论文部分内容阅读
栅格计算因其具有简单的构架成为目前地学分析的主流模型,然而,由于栅格计算平均分配计算和存储资源的弱点,不仅容易产生冗余,更重要的是难以凸显研究对象的突变部分,从而使研究者有可能忽略地学现象的变化特征。为此,本文提出将时空点过程模型应用于地学研究。时空点过程不仅适用于模拟以点事件为基本单元的地学现象,而且由于大多数地学过程可以转化为时空点过程,故其具有更广泛的应用范围。因此,时空点过程不仅是一种数据模型,同时也是地学问题的分析方法,更是观察和理解地学问题的一种新视角。为了实现从点过程数据中提取模式,作者经过多年研究提出了时空点过程层次分解理论框架,该理论与信号处理理论中的谱分析思路类似,首先,假设任意点集为有限多个均匀点过程的叠加,然后,通过点局部密度表达工具K阶邻近距离,将空间点转换为混合概率密度函数,再应用优化方法将混合密度函数进行分解得到丛集点和噪声,最终利用密度相连原理从丛集点中提取模式。该理论框架可适用于绝大多数点集数据,初步实现了点集数据的“傅里叶变换”。
Grid computing has become a mainstream model for geoscientific analysis because of its simple architecture. However, it is not only easy to generate redundancy, but more importantly, it is difficult to highlight the mutation part of the research object because of the average distribution of grid computing and the weakness of storage resources. Thus making it possible for researchers to ignore the changing nature of the phenomenon of learning. Therefore, this paper proposes to apply the spatio-temporal point model to geosciences research. The space-time point process is not only suitable for simulating the geo-learning phenomenon based on point events, but also has a wider range of applications because most of the geo-learning processes can be transformed into space-time point processes. Therefore, the time-space point process is not only a data model, but also a method of analyzing the problem of geosciences, which is also a new perspective to observe and understand the problem of geosciences. In order to extract the patterns from the point process data, the author puts forward the theoretical framework of hierarchical decomposition of the space-time point process after years of research. This theory is similar to the spectral analysis idea in the signal processing theory. First, assume that any set of points is a finite number of uniform point processes Then, the spatial point is transformed into a mixed probability density function by using the point-local density expression tool K-order proximate distance, and then the hybrid density function is decomposed to obtain cluster points and noises using the optimization method. Finally, In the extraction mode. The theoretical framework can be applied to the vast majority of point set data, the initial realization of the “Fourier transform ” of point set data.