Enabling Security and Risk-based Operation of Container Line Supply Chains (CLSCs) under High Uncertainties

在过去的几年里，越来越多的国际社会认识到，迫切需要审查集装箱班轮供应链等海运系统的安全和风险问题。2001年9·11恐怖袭击、2002年美国西海岸港口封锁、2004年马德里通勤列车爆炸、2005年伦敦通勤巴士和地铁爆炸等严重事故震惊了整个国际航运和物流供应业，促使这一紧迫性。CLSC拥有许多复杂的实物和信息流，不仅促进了经济繁荣，而且使自己特别容易受到许多风险的影响，从货物交付延误到环境污染，从恐怖袭击到破坏经济稳定。安全性正在成为衡量海洋系统设计、控制和管理性能的最重要标准之一。安全一词一般可被定义为免于脆弱性，即暴露在威胁引起的严重动乱中。在本研究中，与威胁相关的风险将称为安全风险。虽然传统的基于风险的风险是不良事件发生的概率及其可能后果的程度的组合，但安全风险不同于基于风险的风险，需要以不同的方式建模。因此，安全和风险评估是分析系统中的威胁和危险，并针对系统的潜在脆弱性就适当的战略作出各自决定的过程。以前在这一领域和相关领域的研究极大地增加了我们对脆弱性、风险、威胁和危险的理解。然而，很少有研究从工程学和管理学的角度为CLSC的安全和风险研究提供适当的支持工具。该项目旨在制定一个基于安全和风险的框架以及适用于海洋作业的评估模式。为了实现这一目标，需要研究几个具有挑战性的研究问题。首先，不同的安全和风险变量之间的大多数关系可能出现在不同的空间、时间或功能尺度上，如果在动态和交互的细节水平上或两者之间描述每种关系，而不是将静态和稳定的尺度过程等同对待，可能会更好地反映这些关系。在这个项目中，将开发一种结合贝叶斯网络、模糊集和证据推理的新型混合推理网络，称为ER-RN模型，以估计CLSC中威胁和危险的发生可能性。其次，CLSC中用于安全和风险评估的信息本质上是不确定的，原因是对CLSC领域的了解不完善，对领域状态的了解不完整，管理领域行为的机制的随机性，或它们的组合。因此，处理这种不确定的信息是一个巨大的挑战。在这个项目中，将研究一种新的基于信念规则(BRB)的系统方法，以便使用这些不确定信息通过对威胁的破坏能力、回忆难度和损害概率以及危险的可能后果进行建模来估计与威胁和危险相关联的风险。第三，对安全和风险控制措施的评估要求同时考虑多个标准，例如系统风险、实施安全和风险控制措施的成本以及减少风险和货物转运延误带来的好处。在这项研究中，将发展一种多属性决策方法，它可以处理从所提出的ER-RN和BRB模型中产生的具有不确定性的各种信息。将进行案例研究，以演示所建议的网络、模型和分析方法。

Over the past several years, there has been a growing international recognition that security and risk issues of marine systems such as container line supply chains (CLSCs) need to be reviewed urgently. Serious accidents such as the 9/11 terrorist attacks in 2001, the lock-out of the American West Coast Ports in 2002, the blast on the Madrid commuter trains in 2004 and the blast on the London commuter buses and underground trains in 2005 have shocked the whole international shipping and logistics supply industries and prompted this urgency. CLSCs, with many complex physical and information flows, have not only contributed to economic prosperity but also rendered themselves uniquely vulnerable to many risks ranging from delay of cargo delivery to environmental pollution and from terrorist attacks to damage of economic stability. Security is becoming one of the most important criteria for measuring the performance of the design, control and management of marine systems. The term security may in general be defined as freedom from vulnerability which is an exposure to serious disturbances arising from threats. In this research, risks associated with threats will be referred to as security risks. Whilst conventional hazard-based risk is a combination of the probability of occurrence of an undesirable event and the degree of its possible consequences, security risks are different from hazard-based risks and need to be modelled differently. As a result, security and risk assessment is a process of analysing both threats and hazards in a system and making respective decisions on suitable strategies against the potential vulnerability of the system. Previous research in this and related areas has greatly increased our understanding of vulnerability, risks, threats and hazards. However, few studies have generated appropriate supporting tools for security and risk studies in CLSCs from both the engineering and managerial viewpoints. This project is aimed at developing a security and risk-based framework and also assessment models suitable for marine operations. To achieve this aim, several challenging research questions need to be investigated. First of all, most relationships among different security and risk variables may emerge at a variety of spatial, temporal or functional scales, which might be better represented if each relationship were described at or between the dynamic and interactive levels of detail, rather than treating static and steady scale processes identically. In this project, a novel hybrid reasoning network combining Bayesian networks, fuzzy sets and evidential reasoning, referred to as the ER-RN model, will be developed in order to estimate the occurrence likelihoods of threats and hazards in CLSCs. Secondly, information for security and risk assessment in CLSCs is inherently uncertain, caused by imperfect understanding of the domain of a CLSC, incomplete knowledge about the state of the domain, randomness in the mechanisms governing the behaviour of the domain, or a combination of them. It is therefore a great challenge to handle such uncertain information. In this project, a novel belief rule based (BRB) system approach will be investigated in order to use such uncertain information for estimating risks associated with both threats and hazards by modelling the damage capability, recall difficulty and damage probability of threats as well as the possible consequences of hazards. Thirdly, the assessment of security and risk control measures (SRCMs) requires the simultaneous consideration of multiple criteria such as system risk, the costs of implementing a SRCM and the benefits from reduced risk and cargo transfer delay. In this research, a multiple attribute decision-making method will be developed, which can process various types of information with uncertainty generated from the proposed ER-RN and BRB models. Case studies will be conducted to demonstrate the proposed network, models and analysis methods.