the clustering algorithm quite arbitrarily. However, it is apparent that, as the antecedent parameters are based solely on the identified clusters, the method used for clustering should certainly have an impact on the overall performance of the model. This paper presents the results of a study conducted to investigate the impact of choice of clustering algorithm on the overall performance of a fuzzy‐based hydrologic model. The research is illustrated through a case study of developing a Takagi–Sugeno fuzzy model for reservoir inflow forecasting in the Narmada basin, India. The model was developed using two popular clustering techniques, namely Gustafson–Kessel (GK) and subtractive clustering (SC), and was extensively evaluated for performance based on various statistical indices. The results show that the model performance is comparable at a 1 h lead forecast. However, it is observed that the GK approach results in a better performance than the SC approach in computing forecasts at higher lead times. The analysis suggest that the GK method clusters the input space based on the actual pattern, since it uses a membership‐grade weighted‐distance measure as the measure of closeness, whereas the SC method classifies the input space more logically according to the magnitude of flow available in the data set. Copyright © 2007 John Wiley & Sons, Ltd.     

基于K-means聚类距离准则的R树结点分配算法研究

对于空间数据库,R树索引是非常有效的空间索引。本文针对R树的结点分配算法存在的不足,提出了一种新的结点分配算法—基于K-means聚类距离最小的R树结点分配算法。研究结果表明,新的分配算法比原始的算法,产生的虚结点的最小约束矩形具有更少的空白区域,较明显地提高了空间查询的效率。     

近40年内蒙古候平均气温变化趋势

基于1964—2003年内蒙古有关台站的逐日气温数据, 对前20年 (1964—1983年) 和后20年 (1984—2003年) 的候平均气温中值进行了对比, 并利用k-means方法进行了候平均气温变化趋势的聚类分析, 旨在阐明内蒙古气温变化的时空特点和规律。研究结果表明:内蒙古全境年平均气温普遍上升, 没有下降的现象。不同地区、不同季节气温变化格局不同, 北部变暖比南部更为明显, 冬季和夏末秋初变暖出现得更频繁。绝大部分地区大多数候平均气温都有所上升, 而且候平均气温变化存在纬向地带性。大多数台站四季里都有一些显著升温的候, 显著升温在冬季发生得更为频繁。同时, 有少数几个候出现气温下降的现象, 但未达到显著下降的程度。内蒙古绝大部分地区候最高、最低气温都有所升高, 但二者的差值在缩小。候最低气温出现时间有所提前, 候最高气温出现时间有提前也有推迟, 但总体上不同地区候最高、最低气温出现时间更加同步。因此, 候最高、最低气温的时空均一性有所增强。候最高气温出现时间与候最低气温出现时间的间隔有扩大迹象。总之, 全球气候变暖降低了气温的季节和地区差异。     

分布式增量机制下的交通流大数据聚类分析

时空聚类分析是对时空大数据进行利用的一种有效手段。本文提出了一种分布式增量大数据聚类分析方法,利用分布增量机制不但可以减少重复计算和迁移拷贝次数,而且可以持续对聚类结果进行修正,能够在保持聚类准确性的条件下提升整体运算效率。而聚类算法本身通过数据聚集趋势预分析、聚类算法和结果评价3个步骤,构建了一体化时空邻域,在时间和空间维度保证了聚类结果的准确性。经过试验证明该方法可以实现时空大数据的快速高效信息挖掘。     

自然灾害调查数据的多尺度异常检测方法研究及应用

大范围自然灾害调查,涉及区域环境差异大,数据获取方式多样,参与人员多,各级汇总成果中存在一些异常调查单元,需要人工判读其合理性,但单纯依靠人工从海量数据中有效识别异常是不现实的。本文设计了一种自然灾害调查数据的多尺度异常检测方法,综合运用离群检测方法和空间数据挖掘算法,分别进行异常值和异常空间分布模式检测,能够从海量调查数据中快速提取各级尺度的异常值和异常调查单元,支撑人工判读工作。将该方法应用于全国山洪灾害调查评价汇总数据的审核中,以全国历史山洪灾害点和防治区乡镇面积审核为例,分别快速提取了县乡两级区划中的山洪灾害点密度异常单元和面积值异常的乡镇单元,通过对检测结果进行分析,发现是填报口径不一致、单位错误、记录重复等原因造成的。最后分析了该方法在大范围自然灾害调查中的适用条件和方法。     

