Impact monitoring on complex structure using VMD-MPE feature extraction and transfer learning

Area of Science:

• Aerospace Engineering
• Structural Health Monitoring
• Signal Processing

Background:

• Aircraft structural integrity is critical for safe operation, necessitating effective impact monitoring.
• Traditional methods face challenges with large structures and sensor distances.
• Impact events can compromise aircraft safety.

Purpose of the Study:

• To develop a region-to-point monitoring method for aircraft impact detection.
• To address challenges of large monitored areas and long sensor distances.
• To accurately locate and quantify impact events on aircraft structures.

Main Methods:

• Variational Mode Decomposition (VMD) for signal analysis.
• Multi-scale Permutation Entropy (MPE) for feature extraction.
• Transfer Component Analysis (TCA) for domain adaptation.
• Probabilistic Neural Network (PNN) for classification.
• System identification and weighted centroid algorithm for precise localization.

Main Results:

• Effective decomposition of impact signals into distinct frequency modes.
• Successful reduction of feature discrepancies between source and target domains using TCA.
• Accurate identification of impact areas using PNN.
• Precise determination of impact force history and coordinates.

Conclusions:

• The proposed region-to-point method effectively monitors aircraft impacts.
• The integrated approach enhances the accuracy of impact localization and force estimation.
• This technology contributes to improved aviation safety through advanced structural monitoring.