This report investigates the feasibility and limitations of applying Artificial Intelligence (AI) to enhance predictive maintenance for military readiness spares packages (RSPs). The U.S. Air Force currently manages RSPs to ensure 30-day self-sufficiency, balancing the costs of overstocking against the risks of understocking to mission readiness. This study explores whether and when AI methods can improve parts failure analysis, which traditionally relies on probability distribution models. By developing and testing several machine-learning models on historical A-10C aircraft data, this report compares their performance against the USAF's existing optimization and prediction software.

This report is the third in a series focused on AI applications for warfighters, specifically addressing predictive maintenance and its impact on readiness spares packages. The study aims to inform those interested in AI applications and logistics by presenting a data-driven comparison of AI methods against existing USAF systems. The analysis is grounded in practical application, using A-10C aircraft data to evaluate the performance of developed machine-learning models. Its findings are intended to guide decisions on when and how AI can be most effectively integrated into military operations to enhance efficiency and readiness.