终端区动态交通特征与运行态势研究
发布时间：2018-07-14 14:15
【摘要】：面对空中交通运输需求的快速增加,运用技术手段提升空管运行效率与服务能力以缓解供需矛盾是当前行之有效的方式。由于传统经验型的粗放运行管理模式,未能有效利用空中交通特性、规律及时空特征,以帮助认知运行的动态性与问题所在,因此难以实现针对性、精细化的管理与科学决策。随着空管运行数据采集的不断完善,利用数据挖掘技术发掘空中交通的运行规律、动态模式、场景分类等隐含知识具备了基础条件,相应的智能数据分析与决策技术也成为当前的研究热点与趋势。本文总结了空中交通领域交通特性分析与智能决策支持的研究现状与趋势,以终端区为对象,从管制运行与流量管理的不同应用决策需求出发,深入研究了空中交通特性与运行态势相关的若干数据分析与挖掘方法,以期提供科学有效的决策数据与模型支持,主要内容包括:(1)机场飞行区航班运行时间的特征分析与决策应用。利用聚类方法提取飞行区场面滑行阶段运行时间的聚集分布模式,从航班运行的阶段构成出发,将场面滑行时间的周期性单日变化特征应用于城市对航班运行时间的差异化度量,设计了一种按阶段独立计算、按条件整合的运行时间估算方法,以提供计划编制的依据;针对恶劣气象对航班离场阶段时间的影响,设计了适用于机场的气象影响程度指数(WITI),研究了机场WITI与不同离场延误特征间的变化关系及规律,建立了回归模型,以支持预测气象条件下的终端区离场阶段延误的早期评判。(2)建立了终端区交通流识别方法与交通流相态判别模型。从空中交通流特性分析与状态识别的需求出发,首先利用改进的相似性度量与谱聚类实现了终端区各类交通流的识别与参考航迹提取,抽象了表征交通流相态的特征度量。通过特征间的数值关系与变化规律,并结合领域认知,界定了终端区交通流的自由态、平稳态与拥堵态,及各相态对应的管制运行特性与演化过程。以此为经验,构建了因子分析与遗传EM聚类的交通流相态模糊识别方法,实现交通流相态影响因素分析与隐性特征向量的抽取,为终端区流量时空分布调配与飞行程序优化提供支撑。(3)构建了终端区交通态势模糊评价方法与态势等级预测模型。首先从交通态势评价的指标缺陷及主观性问题出发,通过提取终端区宏观与微观的交通特性,从进离场及总量三个维度构建态势指标描述,进而设计了基于中介真值程度度量与熵理论权值赋值的终端区交通态势模糊评价模型,实现了态势的客观判别与差异化度量。同时,鉴于态势的模糊性与认知差异,以及流量调配对交通态势识别的需求,设计了BP神经网络模型,并利用中介评价结果与实际管制认知结果设定的样本集合进行模型训练与验证,实现了终端区交通态势等级的快速准确预测。(4)设计了终端区ATFM决策支持概念框架与可扩展平台架构。由流量管理的目标与泛化内涵出发,界定了终端区运行决策支持的应用范畴及方法体系,并从流量管理时效阶段与决策应用领域维度,实现具体应用决策方法的关联,构建了可扩展的终端区ATFM决策支持概念框架与一般决策过程框架。以此为依据,进一步建立了决策信息抽取的层次化概念模型,设计了一种具备扩展能力与松耦合特征的多层次平台体系框架,通过决策逻辑、数据挖掘方法及特征/属性度量的分离设计,支持应用工具的快速开发与配置。为终端区运行决策支持方法的应用、研究与工程实施提供科学参考。最后,对本文的研究成果进行了概括性总结,并针对现有研究的缺陷与需求外延,对进一步的研究工作内容及方向进行了展望。
[Abstract]:In the face of the rapid increase in the demand for air transportation, it is an effective way to use technical means to improve the efficiency and service ability of air traffic tube in order to alleviate the contradiction between supply and demand. As the problem lies, it is difficult to realize the pertinence, fine management and scientific decision. With the continuous improvement of the data collection of the air traffic management, the underlying conditions of the hidden knowledge, such as the running law of air traffic, the dynamic mode, the scene classification, and so on, are provided with the data mining technology, and the corresponding intelligent data analysis and decision technology have also become the present. This paper summarizes the current situation and trend of traffic characteristics analysis and intelligent decision support in the field of air traffic. Based on the terminal area as the object, starting from the different application decision requirements of the control operation and the flow management, a number of data analysis and mining methods related to the air traffic characteristics and the transportation situation are deeply studied. In order to provide scientific and effective decision data and model support, the main contents include: (1) the characteristics analysis and decision application of the flight time of the airport flight area. The clustering method is used to extract the aggregation distribution pattern of the running time of the flight stage, starting from the stage of the flight operation, and taking the cyclical single time of the skidding time. The diurnal variation characteristics are applied to the difference degree of the flight time of the city, and a method to estimate the running time according to the conditions is designed to provide the basis for the planning, and the meteorological influence index (WITI) for the airport is designed for the effect of the bad weather on the departure time of the flight. The relationship between the airport WITI and the characteristics of different departure delays is investigated. A regression model is established to support the early evaluation of the delay in terminal departure stage under the forecast weather conditions. (2) a traffic flow identification method and a traffic flow phase state discrimination model are established. First, we use improved similarity measure and spectral clustering to realize the recognition and reference track extraction of all kinds of traffic flow in terminal area, abstract the characteristic measurement of traffic flow phase state, and define the free state of traffic flow in terminal area, flat steady state and congestion state through the domain cognition. The characteristic and evolution process of phase state corresponding control operation. In this way, a traffic flow phase state fuzzy recognition method of factor analysis and genetic EM clustering is constructed to realize the analysis of the factors of traffic flow phase state and the extraction of the recessive characteristic vector, which provides support for the spatial and temporal distribution and optimization of the terminal area. (3) the terminal is constructed. The fuzzy evaluation method and the situation grade prediction model of the regional traffic situation. First, starting from the index defect and subjective problem of the traffic situation evaluation, by extracting the macroscopic and microscopic traffic characteristics of the terminal area, the description of the situation index is constructed from three dimensions of entering the field and the total amount, and then based on the measure of the true value of the intermediary and the entropy theory. At the same time, the BP neural network model is designed in view of the ambiguity of the situation and the difference of cognition and the demand for traffic situation recognition, and the sample set is set up by the results of intermediary evaluation and the actual control cognitive results. According to the training and verification of the model, the rapid and accurate prediction of traffic situation level in terminal area is realized. (4) the concept framework and extensible platform architecture of terminal area ATFM decision support are designed. The application category and method system of the terminal operation decision support are defined by the target and generalization of traffic management. The effect phase and the decision application domain dimension, the realization of the relevance of the specific application decision method, the extensible conceptual framework of ATFM decision support and the general decision process framework. Based on this, a hierarchical conceptual model of decision information extraction is established, and a kind of multiple features with the expansion ability and loosely coupled is designed. The framework of hierarchical platform system is designed to support the rapid development and configuration of application tools through decision logic, data mining and feature / attribute measurement. It provides a scientific reference for the application of decision support methods for terminal operation, and provides a scientific reference for the implementation of the research and engineering implementation. The shortcomings and demands of research are discussed, and further research contents and directions are prospected.
