铁路集装箱中心站物流系统资源调配优化与仿真研究

发布时间：2019-03-27 21:31

【摘要】：集装箱运输费用低、效率高、协作性好,目前已受到各界的普遍重视。铁路集装箱运输作为其中一个关键环节,在整个集装箱运输系统中发挥着重要作用。铁路集装箱中心站是办理货物到发、存储、加工等业务的主要场所,也是现代综合物流体系中的重要组成部分。在日益激烈的市场竞争中,只有不断提高作业效率、降低运营成本,才能得以生存。然而,由于铁路集装箱中心站物流系统比较庞大,且具有高度非线性和动态性的特点,同时其作业过程影响因素众多,各环节之间的逻辑比较复杂,因此对系统优化的研究十分困难。本文对铁路集装箱中心站物流系统相关优化问题分别进行研究,采用不同的优化方法,对系统各环节以及作业资源进行分析和优化,从而达到科学管理和控制的目的。状态空间模型是在动态系统符合马尔科夫性假设基础上,由平稳时间序列分析而来的一种动态时域模型,该模型不仅能够准确地描述系统内部状态,而且能够很好地阐释内部状态与外部输入、输出变量的联系。互信息技术是典型的特征选择高维数据分离度量方法,通过建立高维特征提取向量与输出分类信息之间的内在联系,达到原始高维特征空间降维的目的。本文利用状态空间时间序列模型对于多输入、输出变量复杂问题的适用性,且不需要大量历史数据对系统状态进行描述的特点,建立了铁路集装箱中心站所在区域货运需求预测状态空间时间序列模型。同时,采用互信息技术对原始高维输入数据进行降维,并针对自然灾害、政策变动等一些特殊影响因素,提出加权的互信息计算方法,以提高综合信息提取与数据降维的能力。通过与LIBSVM支持向量回归模型以及局部线性小波神经网络模型的对比实验,证明了该模型对于小样本、高维度区域货运需求预测问题的有效性。排队论是研究系统运作策略相关问题的一种有效方法,由该理论建立起来的模型可进一步分为静态排队模型以及瞬时排队模型。瞬时排队模型由于系统状态描述比较困难、计算相对复杂,因此其应用受到许多限制。铁路集装箱中心站大门系统具有很强的动态性,传统静态排队模型难以对系统进行准确描述。本文针对铁路集装箱中心站大门系统拥塞问题,通过搜集有关数据,统计得出外部卡车到达时间间隔以及大门系统服务时间分布规律,并在此基础上分别建立瞬时排队模型及系统优化模型。采用等可能组合优化求解方法对模型进行求解计算。通过系统仿真对比实验以及三个方面的灵敏度分析,证明模型与方法的合理性及有效性。针对铁路中心站起重机调度与箱位分配决策问题,本文在综合考虑集装箱堆存方向、轨道式集装箱门式起重机安全距离等因素基础上,以最大作业完成时间最小化为目标,建立数学模型。启发式算法作为非确定性多项式完全问题近似求解的重要方法,在解决铁路集装箱中心站有关优化问题方面起着十分重要作用。回溯搜索优化算法是目前较新的一种进化算法,由于其总体结构比较简单,因此能够更加快速、有效地求解高维多模优化模型。同时,为了进一步提高算法寻找最优解的能力,并解决该算法易陷入局部最优解的缺点,本文对回溯搜索优化算法进行了相应的改进,从而提高算法的性能以及对于该问题的适应度。通过数值算例分析,证明铁路中心站起重机调度与箱位分配模型及改进回溯搜索优化算法对于求解该问题的可行性及有效性。铁路集装箱中心站物流系统很难采用传统数学模型予以描述,因此系统仿真方法是解决此类问题的有效途径。Simio仿真软件是面向“智能对象”的新一代三维系统仿真软件,采用独特的三层结构,对系统对象的行为、属性以及过程进行定义,因此具有良好的离散系统仿真能力。本文根据铁路集装箱中心站物流系统整体作业流程,建立相应的Simio系统仿真模型,通过系统实际参数输入,模拟特定条件下系统的运作情况。改变系统相关影响因素以及资源配置数量分别进行多次仿真实验,观察结果变化并得出相应结论。提出铁路集装箱中心站物流系统资源配置优化计算方法,并嵌入系统仿真逻辑,在仿真模型运行的过程中计算系统最优的资源数量。由模型实验结果可以得出,系统仿真方法能够对铁路集装箱中心站物流系统进行有效的模拟和监控,其仿真优化结果有助于管理者实施相关决策。本文采用互信息技术、状态空间时间序列、等可能组合算法、改进回溯搜索算法,以及Simio系统仿真技术对铁路集装箱中心站物流系统进行研究,从区域货运需求预测、大门系统拥塞优化、起重机调度与箱位分配决策、系统资源优化配置方面实现系统的智能控制。本文的研究具有重要的理论意义和实际应用价值,并为今后的研究奠定了基础。
[Abstract]:The cost of container transportation is low, the efficiency is high, the cooperation is good, and it is now widely regarded by various circles. As one of the key links, railway container transport plays an important role in the whole container transportation system. The railway container central station is the main place for handling the goods, storage and processing, and it is also an important part of the modern integrated logistics system. In the increasingly fierce market competition, only the operation efficiency is continuously improved, the operation cost is reduced, and the survival can be realized. However, because the logistics system of the railway container central station is relatively large, and has the characteristics of high nonlinearity and dynamics, and the influence factors of the operation process are numerous, the logic between the various links is more complex, and therefore, the research on the system optimization is very difficult. In this paper, the optimization problems related to the logistics system of the railway container central station are studied respectively, and different optimization methods are adopted to analyze and optimize the links and operating resources of the system, so as to achieve the purpose of scientific management and control. The state space model is a dynamic time-domain model which is analyzed by the stationary time series on the basis of the Markov-based hypothesis of the dynamic system. The model not only can accurately describe the internal state of the system, but also can well explain the internal state and the external input. The contact of the output variable. The mutual information technology is a typical feature selection high-dimensional data separation measurement method, and the purpose of reducing the dimension of the original high-dimensional feature space is achieved by establishing the internal relation between the high-dimensional feature extraction vector and the output classification information. In this paper, the applicability of the state space time series model to the complex problem of multi-input and output variables is used, and a large amount of historical data is not required to describe the system state, and the space time series model of the freight demand forecast state in the region where the railway container central station is located is established. At the same time, using the mutual information technology to dimension the original high-dimensional input data, and aiming at some special influence factors such as natural disaster and policy change, a weighted mutual information calculation method is proposed to improve the ability of comprehensive information extraction and data reduction. Based on the comparison between the support vector regression model of the LIBSVM and the local linear wavelet neural network model, the validity of this model for small samples and high-dimension freight demand forecasting is proved. The queuing theory is an effective method for studying the