Investigating Robust and Stochastic Optimisation Methods For Automated Container Terminals

港口业在英国经济中扮演着战略角色。人们对集装箱码头的自动化越来越感兴趣，特别是来自欧洲港口的兴趣，以此作为提高其效率和可靠性的一种方式。在前壳牌港湾旧址上新建伦敦门户集装箱码头的计划设想了迄今为止最高的自动化程度，为英国提供了一个独特的机会，可以在自动化软件领域开发具有战略重要性的专业知识。这项建议旨在开发基于稳健和随机优化方法的创新方法，以控制码头内集装箱的自动水平和垂直移动，并在原型控制软件中实施这些方法。稳健优化包括定义关键参数的间隔，如集装箱移动的开始时间和持续时间，然后寻找关于参数实现的最大遗憾最小化的解决方案。这种优化通过滚动的计划范围将未来的事件考虑在内。作为稳健优化的替代方案，随机优化通过情景分析来处理不确定性。在该方法中，与系统组件/参数相关的不确定性由从基础优化问题导出的多个场景(或子问题)建模，并且信息结构可以被表示为树，每个场景对应于从树根到树叶的路径。由于稳健优化问题和随机优化问题在一般情况下都是NP难的，因此将设计启发式方法以允许实时实施。原型控制软件在提供给英国港口运营商之前将在模拟中进行测试。

The port industry plays a strategic role in the UK economy. There has been growing interest, particularly from European ports, in the automation of container terminals as a way to improve their efficiency and reliability. Plans for the new London Gateway container terminal on the former Shell Haven site envisage the highest degree of automation to date, providing a unique opportunity for the UK to develop strategically important expertise in automation software. This proposal seeks to develop innovative approaches based on robust and stochastic optimisation methods to control the automated horizontal and vertical movement of containers within the terminal and the implementation of these methods in prototype control software. Robust optimisation involves the definition of intervals for key parameters, like the start time and duration for container movements, and then looks for a solution that minimises maximum regret with respect to realisations of parameters. The optimisation takes future events into account through a rolling planning horizon. As an alternative to robust optimisation, stochastic optimisation deals with the uncertainty via a scenario analysis. In this approach, the uncertainty related to system components / parameters is modelled by a number of scenarios (or sub-problems) derived from an underlying optimization problem and the information structure can be represented as a tree, with each scenario corresponding to a path from the root of the tree out to a leaf. As both the robust and stochastic optimisation problems are in general NP-hard, heuristic methods will be devised to allow implementation in real time. The prototype control software will be tested in simulation before being made available to UK port operators.