节点文献

基于软计算融合的城市道路交通资源选址研究

Location Selection of Urban Road Traffic Resoures Based on Soft Computing

【作者】 汤旻安

【导师】 任恩恩;

【作者基本信息】 兰州交通大学 , 交通信息工程及控制, 2011, 博士

【摘要】 城市道路交通问题是全社会关注的热点问题之一。软计算技术的迅速发展,各种模型和综合分析方法层出不穷,新的理论和研究成果不断出现,并已在实际的理论研究和工程应用中显示出巨大的威力和发展潜力。运用软计算技术进行科学的选址,让有限的城市道路交通资源发挥最大的效能,从而达到更进一步缓解交通矛盾的作用,是论文研究的主要问题。论文在广泛搜集、阅读国内外有关选址—分配和软计算理论与方法的最新文献和成果的基础上,研究了软计算理论中的模糊逻辑、人工神经网络、遗传算法、粗糙集等的基本思想和工作原理以及基本理论,系统研究了城市道路交通资源选址的理论与方法;针对分析不同的城市道路交通资源选址中的困难和不确定性,把软计算的基本思想融入城市道路交通资源选址的策略中,在此基础上将这几种软计算技术有机结合,提出了针对不同问题的城市道路交通资源选址的软计算模型和算法,以提高城市道路交通资源的利用率和实用性,提供合理、科学的实施方法,促进软计算方法在城市道路交通资源选址工程上的应用。通过实例的分析和仿真表明,这些方法能更有效地解决城市道路交通资源选址方面的问题。论文的主要内容如下:深入分析了城市道路交通面临和存在的严峻形势,指出道路交通资源选址在智能运输系统(ITS)中的作用、意义,对前人在选址—分配问题和软计算技术理论应用文献查阅、分析研究的基础上,对软计算理论方法的发展、研究、应用进行了详尽的分析和阐述,提出了以软计算方法进行城市道路交通资源选址的思想。研究了软计算方法的融合集成,通过在分析常用软计算方法优缺点的基础上,对软计算方法的融合集成哲学基础、方法学基础、融合集成原则、融合集成的形式进行了论述,研究了论文中运用的软计算融合集成方法技术路线。研究了模糊逻辑、神经网络、遗传算法融合集成在城市道路交通资源选址中的决策知识规则提取问题。首先研究了模糊逻辑与神经网络的融合集成,运用上述理论技术有效地建立了将城市道路交通资源选址中的资源属性或道路交通调查数据转换成具有专家知识形式的模糊规则的问题求解模型。在此基础上提出了基于遗传算法的NFS思想,实现了由道路交通调查数据到决策信息,由决策信息到决策知识的转换。研究了在城市道路交通资源选址问题求解中面临少数据、小样本、贫信息、不确定性等问题使用灰色神经网络建立预测模型的方法。阐述了灰色系统与神经网络相结合的基本理论,在此基础上建立了以遗传算法优化改进灰色神经网络“白化”参数的预测求解模型,并以换乘需求总量为公交换乘枢纽选址主要因素的实际应用需求为研究对象,进行了仿真实验研究与对比分析。以此,对基于遗传算法优化改进灰色神经网络的方法在求解城市道路交通资源选址问题中的应用进行了深入研究。研究了面对城市道路交通资源选址问题求解中建模因素过剩或庞杂时采用多重GA与BP神经网络相结合的因素筛选进行自变量降维及对系统进行预测的模型方法。阐述了在城市道路交通资源选址问题中运用神经网络的预测理论和技术方法,剖析了诸如轨道交通线路选址中面临多因素高维自变量问题,然后在分析上述多变量选址模型的基础上,给出了基于遗传算法和BP神经网络相结合的因素筛选及系统预测的复合优化模型方法。最后,利用实际数据进行了仿真研究,实现了城市轨道交通干线选址中的因素筛选,对城市轨道交通干线选址做了深入分析。研究了基于粗糙集属性约简与GIS技术相结合的高密度城区停车设施选址决策问题。从分析高密度城区停车设施的地理信息特点入手,提出了运用粗糙集理论与GIS技术相结合进行高密度城区停车设施选址决策条件属性因素约简的模型和算法,深入研究了采用互信息的模糊粗糙集属性约简算法对决策表进行约简得到相对约简决策规则的方法,然后,对规划中的决策分类进行评价和分析。结合实际仿真实例,对研究结果进行了评估和分析。给出了在我国进行道路交通资源选址的一些建议。最后,对全文进行了概括性总结,提出了一些需要完善的研究工作,并指出了在城市道路交通资源选址方面理论和应用上有待进一步研究的问题。

