Study on intelligent and visualization method of ultrasonic testing of composite materials based on deep learning

•The DiMP object tracking model is improved by using Wasserstein distance to realize the visual localization of the ultrasonic probe.•A 1DCNN classification network is designed to classify one-dimensional ultrasonic signals.•A data connection is proposed to make the two models work together.•The pro...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Applied acoustics Ročník 207; s. 109363
Hlavní autori: Hu, Qichun, Wei, Xiaolong, Guo, Hanyi, Xu, Haojun, Li, Caizhi, He, Weifeng, Pei, Binbin
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Elsevier Ltd 01.05.2023
Predmet:
Composite materials
Deep learning
Non-destructive testing
Ultrasonic flaw detection
PrDiMP
Deep learning
Composite materials
D3S
DiMP
Non-destructive testing
Ultrasonic flaw detection
ATOM
KCF
ISSN:0003-682X, 1872-910X
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:•The DiMP object tracking model is improved by using Wasserstein distance to realize the visual localization of the ultrasonic probe.•A 1DCNN classification network is designed to classify one-dimensional ultrasonic signals.•A data connection is proposed to make the two models work together.•The proposed method is embedded in the control framework of ultrasonic A-scan equipment, and preliminary engineering applications are carried out. Intelligent ultrasonic testing technology of composite materials can greatly reduce the dependence on people and improve the efficiency of ultrasonic testing. The combination of ultrasonic testing and visual positioning technology can realize strong robust visual interpretation of ultrasonic testing results. In this paper, the DiMP tracking model is improved by using the Wasserstein distance, and the intelligent tracking and positioning of ultrasonic probe is realized. At the same time, an ultrasonic signal classification network based on 1DCNN depth neural network is built to realize the intelligent detection of ultrasonic signals, and an effective data connection mode is designed to make the two networks work together, so that the intelligent interpretation and visual display of internal defects of composite materials can be realized. The experimental results show that the interpretation accuracy of the method proposed in this paper reaches 98.74%, and the Kappa coefficient reaches 0.97. The comparison results with other models show that the improved model in this paper is more excellent, and the AUC and Precision values are increased by 6.4% and 8.32% respectively compared with the benchmark.
ISSN:0003-682X
1872-910X
DOI:10.1016/j.apacoust.2023.109363