【作者】 李婕；

【导师】 张秋文；

【作者基本信息】 华中科技大学 ， 空间信息科学与技术， 2018， 博士

【摘要】 空间信息科学与技术是当代发展最为迅速的领域之一,其利用重要的时空信息,应用在社会的方方面面。近些年来,在需求的强烈推动下,空间信息科学与技术特别是GIS有了新的大幅度发展,其与智能计算技术相结合,应用到各类专业领域已成为一种趋势。中小科技企业是世界各国和地区高新技术产业的基础,对经济和社会发展起着引领作用。由于研发投入大、产出不确定和风险大等原因,中小科技企业面临较大的生存与发展压力,而客观评价和分析中小科技企业的发展环境,掌握这类企业发展的时空规律,找到影响它们发展的内外部要素,是促进其健康成长的重要基础。由于中小科技企业发展环境具有较大的时空非均匀性和复杂性,需要结合空间信息技术和智能计算技术,对该问题展开有针对性的科学研究。论文对国内外地理空间数据分析与GIS、智能计算、中小科技企业发展环境的相关研究现状进行了分析,在此基础上,总结了研究中的主要问题与难点,重点针对已有研究中在评价模型与时空分析方法等方面存在的不足,考虑中小科技企业生存适宜度、融资信用和创业信心等三种不同评价问题的特性,基于多种智能计算技术,提出了三种地理空间评价混合模型,即GWO-SVM进化支持向量机混合模型、AHP-FCE-CM云模型改进的多层次模糊综合评判混合模型、VMD-GWO-ELM三阶段进化极限学习机混合模型,并进行了应用实验。与此同时,分别将三种混合模型与GIS技术相结合,对专题空间数据进行了辨识与分析,构建了专题空间数据模型和数据结构,提出了基于GIS的中小科技企业发展环境时空分析方法,通过实例应用,对研究区域中小科技企业发展环境的空间分布特征和时间演变规律进行了挖掘与分析。论文的主要研究内容和成果如下:1.基于GWO-SVM进化支持向量机混合模型和GIS的生存适宜度评价及其时空规律研究:首先,在阐述中小科技企业生存适宜度定义、问题特性基础上,从经济、金融、科教和基础设施等方面,构建生存适宜度评价指标体系。其次,针对小样本数据集评价问题,通过采用灰狼优化算法改进支持向量机的参数寻优机制,优化支持向量机的惩罚因子与核函数宽度值,构建评价精度更高、分类能力更强的GWO-SVM混合模型,并应用于中小科技企业生存适宜度评价建模实例。再次,将GWO-SVM和GIS技术结合,对生存适宜度空间数据进行辨识与分类,构建适宜度空间数据模型和数据结构,提出了基于GIS的中小科技企业生存适宜度时空分析方法。最后,将时空分析方法运用到2010-2015年中国29个省份中小科技企业生存适宜度时空评价具体实例中,从时空总体、空间、时间和机理等四个维度,对这些区域中小科技企业生存适宜度进行时空分析,掌握了中国中小科技企业生存适宜度的空间分布特征,揭示了该类企业生存适宜度的时间演变规律,并对其形成机理进行了挖掘与探讨。2.基于AHP-FCE-CM云模型改进的多层次模糊综合评判混合模型和GIS的融资信用评价及其空间分布规律研究:首先,分析中小科技企业融资信用基本含义、评价问题特性,并从宏观和微观两个方面,构建了中小科技企业融资信用评价体系。其次,针对群体决策过程中存在的随机性以及模糊性等问题,利用云模型分别对层次分析法和模糊综合评判法进行改进,构建了一种置信度更高、鲁棒性更强的基于云模型改进的多层次模糊综合(AHP-FCE-CM)评价方法,并应用于中小科技企业融资信用评价建模实例中。再次,将AHP-FCE-CM模型与GIS技术相结合,对融资信用空间数据进行辨识与分类,构建了融资信用空间数据模型和数据结构,提出基于GIS的中小科技企业融资信用时空分析方法。最后,将这种时空分析方法运用到中小科技企业融资信用时空评价实例中,对2014年中国新三板29家中小科技企业融资信用状况进行时空分析,获取了样本中小科技企业融资信用等级的空间分布特征,并从宏观和微观两个层面,挖掘和揭示了形成这种空间分布规律的可能原因。3.基于VMD-GWO-ELM三阶段进化极限学习机混合模型和GIS的创业信心评价及其时空规律研究:首先,在阐述中小科技企业创业信心的基本含义、评价问题特性基础上,采用百度搜索指数,构建了中小科技企业创业信心指标。其次,针对非平稳时间序列的机器学习问题,通过引入变分模态分解方法分解极限学习机的时间序列输入变量,并采用灰狼算法改进极限学习机的参数寻优机制,从而优化极限学习机的权值与偏置,提出了一种基于非平稳时间序列分解和数据驱动的三阶段进化极限学习机(VMD-GWO-ELM)评价方法,并应用于中小科技企业创业信心评价建模实例中。再次,将VMD-GWO-ELM模型与GIS相结合,对创业信心空间数据进行辨识与分类,构建相应的创业信心空间数据模型与模型结构,提出基于GIS的中小科技企业时空分析方法。最后,将这种时空分析方法运用到中小科技企业创业信心时空评价具体实例中,对2010-2017年湖南和湖北两省中小科技企业创业信心进行时空分析,挖掘并揭示了研究区域中小科技企业创业信心的时间演变规律及其基本模式。论文采用多种智能计算方法,针对地理空间评价问题的具体特征而建立的混合模型,极大地提高了中小科技企业发展环境的评价水平,采用基于GIS的时空分析方法,以可视化的方式挖掘和揭示了区域中小科技企业发展环境的时空规律及其内在机理,研究成果可以为政府、投资者和企业家的相关管理与决策提供重要依据,具有重要的理论意义和实用价值。更多还原

