Leveraging Machine Learning and AI to Identify Alternative Parts to Increase Parts Availability and Improve Fleet Readiness

Abstract

As competition between the United States and near-peer adversaries intensifies, the U.S. Navy faces increasing challenges to its sea dominance. Fleet readiness, backed by superior Naval capabilities, is critical to credibly project U.S. power and deter conflict in the region.

The speed and agility of the U.S. industrial base to maintain operational availability (Ao) is foundational to readiness. However, obsolescence issues such as parts shortages plague weapon systems, negatively impacting Ao. Leveraging artificial intelligence and machine learning (AI/ML) processes to quickly identify potential alternative parts can greatly speed up the time required to identify replacement parts.

Currently, to remedy these issues, engineers must manually scour hundreds of sources and compare a multitude of technical characteristics to identify alternative parts, a time- and labor-intensive process. To address this need, this study developed an LLM-base AI model to quickly compare multiple parts, rank them based on similarity to the part under investigation, and ultimately identify feasible alternatives. The output is a prioritized ranking of parts, based on model-determined similarity of form, fit and function of the parts. The model-recommended parts are then analyzed for current stock on hand to identify the most viable parts that could also be quickly accessed.