【Abstract】 Urban traffic problem is one of the hot issues of social concern. The rapid development of soft computing technology, emerging various models and comprehensive analysis as well as new theories and research results has shown great power and development potential in practical theoretical researches and engineering applications. Use soft computing technology to carry out scientific location selection to enable limited urban road traffic to achieve maximum performance, thus play the role of further easing traffic contradictions, which is the main issue of this paper.On the basis of extensively collecting and reading the latest literature and achievements concerning location selection—allocation and soft computing theory and method at home and abroad, this paper has studied the basic idea, principle and basic theory of fuzzy logic, artificial neutral networks, generic algorithms and rough sets in soft computing theory, systematically researching the theory and methods of urban road traffic location selection. Targeting on the difficulties and uncertainties in the location selection of different urban road traffic resources, integrate the basic idea of soft computing idea into the strategy of urban road traffic resource location selection, organically combine these soft computing technologies on this basis, proposing the corresponding soft computing models and algorithms to improve the utilization and availability of urban road traffic resources, provide reasonable and scientific implementation methods, thereby promote the application of soft computing methods in the engineering of urban road traffic resource location selection. Analysis and simulation shows these methods can more effectively solve the problem of urban road traffic resource location selection. The main contents of this article are as follows:First, it carries out in-depth analysis on the severe situation faced by urban road traffic, pointing out the role and significance of road traffic resource location selection in intelligent transportation systems (ITS), on the basis of collecting and reading the existing location selection and allocation as well as soft computing technology theory, then makes detail introduction on the development, research and application of soft computing method, this paper proposes the idea of conducting urban road traffic resource selection based on soft computing method.Study the integration of soft computing method, based on the analysis of merits and demerits of common soft computing method, discuss the philosophical foundation, methodological foundation, integration principles and integration forms of soft computing method, researching the technology route of soft computing integration in this paper.Study the decision knowledge rule extraction in the research of urban road traffic resource location selection integrated by fuzzy logic, neutral networks and genetic algorithms. It firstly focuses on the integration of fuzzy logic and neutral network, discusses efficient transformation of urban road traffic location selection resource property or road traffic survey data to fuzzy rules with expert knowledge. The paper proposes genetic algorithm based NFS thought, realizing the transformation from data to information and from information to knowledge.Research the method of using gray neutral network prediction in solving small data, small samples, poor information and uncertainty in face of urban road traffic resource location selection; describe the basic theory of grey system and neutral network and analysis the main factor of transfer need amount for bus transfer hub selection. On this basis, the transfer need amount prediction solution model improving grey neutral network "whitening" parameters by genetic algorithm, based on the practical application, carry out simulation experiment study and comparative analysis, thus conduct in-depth study on the application of generic algorithm based grey neutral network improvement method.Study the model of independent variable dimensionality reduction and system prediction by surplus or numerous and jumbled modeling factors in urban road traffic resources location selection with the combination of multiple heterogeneous genetic algorithm and neural network. First, it makes an overall introduction of applying neutral network prediction theory and technology in urban road traffic resource location selection, analyzes the multi-factor and high-dimensional variables faced by rail transit route location selection, and then puts forward the factor screening and system prediction model method with the composite optimization of generic algorithm and BP neural network. Finally, conduct simulation research based on the practical application need data to realize the factor screening in urban rail transit lines, and make in-depth analysis on urban rail transit lines location selection.Study high-density urban parking facility locations selection based on the combination of rough set attribute reduction and GIS technology. Start from the analysis of high-density urban packing facility geographic information features, it proposes using the combination of rough set theory and GIS technology to get the model and algorithms of high-density urban parking facility location selection decision factors’attribute reduction. Make in-depth study on the attribute reduction by using mutual information fuzzy rough set to obtain the relative reduced decision rules, and then carry out evaluation and analysis on the decision classification in planning. With the actual simulation results, carry out assessment and analysis on research results.Some problems about location selection of urban road traffic resources that should be noted are discussed.Finally, make a summary of the full text, propose the researches requiring improvement and point out issues needing further study of urban road traffic resources location selection in terms of theory and application.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络