Life-Cycle Modeling of Structural Defects via Computational Geometry and Time-Series Forecasting

Area of Science:

• Structural Engineering
• Computational Mechanics
• Data Science

Background:

• Structural integrity assessment relies on evaluating geometric defects over time.
• Current monitoring methods struggle to link defect detection with predictive simulations.
• Predictive modeling of defect evolution is crucial for estimating remaining structural life.

Purpose of the Study:

• To present a new approach for predictive modeling of geometric defects using point cloud data.
• To develop a method for consistent temporal tracking and analysis of defect evolution.
• To enhance structural health monitoring by integrating defect analysis with predictive simulations.

Main Methods:

• Parametrization of point cloud segments using the convex hull algorithm to extract defect features.
• Adaptation of stochastic dynamic models (ARIMA, VAR) to parameterized hull features for modeling evolution.
• Generation of 2D point clouds over simulated life cycles for analysis.

Main Results:

• The convex hull approach provides consistent and accurate representations of defect evolution across various defect types.
• The method demonstrates robustness to noisy measurements.
• Model accuracy is significantly influenced by assumptions about the underlying dynamical process.

Conclusions:

• The developed convex hull method offers a reliable way to track and predict geometric defect evolution in structures.
• Validation on experimental fatigue testing data confirms high accuracy.
• This approach will aid in finite element model updating for predictive structural capacity analysis.