ML-Assisted Composite Impact Damage Reconstruction

Overview

The Structural Integrated Research Collaboration (SIRCA) program aims to enable on-the-fly risk assessment of impact damage to composite aerospace structures using partial or incomplete non-destructive inspection (NDI) data.

LAMMA focuses on the composite damage modeling component — predicting the delamination pattern in composite laminates due to low-velocity impact (LVI).

Part 1 — Discrete crack modeling

The first part focuses on modeling discrete cracks in the impact region of a composite plate to capture the distinct crack and delamination patterns caused by LVI. These models are used to generate synthetic data for training machine learning algorithms.

Part 2 — ML-assisted inspection data reconstruction

Ultrasound scan images of composite impact damage can be noisy and ambiguous about which layer of the composite is damaged. LAMMA applies machine learning tools to evaluate ultrasound scans and estimate the exact pattern and depth of delamination — enabling accurate, real-time damage assessment from existing USAF inspection tools.

Funding

Air Force Research Lab (AFRL) — SIRCA program.