Reinforcement Learning and Scenario-Based Order for Modeling Enterprise Resilience of Maritime Container Ports

Author: ORCID icon
Loose, Davis, Systems Engineering - School of Engineering and Applied Science, University of Virginia
Loose, Davis, Systems and Information Engineering, University of Virginia

Global logistics systems met a crisis from the pandemic, diminished workforce, supply reductions, and demand surges. Maritime ports in particular are vulnerable to these disruptions. There is a need for methods to address system resilience. This dissertation introduces the cyber-physical systems requirements methodology (CPSRM), an approach for developing resilience of cyber-physical systems to disruptions. The CPSRM and associated tools are demonstrated in four parts on a maritime port and surrounding region as follows. First, it describes an approach to the development of a system specification as well as a hazard and gap analysis of resilience techniques. Second, it describes a mathematical simulation to account for key factors, focusing on bottlenecks in the supply chain. Third, it adapts reinforcement learning to understand and control these processes in scenarios of disruption. Fourth, it describes how to manage the disruption of system orders by the scenarios. The CPSRM improves on existing methods by incorporating particular tools from cybersecurity and risk analysis; a) red and blue team exercises for the negotiation of system requirements and b) quantification of risk as the degree of order disruption. The approach is of interest across topics of systems engineering, particularly for requirements elicitation, gap analysis, modeling and simulation, reinforcement learning, performance evaluation, and risk analysis. Practitioners will benefit by using the CPSRM to design and evaluate alternatives for system resilience.

PHD (Doctor of Philosophy)
Systems Engineering, risk analysis, MuZero, scenario-based preferences, enterprise systems, modeling and simulation, augmented intelligence , hazard analysis
Issued Date: