基于PSO-SVM的我国新型城镇化发展水平评价

发布时间：2018-03-18 00:33

本文选题：新型城镇化　切入点：PSO-SVM　出处：《安徽建筑大学》2015年硕士论文　论文类型：学位论文

【摘要】：本文以新型城镇化内涵及发展模式为理论基础,以支持向量机及粒子群算法为方法指导,结合定性同定量分析、规范研究与实证研究,实现对我国新型城镇化发展水平的评价分析,并根据实证研究的结果将全国河北、山西、内蒙古等27个省份划分为四大区域进行横向比较27个省份的新型城镇化发展现状,进而得出推动我国新型城镇化发展及加快我国新型城镇化建设的对策及建议。第一章首先分析了论文的选题背景和研究的目的与意义,分析了推动新型城镇化发展对我国整体发展的意义之所在,其次根据时间脉络对有关新型城镇化内涵及发展模式的国内外研究进行综述,再次介绍了本论题所运用的研究方法以及创新之处,最后以技术路线图的形式展现了本文的研究思路。第二章分别阐述并分析了我国新型城镇化发展水平评价指标方法和我国新型城镇化发展水平测度方法的现状。通过对评价指标方法的评述,总结出我国新型城镇化相关指标体系的问题,为后续章节相关指标体系的构建奠定一定现实基础,并分析了当前新型城镇化相关测度方法的不足,简略阐述了粒子群算法优化支持向量机评价模型的优越性。第三章在严格遵循指标体系建立基本原则的基础上,参考现今国内学者对新型城镇化相关评价研究的成果,对源自全国河北、山西、内蒙古等27个省份2012年度总数达1600多个指标数据进行标准化预处理,并将SPSS18.0应用于这些标准化后的数据,由此输出显著性与相关性矩阵,并根据该矩阵对初选指标进行指标筛选,进而构建了我国新型城镇化发展水平评价的指标体系。第四章首先描述了我国新型城镇化发展现状,并分开阐述了支持向量机同粒子群算法的相关原理并叙述了PSO-SVM的运行流程,意在强调PSO-SVM方法的科学性及合理性。其次,基于粒子群算法优化支持向量机对我国新型城镇化发展水平评价问题进行实证研究。本章根据第三章所确定的评价方法以及评价指标体系,利用河北、山西、内蒙古等27个省份的筛选后的指标数据,在Matlab平台上建立了基于PSO-SVM的我国新型城镇化发展水平评价模型,利用相关程序语言实现运行。程序运行结果显示,均方误差是0.00565827;与综合评价值曲线的拟合相关系数是99.7104%,这说明回归结果具有一定科学性,通过PSO优化SVM所得到的最佳惩罚因子及核函数分别为10.0179和0.0527。最后,本章根据仿真结果对我国27个省份的新型城镇化发展水平进行了划区域评价及结果分析,并有针对性的提出相应的对策及建议,意在为我国27个省份持续推动新型城镇化建设提供一定建设性的建议。第五章首先对全文的研究进行总结,得出本论题的研究结论,相关结论如下:(1)东部地区新型城镇化发展水平较高,优势较为明显;(2)中部地区新型城镇化发展水平相对滞后;(3)西部地区新型城镇化发展水平呈现阶梯分布。其次,对本论题未来进一步的研究方向从评价指标体系及测度方法两个方面做出了展望。
[Abstract]:In this paper, and the development of new urbanization connotation model as the theoretical basis, based on the support vector machine and particle swarm algorithm as the guidance of methodology, combining qualitative analysis with quantitative analysis, normative research and positive research, analysis and evaluation of the level of development of new towns in China, and according to the empirical research results will be the Hebei, Shanxi, Inner Mongolia the 27 provinces are divided into the current situation of the development of the new urbanization comparing the four regions of 27 provinces, then promote the development of China's new urbanization and the countermeasures and suggestions about accelerating the construction of the new urbanization of our country. The first chapter analyzes the purpose and significance of the background of the research and analysis, to promote the development of the new urbanization of China's overall development significance, secondly according to the time sequence of the relevant research on new urbanization connotation and mode of development at home and abroad are reviewed, introduced again Research methods used in this thesis and innovation, the technology roadmap reflects the idea of this paper. The second chapter describes and analyzes the status quo of China's new urbanization development level evaluation method and method for the development of China's new urbanization level measure. Through the analysis of evaluation index method, summary a relevant index system of the new towns in China, lay a realistic basis for the following chapters to construct relevant index system, and analysis of the current shortage of relevant measurement methods of the new town, explains the advantages of particle swarm optimization support vector machine evaluation model. In the third chapter, based on follow the establishment of index system of basic on the principle of reference to today's domestic scholars research results related to evaluation of the new town of Hebei, from the national, Shanxi, Inner Mongolia and other 27 provinces 201 The year 2 a total of more than 1600 data standardization processing, and the application of SPSS18.0 in the standardized data, the resulting output significantly and correlation matrix, and according to the matrix of the primary index of index selection, and then constructed the evaluation index system of the development level of the new urbanization in China. The fourth chapter describes the current development of new urbanization in China, and expounds the principle of separate support vector machine with particle swarm algorithm and describes the operation process of PSO-SVM, the PSO-SVM method is intended to emphasize the scientificity and rationality. Secondly, the particle swarm optimization support vector machine development level evaluation of new towns in China, empirical research based on this chapter. The third chapter according to the evaluation method and evaluation index system, the use of Hebei, Shanxi, Inner Mongolia and other screening index data of 27 provinces in the Matl AB platform is established on the level of development of China's new urbanization evaluation model based on PSO-SVM, realize the operation of using programming language. The program running results show that the mean square error is 0.00565827; and the comprehensive evaluation value curve fitting correlation coefficient is 99.7104%, which shows that the regression result is scientific, through optimal penalty factor and kernel PSO function optimization of SVM are 10.0179 and 0.0527. respectively at the end of this chapter, according to the simulation results of new towns in 27 provinces of our country the development level of the designated area evaluation and result analysis, and put forward the corresponding countermeasures and suggestions, in order to provide some constructive for the 27 provinces in our country continuously promote the construction of new urbanization. The fifth chapter summarizes the research conclusion, the research of this thesis, the related conclusions are as follows: (1) the new urbanization development of the eastern region The level is higher, the more obvious advantages; (2) new towns in the central region development is lagging behind the level; (3) the new urbanization of western region development shows a ladder distribution. Secondly, the further research direction of the topic of the future from two aspects of the evaluation index system and measure method is prospected.

【学位授予单位】：安徽建筑大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F299.21;F224

【参考文献】