集装箱码头闸口交通需求智能预测研究

发布时间：2018-01-08 16:08

本文关键词：集装箱码头闸口交通需求智能预测研究　出处：《河北工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：集装箱码头闸口是集卡进出码头的必经关口,是港区集疏运作业的关键节点和瓶颈。其开设的不合理,不仅使客户长时间排队等待,也造成由于排队长度过长交通量过大而使港口道路拥堵严重,同时给码头企业的运输作业形成巨大的压力,使得集装箱码头企业由于这种需求的不确定性而使运输组织无法合理安排,不均衡性更加严重,企业经济损失严重,因此对集装箱码头闸口交通需求进行预测具有重要的理论和工程应用价值。同时,虽然集装箱码头闸口交通需求具有非线性特点,但集装箱码头企业运输作业随船期表的特性,又使其运输组织特别是闸口交通需求呈现出一定的规律性、内随机性。如何利用现代信息智能技术和其高度非线性特点,对闸口交通需求进行预测研究,成为本研究的出发点。本论文主要工作可概括为以下几方面:(1)港区集疏运交通系统分析。在讨论系统基本要素、系统环境、交通特点等基础上,设计了包括信息子系统、历史信息数据库、知识库、在线预测子系统、离线预测子系统等构成的集装箱码头闸口交通智能预判系统,并对各子系统的功能、原理以及智能预判系统的功能、原理予以说明,为集装箱码头闸口交通需求的智能预测提供基础平台和工程应用可能。(2)基于曲线拟合和SVM的码头闸口交通需求预测研究。运用概率分布拟合方法,建立基于历史信息的集装箱码头闸口的概率分布模型,并采用SVM方法对每一班船集卡数量进行预测,从而对各个时间段的集卡车数量进行预测,并通过具体实例予以验证。(3)基于实时信息的码头闸口交通需求预测方法。在PDFM方法确定概率分布的基础上,建立了一种基于实时信息的概率修正预测模型,通过实例进行验证,显示出该预测方法的高精度性。(4)基于季节性ANN的码头闸口交通需求预测方法。在集装箱码头闸口交通需求源于船期表具有季节性和非线性特点系统分析的基础上,提出基于每条班线船期表来预测其对码头闸口产生交通需求的思想,采用季节性时间序列方法处理集港车辆到达码头闸口随时间的数量分布,建立处理后的时间序列数据与预测交通量之间非线性关系的人工神经网络模型。在天津港集装箱码头闸口进行具体例子应用,证明了该方法优于概率分布拟合方法和基于实时信息的概率修正预测方法,显示其可行性。
[Abstract]:The gate of container terminal is the key node and bottleneck of collecting card entering and leaving wharf, and it is not only the unreasonable opening of container terminal, but also makes customers wait in line for a long time. It also causes the port roads to be congested seriously because of the excessive traffic volume of the long queue, and at the same time, it forms a huge pressure on the transport operations of the wharf enterprises. Because of the uncertainty of the demand, the container terminal enterprise can not arrange the transportation organization reasonably, the imbalance is more serious, and the economic loss of the enterprise is serious. Therefore, the prediction of container terminal gate traffic demand has important theoretical and engineering application value, at the same time, although container terminal gate traffic demand has nonlinear characteristics. However, the characteristics of the shipping schedule of container terminal enterprises make the transportation organization, especially the traffic demand of gate, show certain regularity. Internal randomness. How to use modern information intelligence technology and its highly nonlinear characteristics to predict the traffic demand of sluice gates. The main work of this paper can be summarized as follows: 1) Port area transportation system analysis. On the basis of discussing the basic elements of the system, system environment, traffic characteristics and so on. The intelligent prejudgment system of container terminal gate traffic is designed, which includes information subsystem, historical information database, knowledge base, on-line prediction subsystem and off-line prediction subsystem. The functions of each subsystem are also discussed. The principle and function of intelligent prejudgment system are explained. This paper provides a basic platform and engineering application for intelligent prediction of gate traffic demand of container terminal. (2) based on curve fitting and SVM, the traffic demand prediction of terminal gate is studied. The probability distribution fitting method is used. The probability distribution model of container terminal gate based on historical information is established, and the SVM method is used to predict the number of container trucks in each time period. The method of traffic demand prediction based on real-time information is verified by an example. Based on the PDFM method, the probability distribution is determined. A probabilistic modified prediction model based on real-time information is established and verified by an example. It shows the high accuracy of the prediction method. Based on seasonal ANN, the traffic demand forecast method of terminal gate is based on the systematic analysis of the seasonal and nonlinear characteristics of the container terminal gate traffic demand derived from the seasonality and nonlinear characteristics of the ship schedule. This paper puts forward the idea of forecasting the traffic demand for the gate of the wharf based on the schedule of each shift line, and adopts the seasonal time series method to deal with the distribution of the number of vehicles arriving at the terminal with time. An artificial neural network model of the nonlinear relationship between the time series data and the traffic volume is established. The model is applied to the gate of Tianjin Port Container Terminal. It is proved that this method is superior to the probability distribution fitting method and the probability correction prediction method based on real time information.
【学位授予单位】：河北工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491;U691

【参考文献】