operation strategy of the system. The model established by the theory can be further divided into the static queuing model and the instantaneous queuing model. The instantaneous queuing model is relatively complicated because the system state description is difficult and the calculation is relatively complex, so its application is limited. The gate system of the railway container central station has a strong dynamic, and the traditional static queuing model is difficult to accurately describe the system. Aiming at the problem of the congestion of the gate system of the railway container central station, by collecting the relevant data, the time interval of the arrival time of the external truck and the time distribution of the service time of the gate system are obtained, and the instantaneous queuing model and the system optimization model are set up on the basis of this. The method is used to calculate the model by using the possible combination optimization solution method. The rationality and validity of the model and method are proved through the system simulation and contrast experiment and the sensitivity analysis of the three aspects. In order to solve the problem of scheduling and distribution of the crane in the central station of the railway, this paper, based on the consideration of the factors such as the stacking direction of the container, the safe distance of the rail-type container gantry crane, minimizes the time of the maximum operation, and sets up a mathematical model. The heuristic algorithm is an important method for solving the problem of the complete problem of the non-deterministic polynomial, and plays an important role in solving the problem of the optimization of the railway container central station. The backtracking search optimization algorithm is a new evolutionary algorithm, because the overall structure is relatively simple, so the high-dimensional multi-mode optimization model can be solved more quickly and effectively. At the same time, in order to further improve the ability of the algorithm to find the optimal solution, and to solve the disadvantage that the algorithm is easy to fall into the local optimal solution, this paper makes a corresponding improvement to the backtracking search optimization algorithm, so as to improve the performance of the algorithm and the fitness of the problem. By means of numerical examples, it is proved that the scheduling and box-level allocation model of the railway central station and the improved backtracking search optimization algorithm are feasible and effective to solve the problem. It is difficult to describe the traditional mathematical model in the logistics system of the railway container central station, so the system simulation method is an effective way to solve such problems. Simo simulation software is a new-generation three-dimensional system simulation software facing the "Smart Objects", and adopts a unique three-layer structure, and defines the behavior, the property and the process of the system object, thus having good discrete system simulation capability. According to the overall operation process of the logistics system of the railway container central station, this paper establishes the corresponding Simio system simulation model, and simulates the operation of the system under specific conditions through the system's actual parameter input. The influence factors of the system and the number of resource allocation are simulated, and the results are changed and the corresponding conclusions are obtained. The method for optimizing the resource allocation of the logistics system of the railway container central station is put forward, and the simulation logic of the system is embedded, and the optimal resource quantity of the system is calculated in the process of running the simulation model. The result of the model experiment can be obtained, and the system simulation method can effectively simulate and monitor the logistics system of the railway container central station, and the simulation optimization result can help the manager to implement the relevant decision. in this paper, we use cross-information technology, state space time series, and other possible combination algorithms, improve the backtracking search algorithm, and Simio system simulation technology to study the logistics system of the railway container central station, from the regional freight demand forecast, the gate system congestion optimization, The intelligent control of the system is realized in the aspects of the scheduling of the crane and the decision of the allocation of the box position and the optimal configuration of the system resources. The research of this paper has important theoretical and practical value, and lays the foundation for future research.
【学位授予单位】：西南交通大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：U294.3

【参考文献】