【Abstract】 Spatial information science and technology is one of the most rapidly developing areas of science and technology in modern times,which is widely applied in various kinds of aspects of society by employing essential spatio-temporal information.In recent years,with the strong impetus of demand,spatial information science and technology,especially geographic information system(GIS)technology,has developed greatly.It has become a trend to combine GIS technology with intelligent computing technology in various professional fields.Scientific and technological small–medium enterprises(STSMEs)are the basis of high-tech industries around the world,playing a leading role in the global economic and social development.Owing to kinds of developing difficulties,i.e.,high research and development investment,output uncertainties and high risks,the STSMEs are often hard to survive and develop.Evaluation and analysis on development environment of the STSMEs,finding their spatio-temporal laws as well as internal and external factors that affect their development,can help to solve the survival and development issues for these enterprises.By analyzing the research status and progress on geospatial data analysis and GIS,intelligent computation,developing environment of STSMEs,the main defects and difficulties of researchs,i.e.,lacks of advanced evaluation models and spatio-temporal analysis method,are summarized in this thesis.Considering the characteristics of three different evaluation issues,such as survival suitability,financing credit and entrepreneurial confidence of STSMEs,three new hybrid models for evaluating geospatial problems are proposed through combining various intelligent computing techniques.They are GWO-SVM evolutionary support vector machine model,AHP-FCE-CM cloud model improved multilevel fuzzy comprehensive evaluation model,and VMD-GWO-ELM three-stage evolutionary extreme learning machine model.The application experiments of these models are performed on evaluating the STSMEs’ development environment.Combining the three hybrid models with GIS technology,the thematic spatial data are identified and analyzed,the thematic spatial data model and data structure are constructed,and the spatial-temporal analysis methods for the development environment of STSMEs based on GIS are put forward in the thesis,which are used for analyzing spatial distribution characteristics and time evolutional laws of STSMEs’ development environment.The main contents of the thesis are listed as follows:1.The research on evaluation for survival suitability and its spatio-temporal laws of STSMEs based on GWO-SVM evolutionary support vector machine hybrid model and GIS: on the basis of the accurate definition description and evaluation problem characteristics of STSMEs’ survival suitability,the thesis builds up the evaluation index systems from the four aspects which are economy,finance,scientific education and infrastructure.The thesis uses the grey wolf optimization algorithm to improve the parameter optimization mechanism of support vector machine,and constructs a GWO-SVM evolutionary support vector machine hybrid model based on small sample data sets,which is applied to evaluate survival suitability of STSMEs.By combining GWO-SVM model and GIS technology,with the identification of spatial data and the construction of spatial data model and structure,a spatial-temporal analysis method for the survival suitability of STSMEs is proposed,which is applied to spatial-temporally analyze survival suitability of STSMEs in 29 provinces of China in 2010-2015 years on four perspectives such as whole space-time environment,spatial distribution,time evolution and mechanism.The spatial distribution characteristics,time evolutional laws and mechanism of survival suitability of STSMEs in China are revealed.2.The research on financing credit evaluation and its spatial distribution laws of STSMEs based on AHP-FCE-CM cloud model improved multilevel fuzzy comprehensive evaluation hybrid model and GIS: according to the basic implication and evaluation problem characteristics of STSMEs financing credit,the thesis constructs a financing credit evaluation index system for STSMEs based on macro and micro aspects.In order to solve the problems of randomness and fuzziness in the process of group decision making,the thesis improves the AHP and fuzzy comprehensive evaluation method by employing cloud model,and establishes an AHP-FCE-CM cloud model improved multilevel fuzzy comprehensive evaluation hybrid model with higher robustness,which is applied to evaluate financing credit of STSMEs.Through incorporating AHP-FCE-CM model and GIS technology,a spatial-temporal analysis method for financing credit of STSMEs is proposed,which is applied to spatial-temporally analyze financing credit of STSMEs using the data of 29 STSMEs in new third board in China in 2014.The spatial distribution characteristics and laws,as well as their possible formation reasons of sample enterprises are discovered and discussed in the thesis.3.The research on entrepreneurship confidence and its spatio-temporal laws of STSMEs based on VMD-GWO-ELM three-stage evolutionary extreme learning machine hybrid model and GIS: on the basis of implication description and evaluation problem characteristics of STSMEs’ entrepreneurial confidence,the thesis sets up the entrepreneurial confidence index system using Baidu search index.In order to mining and predict nonstationary time series,the variational mode decomposition is used to decompose the input variables of the extreme learning machine,and builds up a VMD-GWO-ELM three-stage evolutionary extreme learning machine hybrid model based on non-stationary time-series decomposition and data-driven,which is applied for evaluating entrepreneurial confidence of STSMEs.By combining VMD-GWO-ELM model and GIS technology,a spatial-temporal analysis method for entrepreneurship confidence of STSMEs is proposed,which is applied to spatial-temporally analyze entrepreneurship confidence of STSMEs by using the data of STSMEs in Hunan and Hubei provinces from 2010 to 2017.The time evolutional laws and basic patterns of entrepreneurial confidence of STSMEs in specific regions in China are revealed.In order to solve geospatial evaluation issues with different characteristics,the thesis establishes hybrid evaluation models based on a variety of intelligent computing methods,which can greatly improve the evaluation level of the development environment of STSMEs.The thesis proposes the spatio-temporal analysis methods for development environment of STSMEs based on GIS,mining and revealing the spatio-temporal laws and internal mechanism of development environment of STSMEs.The research findings have significant theoretical significance and practical value.It could provide technical support for decision-making by the government,investors or entrepreneurs.更多还原