基于粒子群算法和支持向量机的中心城市承载力预测研究

发布时间：2018-02-04 01:28

本文关键词： 城市承载力相空间重构支持向量机粒子群算法　出处：《浙江工商大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着城市化进程的加快和社会经济快速发展,很多城市盲目发展和建设,城市资源利用效率低下,导致土地浪费现象突出、淡水和能源资源紧张、环境污染严重、生态系统严重退化、交通拥堵日益严重等城市危机险象频生。而城市承载力的大小严重关系到城市能否持续健康的发展,人们能否享受更高的生活质量。因此,人们迫切希望寻求一种切实有效的模型对城市承载力进行预测,进而对人们的生活和政府的规划进行有效的指导。由此本文将城市水资源、土地资源、交通和大气环境作为研究的重点对象,结合统计年鉴和水资源公报等发掘已有的城市数据,分别构建城市水资源承载力评价指标、土地资源承载力评价指标、交通承载力评价指标、大气环境承载力评价指标和综合承载力评价指标,以期为解决城市在快速发展的同时衍生出的诸多问题提供方向。此外,为了较全面展现我国中心城市目前的城市承载力发展状况,本文基于选取的评价指标,选择了京津冀、长三角、珠三角等重要区域及直辖市、省会城市和计划单列市共64个城市来研究其承载力现状。研究发现:2014年度,我国城市综合承载力普遍较差。具体来说,逾50%的城市水资源承载力处于预警甚至危机的状态;逾56%的城市的土地资源承载力处于危机状态;80%左右的城市交通承载力极其脆弱;空气质量达优的城市仅有一个,占1.56%。这一切都表明:我国城市的城市承载力已经受到了严重的挑战,人们的正常生活和社会的健康发展已经受到了严重的影响。本文针对这一现象,以基于有限数据的机器学习方法——支持向量机(Support Vector Machine,SVM),从四个方面的城市承载力历史数值角度出发对未来的相应的城市承载力数值进行预测。起初,分别对这四个承载力单列的时间序列数据进行相空间重构生成时序矩阵,扩大信息量,确定水资源、土地资源、交通和大气环境它们各自的承载力最优嵌入维度分别为4、5、2、6.然后运用支持向量回归模型对时序矩阵中的数据进行建模。鉴于模型结果会因为支持向量回归机参数的选择不同导致结果差异明显,本文依据核函数参数敏感度强于核函数敏感度的理论,采取以下两种方式选择参数来提高支持向量机回归预测模型效果:其一,直接使用支持向量机中默认参数;其二,采用粒子群优化算法(PSO)择优选取惩罚因子与核参数。模型效果对比结果显示PSO-SVM比一般SVM的预测结果更加精确,其实用性更好。之后使用PSO-SVM模型对杭州市未来五年的水资源、土地资源、交通承载力以及大气环境承载力进行预测。结果显示未来五年杭州市的综合承载能力呈现下降趋势,水资源和大气环境承载力处于较好的状态,一般将不会对杭州市的发展产生较坏的影响。而土地资源承载力和交通承载力则相对比较低下,最有可能对杭州市未来发展可能造成阻碍。
[Abstract]:With the acceleration of urbanization and the rapid development of social economy, many cities are blindly developing and building, urban resource utilization efficiency is low, leading to the phenomenon of land waste, freshwater and energy resources are tight. Serious environmental pollution, serious degradation of ecosystem, traffic congestion and other urban crisis risk frequently. And the size of urban bearing capacity is related to the sustainable and healthy development of the city. Whether people can enjoy a higher quality of life, therefore, people are eager to find a practical and effective model to predict the urban carrying capacity. Therefore, this paper focuses on urban water resources, land resources, traffic and atmospheric environment. Combined with the statistical yearbook and water resources bulletin to excavate the existing urban data, respectively to build the urban water resources carrying capacity evaluation index, land resources carrying capacity evaluation index, traffic carrying capacity evaluation index. The evaluation index of atmospheric environmental carrying capacity and the comprehensive carrying capacity evaluation index are expected to provide the direction for solving many problems arising from the rapid development of the city at the same time. In order to fully show the current development of urban bearing capacity of central cities in China, this paper selected the Beijing-Tianjin-Hebei, Yangtze River Delta, Pearl River Delta and other important regions and municipalities directly under the Central Government based on the selected evaluation indicators. There are 64 cities in provincial capitals and planned cities to study the status of carrying capacity. The study found that the comprehensive carrying capacity of cities in China is generally poor in the year of 2014. To be specific, the comprehensive carrying capacity of cities in China is generally poor. More than 50% of the urban water resources carrying capacity is in the state of early warning or even crisis; The carrying capacity of land resources in more than 56% cities is in a state of crisis. About 80% of the urban traffic carrying capacity is extremely fragile; There is only one city with excellent air quality, accounting for 1.56%. All this shows that the urban carrying capacity of Chinese cities has been seriously challenged. The normal life of people and the healthy development of society have been seriously affected. Support vector machine support Vector machine (SVM) is used as a machine learning method based on finite data. From four aspects of the historical value of urban bearing capacity from the point of view of the future of the corresponding urban carrying capacity of the prediction. At first. The time series data of these four single columns of bearing capacity are reconstructed to generate time series matrix to enlarge the amount of information and to determine the water resources and land resources. The optimal embed dimensions of transport and atmospheric environment are respectively 4? 5? 2??? 6. Then the support vector regression model is used to model the data in the time series matrix. Based on the theory that kernel function parameter sensitivity is stronger than kernel function sensitivity, this paper adopts the following two ways to select parameters to improve support vector machine regression prediction model effect: first. Direct use of default parameters in support vector machines; Secondly, the particle swarm optimization (PSO) algorithm is used to select the penalty factor and kernel parameter. The model results show that PSO-SVM is more accurate than the general SVM. Its practicability is better. Then the PSO-SVM model is used to analyze the water resources and land resources of Hangzhou in the next five years. The results show that the comprehensive bearing capacity of Hangzhou city in the next five years shows a downward trend, water resources and atmospheric environmental carrying capacity is in a better state. Generally, it will not have a bad effect on the development of Hangzhou, but the carrying capacity of land resources and traffic is relatively low, which is most likely to hinder the future development of Hangzhou.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：C912.81;C812

【相似文献】