Deep Learning-Assisted Microscopic Polarization Inspection of Micro-Nano Damage Precursors: Automatic, Non-Destructive Metrology for Additive Manufacturing Devices

Area of Science:

• Advanced Manufacturing
• Materials Science
• Artificial Intelligence

Background:

• Additive Manufacturing (AM) offers advantages for complex component fabrication but faces challenges in quality monitoring.
• Industrial-grade precision requires addressing multi-scale damage precursors, noise interference, and limited training data for intelligent systems.

Purpose of the Study:

• To develop an automated, non-destructive testing (NDT) methodology for detecting micro-nano damage precursors in AM.
• To overcome existing technical bottlenecks in intelligent AM quality monitoring systems.

Main Methods:

• Proposed an innovative microscopic polarization YOLOv11-LSF intelligent inspection framework.
• Implemented a multi-scale perception module with Large Separable Kernel Attention for enhanced feature detection.
• Designed a Slim-neck architecture using VoV-GSCSP module and GSConv for reduced model complexity.
• Developed a simulation strategy with physical features for integrated virtual and real training samples.

Main Results:

• Achieved accuracy (P) increase of 1.6%, recall (R) of 1.6%, mAP50 of 1.5%, and mAP50-95 of 2.8% compared to the baseline.
• Reached 99% detection accuracy for porosity-related micro-nano damage precursors and 94% for cracks.
• Demonstrated robust performance with small sample adaptation and under complex conditions.

Conclusions:

• The YOLOv11-LSF framework provides a reliable technical solution for industrial-grade AM quality monitoring.
• This research advances smart manufacturing quality innovation and enables cross-scale micro-nano damage inspection.