A damage identification method for aviation structure integrating Lamb wave and deep learning with multi-dimensional...

Area of Science:

• Aerospace Engineering
• Materials Science
• Artificial Intelligence

Background:

• The aerospace industry requires advanced structural health monitoring (SHM) for complex aviation structures.
• Existing methods struggle with multi-dimensional damage information extraction from sensing signals.

Purpose of the Study:

• To develop an effective method for damage identification in aviation structures.
• To improve the localization and quantification of damage using Lamb waves and deep learning.

Main Methods:

• Lamb wave signals were collected from aircraft structures.
• Signals were processed into 1D (damage index) and 2D (Gramian Angular Field) features.
• A deep learning model with multi-dimensional feature fusion (1D and 2D branches with Inception-v1, BiLSTM) was developed.

Main Results:

• The method successfully located and quantified single-source offset and multi-source cumulative damage.
• Transfer learning enabled accurate damage identification across different sensor arrays with reduced data and time.

Conclusions:

• The proposed multi-dimensional feature fusion method enhances spatial and temporal damage representation.
• The approach demonstrates significant accuracy and robustness in aviation structure damage identification.