【学位授予单位】：南京航空航天大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：V355.1
，
本文编号：2121907
[Abstract]:In the face of the rapid increase in the demand for air transportation, it is an effective way to use technical means to improve the efficiency and service ability of air traffic tube in order to alleviate the contradiction between supply and demand. As the problem lies, it is difficult to realize the pertinence, fine management and scientific decision. With the continuous improvement of the data collection of the air traffic management, the underlying conditions of the hidden knowledge, such as the running law of air traffic, the dynamic mode, the scene classification, and so on, are provided with the data mining technology, and the corresponding intelligent data analysis and decision technology have also become the present. This paper summarizes the current situation and trend of traffic characteristics analysis and intelligent decision support in the field of air traffic. Based on the terminal area as the object, starting from the different application decision requirements of the control operation and the flow management, a number of data analysis and mining methods related to the air traffic characteristics and the transportation situation are deeply studied. In order to provide scientific and effective decision data and model support, the main contents include: (1) the characteristics analysis and decision application of the flight time of the airport flight area. The clustering method is used to extract the aggregation distribution pattern of the running time of the flight stage, starting from the stage of the flight operation, and taking the cyclical single time of the skidding time. The diurnal variation characteristics are applied to the difference degree of the flight time of the city, and a method to estimate the running time according to the conditions is designed to provide the basis for the planning, and the meteorological influence index (WITI) for the airport is designed for the effect of the bad weather on the departure time of the flight. The relationship between the airport WITI and the characteristics of different departure delays is investigated. A regression model is established to support the early evaluation of the delay in terminal departure stage under the forecast weather conditions. (2) a traffic flow identification method and a traffic flow phase state discrimination model are established. First, we use improved similarity measure and spectral clustering to realize the recognition and reference track extraction of all kinds of traffic flow in terminal area, abstract the characteristic measurement of traffic flow phase state, and define the free state of traffic flow in terminal area, flat steady state and congestion state through the domain cognition. The characteristic and evolution process of phase state corresponding control operation. In this way, a traffic flow phase state fuzzy recognition method of factor analysis and genetic EM clustering is constructed to realize the analysis of the factors of traffic flow phase state and the extraction of the recessive characteristic vector, which provides support for the spatial and temporal distribution and optimization of the terminal area. (3) the terminal is constructed. The fuzzy evaluation method and the situation grade prediction model of the regional traffic situation. First, starting from the index defect and subjective problem of the traffic situation evaluation, by extracting the macroscopic and microscopic traffic characteristics of the terminal area, the description of the situation index is constructed from three dimensions of entering the field and the total amount, and then based on the measure of the true value of the intermediary and the entropy theory. At the same time, the BP neural network model is designed in view of the ambiguity of the situation and the difference of cognition and the demand for traffic situation recognition, and the sample set is set up by the results of intermediary evaluation and the actual control cognitive results. According to the training and verification of the model, the rapid and accurate prediction of traffic situation level in terminal area is realized. (4) the concept framework and extensible platform architecture of terminal area ATFM decision support are designed. The application category and method system of the terminal operation decision support are defined by the target and generalization of traffic management. The effect phase and the decision application domain dimension, the realization of the relevance of the specific application decision method, the extensible conceptual framework of ATFM decision support and the general decision process framework. Based on this, a hierarchical conceptual model of decision information extraction is established, and a kind of multiple features with the expansion ability and loosely coupled is designed. The framework of hierarchical platform system is designed to support the rapid development and configuration of application tools through decision logic, data mining and feature / attribute measurement. It provides a scientific reference for the application of decision support methods for terminal operation, and provides a scientific reference for the implementation of the research and engineering implementation. The shortcomings and demands of research are discussed, and further research contents and directions are prospected.
【学位授予单位】：南京航空航天大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：V355.1
，
本文编号：2121907
本文链接：https://www.wllwen.com/shoufeilunwen/gckjbs/2121907.html
