About this Abstract |
Meeting |
2025 TMS Annual Meeting & Exhibition
|
Symposium
|
Algorithms Development in Materials Science and Engineering
|
Presentation Title |
Explainable Deep Learning Model for Defect Detection during Autoclave Composite Manufacturing |
Author(s) |
Deepak Kumar, Pragathi Agraharam Chan, Yongxin Liu, Sirish Namilae |
On-Site Speaker (Planned) |
Sirish Namilae |
Abstract Scope |
The application of artificial intelligence (AI) in composite manufacturing has enabled new opportunities for automated quality inspection. The practical application of AI in composite processing is currently constrained by the absence of real processing data and deep-learning model capabilities. In this study, we used a custom-built autoclave with viewports and an interior lighting setup to develop a novel dataset of the composite curing process using digital image correlation (DIC). Later, using a unique explainable AI technique, a zero-bias deep neural network (ZBDNN) model was developed by transforming the final dense layer of the standard DNN model into a dimensionality reduction layer and similarity matching layer. ZBDNN model performance was then compared against a one-class support vector machine (OC-SVM) and an autoencoder for abnormality detection. ZBDNN outperformed both models with an abnormality detection accuracy of 99.41, followed by the autoencoder with 91.26, and OC-SVM with 68.38. |
Proceedings Inclusion? |
Planned: |
Keywords |
Composites, Other, Other |