Damage Localization for Type IV Composite Pressure Vessels Based on YOLOv11 with Dynamic Convolution, EMA and AE Time-Frequency Analysis

Abstract

This work proposes a damage localization framework for Type IV composite pressure vessels using acoustic emission time-frequency analysis and an improved YOLOv11 architecture. Dynamic convolution and EMA attention mechanisms are integrated to enhance feature extraction capability and improve localization accuracy under complex industrial environments. Experimental results demonstrate the effectiveness of the proposed framework for structural health monitoring applications.

Key Contributions

• Proposed an AE-based damage localization framework for composite pressure vessels
• Integrated dynamic convolution and EMA attention mechanisms into YOLOv11
• Improved localization robustness under industrial noise conditions
• Combined time-frequency analysis with deep-learning-based visual detection

@inproceedings{guo2026pressurevessel,
    author    = {Xin-Yu Guo and Liu-Yi Yu and Yan-Zuo Chang},
    title     = {Damage Localization for Type IV Composite Pressure Vessels Based on YOLOv11 with Dynamic Convolution, EMA and AE Time-Frequency Analysis},
    booktitle = {2026 IEEE 8th International Conference on Communications, Information System and Computer Engineering (CISCE)},
    year      = {2026},
    address   = {Guangzhou, China},
    pages     = {748--755},
    doi       = {10.1109/CISCE69494.2026.11504488},
    keywords  = {Acoustic Emission, YOLOv11, Damage Localization, Composite Pressure Vessel},
}

Recommended citation: Xin-Yu Guo, Liu-Yi Yu, and Yan-Zuo Chang, “Damage Localization for Type IV Composite Pressure Vessels Based on YOLOv11 with Dynamic Convolution, EMA and AE Time-Frequency Analysis,” in Proc. IEEE CISCE 2026, Guangzhou, China, 2026.
Download Paper