支持向量机在地球化学勘查中的应用

将K均值聚类和v -SVM结合, 推出一种改进的分类方法。该方法根据指标的相关性, 构建了具有强分辨能力的分类模型, 并将模型应用于内蒙某地勘查地球化学研究, 查明了所研究元素之间的相似性, 建立了v - SVM分类机, 实现了组合异常的圈定。     

应用于遥感图像分割的原型提取谱聚类集成算法

针对遥感图像数据量大、类别归属复杂的特点,提出了一种用于遥感图像分割的原型提取谱聚类算法。该算法首先采用广义模糊c-均值聚类算法对遥感图像进行过分割,将得到的聚类中心作为每个分割区域的代表点;然后,通过构造代表点之间的相似性矩阵,利用谱图划分方法对代表点进行聚类;最后,根据代表点的聚类结果对图像像素点进行重新归类来获得遥感图像的最终分割结果。此算法涉及到3个参数,为了克服算法对于参数的敏感性和内在的随机性,进一步引入集成策略,给出了原型提取谱聚类的集成算法。     

基于EM及熵值理论的区域经济与影响因素研究

基于EM算法的聚类方法属于软聚类范畴,在应用中具有较高的灵活性,而熵值理论可以较好地反映出分子对系统凝聚状态的影响度。以河南省为例应用EM聚类方法分析了各地市的经济发展水平,运用熵值理论对影响经济发展的各因素进行了评价,并计算得出各因素的影响度,以此探讨河南省各地市经济发展不平衡的特点和原因,并为经济的全面协调发展提供了对策和思路。     

Fractal Dimension of the 1999 Chamoli Earthquake from Aftershock Studies in Garhwal Himalaya

The Aftershock sequence of Chamoli earthquake (M _w 6.4) of 29 March 1999 is analyzed to study the fractal structure in space, time and magnitude distribution. The b value is found to be 0.63 less than which is usually observed worldwide and in the Himalayas. This indicates that the numbers of smaller earthquakes are relatively less than the larger ones. The spatial correlation is 1.64, indicating that events are approaching a two-dimensional region meaning that the aftershocks are uniformly distributed along the trend of the aftershock zone. Temporal correlation is 0.86 for aftershocks of M 1, indicating a nearly continuous aftershock activity. However, it is 0.5 for aftershocks of M 1.75, indicating a non continuous aftershock activity. From the assessment of slip on different faults it is inferred that 70% displacement is accommodated on the primary fault and the remainder on secondary faults.     

Global multiresolution models of surface wave propagation: comparing equivalently regularized Born and ray theoretical solutions

I invert a large set of teleseismic phase-anomaly observations, to derive tomographic maps of fundamental-mode surface wave phase velocity, first via ray theory, then accounting for finite-frequency effects through scattering theory, in the far-field approximation and neglecting mode coupling. I make use of a multiple-resolution pixel parametrization which, in the assumption of sufficient data coverage, should be adequate to represent strongly oscillatory Fréchet kernels. The parametrization is finer over North America, a region particularly well covered by the data. For each surface-wave mode where phase-anomaly observations are available, I derive a wide spectrum of plausible, differently damped solutions; I then conduct a trade-off analysis, and select as optimal solution model the one associated with the point of maximum curvature on the trade-off curve. I repeat this exercise in both theoretical frameworks, to find that selected scattering and ray theoretical phase-velocity maps are coincident in pattern, and differ only slightly in amplitude.