Exploratory Research on Engineering the Transport Industries (ETI): An Internet-Based Distributed Architecture for Real-Time Control of Freight Transportation Operations

这个探索性研究工程运输行业（ETI）项目是开发多代理架构，用于协调供应链中出现的团体或组织之间的决策，并支持货运网络。 决策单位可以是大型铁路或卡车运输网络中不同终端的管理人员，也可以是供应链中不同供应商的管理人员。 更好的决策需要更好的信息和更正式的决策过程。 该研究将首先定义用于决策的四大类信息。 将设计一个基于数学的决策函数的一般类，这将形式化如何使用每一类信息。 这些功能将使我们能够研究一个单位的决策如何影响另一个单位的决策。 不确定性的存在，以及缺失或不完整的信息，将明确建模。 我们的期望是实现这些决策代理在一个松散耦合的环境中，使用互联网作为主要的通信手段。 所有测试都将使用主要铁路和卡车运输公司提供的数据进行。如果成功，该架构将改善大型运输和物流网络决策者之间的整体协调，加快网络响应速度，降低成本，提供更好的服务。 该技术将捕获复杂操作中固有的分散决策过程，但将开发一个提供全局优化行为的过程。 这个过程应该产生可以实时跟踪的计划，利用对未来决策的预测（以及对系统其他要素的预测）来做出更好的决策。 我们的期望是深入了解改进信息的价值，这将允许量化更好信息的成本与收益。 正是这种权衡是阻碍在运输和物流信息基础设施的许多要素中使用更先进的信息技术的主要因素，研究应该通过帮助我们了解采用信息技术的经济障碍来加快采用信息技术。

This Exploratory Research on Engineering the Transport Industries (ETI) project is to develop multi-agent architecture for coordinating decision-making among groups or organizations arising within supply chains and supporting freight transportation networks. Decision-making units can be managers at different terminals in a large rail or trucking network, or managers at different suppliers within a supply chain. Better decision making occurs with better information, and a more formalized decision-making process. The research will first define the four major classes of information that are used in decision making. A general class of mathematically based decision functions will be designed, which will formalize how each class of information is used. These functions will allow us to study how decisions by one unit impact the decisions of another unit. The presence of uncertainty, and missing or incomplete information, will be explicitly modeled. The expectation is to implement these decision agents in a loosely coupled environment, using the Internet as the primary means of communication. All testing will be done using data already available from major railroads and trucking companies.If successful, the architecture will improve overall coordination between decision-makers in large transportation and logistics networks, accelerating network response, with lower cost and better service. The technology will capture the inherently decentralized process of decision making in complex operations, but will develop a process, which offers global optimizing behavior. The process should produce plans that can be followed in real-time, using forecasts of decisions in the future (and on other elements of the system) to make better decisions now. The expectation is to develop insights into the value of improved information, which will allow quantifying the cost of better information against the benefits. It is this tradeoff that is the primary deterrent against the use of more advanced information technologies in many elements of the transportation and logistics information infrastructure, and the research should speed adoption of information technologies by helping us understand the economic barriers to adoption.