AI-Powered Predictive Analytics for Demand Forecasting in U.S. Aerospace Manufacturing: Enhancing Operational Efficiency

Abstract

In aerospace manufacturing, demand forecasting represents one of the most difficult challenges of the industry in the twenty-first century. Aeronautics is a low-volume, high-value production culture, where large-scale decisions cannot be retracted after being made. An error in decisions to build a huge manufacturing system will cost billions of dollars and take many years to recover [1]. In addition to the large capital investment, aerospace production planning should avoid the wastage of very expensive components and resources. Thus, conducting an analysis of future market demand and risks is essential before making mass-production decisions.

The demand for aircraft is highly variable and the past market histories for each aircraft model are quite different. There have been several methodology developments for demand forecasting, including both probabilistic methods and non-probabilistic methods [2]. However, the demand forecast remains a challenge for aviation production planning because current forecasting methods cannot handle: (1) Dependency between observations (2) Different volatility starting points for the same observation.