Date of Graduation

5-2025

Document Type

Thesis

Degree Name

Bachelor of Science in Mechanical Engineering

Degree Level

Undergraduate

Department

Mechanical Engineering

Advisor/Mentor

Jensen, David

Committee Member

Campbell, Jenn

Abstract

Numerous approaches exist for computational modeling of complex, dynamic systems. One particular option is the use of agent-based modeling, which simulates the behavior of numerous autonomous agents interacting with each other and with the system at large. Current agent-based modeling research has largely focused on studying the system level behavior of a model, as this is generally the main concern for any real world application. However, the characteristics and behaviors of agents deserves consideration as well. The issue with this area of interest lies in the fact that the agent data cannot be as quickly obtained and analyzed as system level data. In order to develop this idea, this paper focuses on utilizing system level data to determine the state of agents in a model. Specifically, it will consider systems where faults have occurred and find the point(s) of failure. A Python model will be developed and study for this purpose, focusing particularly on an electrical network. From this research, it is found that agent- based models can be effective for identifying faults and their source in a system. These models are limited by a lack of learning on the part of agents and could be improved through the implementation of reinforcement learning algorithms.

Keywords

Agent-Based Modeling; Failure Analysis; System Modeling

Additional Files
electrical_model.ipynb (246 kB)
The program developed and used for this research.
Download

Share

COinS