基于非物质文化数据选取的CA_Markov模型精度改进研究

发布时间：2018-06-17 09:36

本文选题：CA_Markov + 时间序列分析　；参考：《河北师范大学》2013年硕士论文

【摘要】：CA_Markov模型是CA模型与Markov链模型的结合体，是一种基于栅格运算的定量分析模型。在CA_Markov模型的运用中涉及到数据的三种特征，分别是时间特征、空间特征和属性特征。在现有研究中，研究者通过人为改变空间特征和属性特征的方法达到提高CA_Markov模型模拟精度的目的，没有充分考虑到时间特征对CA_Markov模型模拟精度的影响。随着遥感影像等数据源的不断丰富，研究者已具备挑选基础数据的条件。现有研究中CA_Markov模型的模拟数据一般只有二到五期，并以其中一期为检验数据进行模拟精度评价，研究过程缺少对模拟数据的筛选过程。本文通过将时间序列分析模型与CA_Markov模型相结合，从分析和运用数据的时间特征角度入手，以蔚县剪纸艺人分布变化研究为例，分析文化景观分布空间随时间的变化规律，并将结果运用到CA_Markov模型模拟运算中。主要结论如下：蔚县剪纸文化发展具有明显的时间特征，著名艺人出现、重要历史政治事件等是蔚县剪纸历史发展阶段的时间分割点。蔚县剪纸文化发展的每个阶段都有各自的发展特点，从1834年至1930年的96年中，知名艺人数由1人增加为35人，年增长率为35.42%。有记载的艺人村庄数由1个增加为11个，年增长率为10.42%，这一阶段是蔚县剪纸艺术的缓慢发展时期。从1931年至1966年的35年中，艺人数由35人增加为78人，，年增长率为119.44%。有记载的艺人村庄数由11个增加为28个，年增长率为47.22%，这一阶段是蔚县剪纸的快速发展时期。从1967年至2006年的39年中，其知名艺人数量艺人数由78人减少为30人，年增长率为-120.00%。有记载的艺人村庄数由28个减少为10个，年增长率为-45.00%，这一阶段是文革衰落期和文革后产业化转型期。蔚县剪纸文化的传播在空间分布上形成明显的剪纸生产区，区内以中短距离为主，传播区域沿壶流河两岸扩散。1889年至1909年属于蔚县剪纸缓慢增长期，形成了县城—南张庄和李家庄双中心发展模式，剪纸技术在两个生产区均为短距离传播。整体上，村庄数量由1889年的5个增加到1909年的9个，数量增加80%，年增长率为20%，村庄分布由蔚县中部向东北和西南短距离扩散。1909年至1936年属于蔚县剪纸由缓慢增长期向快速增长期发展的过渡阶段。整体上，村庄数量由1909年的9个增加到1936年的14个，数量增加55.56%，年增长率为18.52%，村庄分布除由蔚县中部向东北和西南短距离扩散外亦伴有南北方向扩散。1943年至1963年属于蔚县剪纸快速增长期，这一时期新增了位于蔚县东北部的冀家嘴生产区。整体上，村庄数量由1943年的16个增长为1963年的29个，数量增加81.25%，增长率为65%，村庄分布由蔚县中部向东北和西南短距离扩散显著，南北方向亦有明显扩散。蔚县剪纸技艺的传播具有派系特征，每个派系都有其自身的文化传播特点。县城—南张庄生产区是王老赏派系的主要传播区，艺人分布相对集中，该区域传播距离以短距离为主，中短距离为辅，因此作者将其归类为中短距离传播区。李家庄生产区是温新春派系的主要传播区之一，温新春派系注重空间扩散传播，区内以短距离传播为主兼有中短距离传播。本文构建TS_CA_Markov（即时间序列分析_元胞自动机_马尔科夫链）复合模型，并给出了可行的数据筛选和模拟运算流程：构建蔚县艺人时空数据库→相关数据选取→时间序列分析→制定模拟规则→模拟运算。本文以蔚县剪纸艺人研究为例，选取艺人数量时间序列数据作为蔚县艺人时空分布时间序列数据的相关数据，对艺人数量时间序列数据进行时间序列分析，选择残差较小数据作为实验组数据，残差较大数据作为对比组数据。利用专家打分的方法，从艺人故事、从艺时长和所获荣誉三个方面对蔚县剪纸艺人影响力进行评价，制定新的模拟规则，并应用于CA_Markov模型的模拟运算中。实验结果显示对比组预测正确村庄数为9个左右，正确率达到64.29%，实验证明CA_Markov模型适用于对文化景观模拟。实验组预测正确村庄数为12个左右，正确率达到85.71%。经假设检验实验组模拟精度显著优于对比组，证明通过模拟数据的筛选可以显著提高CA_Markov模型模拟精度。实验同时证明了标准残差是模拟数据选取的关键指标。
[Abstract]:The CA_Markov model is the combination of the CA model and the Markov chain model. It is a quantitative analysis model based on the grid operation. In the application of the CA_Markov model, it involves three characteristics of the data, which are time characteristics, spatial characteristics and attribute characteristics. In the present research, the researchers have passed the method of changing the spatial and attribute characteristics by the researchers. In order to improve the accuracy of CA_Markov model, the effect of time characteristics on the simulation precision of CA_Markov model is not fully taken into account. With the continuous enrichment of data sources such as remote sensing images, the researchers have had the conditions to select the basic data. In the present study, the model data of the CA_Markov model are usually only two to five periods, and one of them is one of them. In order to evaluate the accuracy of the test data, the process of screening the simulated data is lacking. By combining the time series analysis model with the CA_Markov model, this paper starts with the analysis and application of the time characteristics of the data, and analyzes the distribution of cultural landscape in Weixian County as an example, and analyzes the spatial distribution of cultural landscape with time. The results are applied to CA_Markov model simulation. The main conclusions are as follows:
The development of Weixian County paper-cut culture has obvious time characteristics. The appearance of famous artists and important historical and political events are the time dividing points in the development stage of Weixian County paper-cut. Each stage of the development of Weixian County paper-cut culture has its own development characteristics. From 1834 to 1930, the number of famous artists increased from 1 to 35, and the annual growth rate was increased. The number of artist villages recorded for 35.42%. has increased from 1 to 11, with an annual growth rate of 10.42%. This stage is a period of slow development in Weixian County paper-cut art. From 1931 to 1966, the number of artists increased from 35 to 78, and the annual growth rate of 119.44%. has increased from 11 to 28, with an annual growth rate of 47.22%. One stage is the rapid development period of paper cutting in Weixian County. From 1967 to 2006, the number of famous artists was reduced from 78 to 30 in the 39 years. The annual growth rate of recorded artists was reduced from 28 to 10, and the annual growth rate was -45.00%. This stage was the period of the decline of the cultural revolution and the industrial transformation period after the cultural revolution.
