【作者】 刘峰涛；

【导师】 贺国光；

【作者基本信息】 天津大学 ， 管理科学与工程， 2007， 博士

【摘要】 科学正确地认识管理和控制对象是实现有效管理和控制的前提。把交通运输系统视为一个线性、定常系统的假设已经被近年的研究所颠覆,“复杂的、非线性系统”的认识已在学术界形成共识。对交通运输系统复杂性、非线性的定量刻画研究也逐渐成为热点问题。目前这些研究存在一些难点:刻画理论缺少系统的基础理论和完整的刻画程序;刻画对象需要的数据量较大;刻画方法不能实现可比化度量;对交通运输系统的复杂性认识还停留在浅层次的阶段,如何针对系统的复杂特性进行管理和控制尚处空白。基于此,选择和改进短序列条件下的可比化测度方法,定量研究和分析交通运输系统的复杂性,并把复杂性测度与交通运输系统的管理和控制结合起来,是交通运输系统复杂性刻画研究的深化,其理论和现实意义显而易见。本文的主要内容和创新之处可概述如下:1、引入Lempel-Ziv算法、近似熵、统计复杂度等非线性动力学的复杂性测度方法。通过对16个实测城市交通流序列(或数组)和6个实测高速公路交通流序列的分析,实现了复杂性测度方法的检验、选择、改进和交通流系统的复杂性测度,并考察了复杂度与混沌、分形的关系,解决了传统刻画方法的“短序列、可比较”难题,得到了一些有用的结论。2、以ARIMA模型、RBF神经网络、非参数估计的混沌局部预测等预测方法为基础,讨论交通流的复杂度与可预测性的相关关系、交通流预测方法的复杂度适用条件。另外,通过仿真实验来研究复杂度在交通控制中的应用,提出了复杂度在交通控制中的简单应用方案。3、根据宏观交通运输系统的特点,提出了能够对超短序列进行复杂性测度的改进Lempel-Ziv算法,实现了中、日两国宏观交通运输系统的测度,得出了宏观交通运输系统大致是线性系统等结论。4、研究了宏观交通运输系统的复杂度与可预测性、宏观交通运输系统复杂度与GDP复杂度的相关关系。5、完成了系统复杂性测度的理论架构。从现象学和意向性科学入手,提出意向性结构模型,明确系统复杂性测度是一个二步意向性解释过程;根据测度论来明确复杂性测度的数学基础和条件;提出复杂性测度的一般程序。更多还原

【Abstract】 The precondition realizing efficient management and control is to get to know the object of management and control, scientifically, correctly. The assumption that the traffic transportation system is a linear, constant system has been overthrown by recent research and the facts that the traffic transport system is complex, nonlinear system are extensively accepted by academe. The quantitative depiction research on complexity, non-linearity of transportation system gradually becomes hotspot research field. Presently, the difficulties in researches are following listed: depicting theory is shortage of systemic basic theory and rounded depicting procedure. Depicting objects need a mass of data. Depicting methods could not realize comparable measurement. The cognition of complexity of traffic transportation system still rests on lower level and how to manage and control traffic transportation system against complexity characteristics is still blank. On basis of these facts, the following researches deepen depicting research of traffic transportation system complexity: choosing and mending comparable measurement under condition of short series, quantitatively researching and analyzing complexity of traffic transportation system, combining complexity measurement with management and control of traffic transportation system. Academic and realistic meaning is obvious for these studies.The key points and main achievements in paper are listed as follows:1.By virtue of analyzing sixteen series (or arrays) of traffic flow in city and six series of traffic flow in highway, the test, choice, improvement of complexity measurement methods and the complexity measurement of traffic flow system are realized. Moreover, the difficulties in short series, compares of traditional depiction methods are solved through reviewing the relations among complexity, chaos, fractal. Finally, some useful conclusions are drawn.2. Based on predicting methods, such as, ARIMA model, RBF neural networks, chaos local prediction of nonparametric estimation, the correlativity between traffic flow complexity and predictability and the complexity applying condition of different prediction methods are discussed. Moreover, the simple applying project of complexity in traffic control are put forward by emulate test to study complexity appliance in traffic control.3. Improved Lempel-Ziv algorithm, which could measure the complexity of super-short series, is put forward. The measures of macro-traffic transportation systems in China and Japan are realized. The conclusions that macro-traffic transportation system approximately is linear system, etc. are drawn.4. The following researches are carried out, i.e. complexity and predictability of macro-traffic transportation system, correlativity between complexity of macro-traffic transportation system and complexity of GDP.5. The construction of theory frame of system complexity measure has been finished. Starting with phenomenology and intentional science, intentional structure model is put forward and it is clear that system complexity measure is a two-step intention explanation process. The mathematics basis and condition of complexity measure are confirmed according to measure theory. General procedure of complexity measure is brought forward.更多还原