The spread of Weixian County paper-cut culture forms an obvious paper cutting production area in the space distribution. The area is mainly medium and short distance in the area. The spread region spread along the two sides of the Juliu River from.1889 to 1909, which belongs to the slow growth period of Weixian County paper cutting, forming a dual center development model of the county town south Zhang Zhuang and Li Jia Zhuang. The paper cutting technology is short distance in the two production areas. On the whole, the number of villages increased from 5 in 1889 to 9 in 1909. The number of villages increased by 80%, the annual growth rate was 20%. The distribution of villages from central Weixian County to northeast and southwest from.1909 to 1936 was the transitional stage of Weixian County paper-cut from slow growth to rapid growth period. On the whole, the number of villages was 9 in 1909. The increase to 14 in 1936 has increased by 55.56%, and the annual growth rate is 18.52%. The distribution of villages is also accompanied by north-east and south-west diffusion in the middle of Weixian County, which is also accompanied by north-south diffusion from.1943 to 1963, which belongs to the rapid growth period of Weixian County paper-cut. In this period, the jizjia mouth production area in the eastern North of Weixian County was added. From 16 in 1943 to 29 in 1963, the number increased by 81.25%, the growth rate was 65%. The distribution of villages from central Weixian County to northeast and south-west was significant, and the north and south direction also had obvious diffusion.
The spread of paper cutting techniques in Weixian County has factional features, each of which has its own cultural transmission characteristics. The county town - Nanchang Zhuang production area is the main dissemination area of Wang Lao's faction, the distribution of artists is relatively concentrated, the distance of the region is mainly short distance and the short distance is supplemented, so the author classifies it as a medium and short distance communication area. Li's family. The Zhuang production area is one of the main transmission areas of the new spring faction, and the warm and new spring factions pay attention to the spread of space diffusion, and the short distance transmission is the main part of the region.
In this paper, the composite model of TS_CA_Markov (temporal sequence analysis cellular automaton Markov chain) is constructed, and the feasible data selection and simulation operation flow are given: the construction of the spatio-temporal database of the Weixian County artist, the selection of related data, the time series analysis, the formulation of the simulation rules and the simulation operation.
Taking the study of Weixian County paper-cut artist as an example, this paper selects the time sequence data of the artist quantity as the data of time series data of Weixian County artists' time and space distribution, analyzes the time series of the time series data of the artist, chooses the small residual data as the experimental group data, and the large data of the residual difference as the contrast group data. The method is to evaluate the influence of the Weixian County paper-cut artists from three aspects of the artist's story, the length of the art and the honor. The new simulation rules are formulated and applied to the simulation of the CA_Markov model.
The experimental results show that the number of correct villages in the contrast group is about 9 and the correct rate is 64.29%. The experiment proves that the CA_Markov model is suitable for the simulation of cultural landscape. The number of correct villages in the experimental group is about 12. The accuracy of the experimental group is up to 85.71%. and the simulation accuracy of the experimental group is better than that of the contrast group. The simulation accuracy of the CA_Markov model can be significantly improved. The experimental results also show that the standard residuals are the key indicators for the simulation data selection.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P208

【参考文献】