Automated visual detection of geometrical defects in composite manufacturing processes using deep convolutional neural networks

Djavadifar, Abtin; Graham-Knight, John Brandon; Körber, Marian; Lasserre, Patricia; Najjaran, Homayoun

doi:10.1007/s10845-021-01776-1

Cited by 19 publications

(10 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This value represents the 10% of the maximum number of epochs and it is consistent with the values reported in previous studies. 43,47,72 The batch size was one, as reported in previous studies, 11,13,19 allowing the maximisation of the frequency at which the weights are updated during each epoch. The combination of hyperparameters values and the model state (weight values) at the epoch minimising the control set loss was saved (Table 2) and used for a further segmentation of voids and dry areas in the Validation Sample scan.…”

Section: Cnn Hyperparameters Selectionmentioning

confidence: 99%

“…20 The decision process regarding selecting the hyperparameters has been documented in only a couple of studies investigating composite, 10,21 but both lack a formal justification of the choice of hyperparameters, or followed previously reported recommendations. 19,20 The primary objective of this current study was to investigate the relationship between the annotation effort, computing cost and segmentation performance of Deep Learning models. For this reason, a range of state-of-the-art CNN architectures were used to segment the interlaminar voids and dry areas of CT Scan micrographs of uncured composite laminates.…”

Section: Introductionmentioning

confidence: 99%

“…As a consequence of the ample choice of CNN networks, authors have compared the performance of a selection of architectures for different segmentation tasks, such as wrinkle detection in digital images 19 or fibre segmentation in ceramic-matrix composites from high-resolution X-Ray micrographs. 20 The decision process regarding selecting the hyperparameters has been documented in only a couple of studies investigating composite, 10,21 but both lack a formal justification of the choice of hyperparameters, or followed previously reported recommendations.…”

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

The effect of convolutional neural network architectures on phase segmentation of composite material X-ray micrographs

Galvez-Hernandez

Kratz

2023

Journal of Composite Materials

View full text Add to dashboard Cite

Porosity severely reduces the mechanical performance of composite laminates and methods for automatic segmentation of void phases are growing. This study investigates porosity in composite materials that take the form of interlaminar voids and dry tow areas. Deep Learning was used for the segmentation of X-ray micrographs via the implementation of eight state-of-the-art Convolutional Neural Network (CNN) architectures trained with data sets containing twenty-five, fifty, and one-hundred images. The combination of hyperparameters providing the highest accuracy for each architecture and training set size was achieved through the optimisation of six relevant hyperparameters, including the cut-off probability applied to output probability maps. Additionally, the properties of the CNN architectures ( e.g., layer typology, connections, density…) were found to play a determining role, not only in the segmentation results but also in the associated computing effort. U-Net and FCDenseNet outperformed the FCN-8s, FCN-16, SegNet, LinkNet, ResNet18 and Xception CNN architectures. However, the CNNs generally outperformed the standard thresholding approaches, especially in sub-volumes containing low porosity (1.07%) where the influence on strength is very sensitive in high-performance composites. In low porosity samples, U-Net and FCDenseNet consistently segmented voids to 85% + accuracy, whereas thresholding was only half as accurate, at around 40%. The results provide a strong motivation to replace thresholding as a segmentation method for composite X-ray micrographs. In terms of efficiency, the reduced complexity of the U-Net network allowed for an average reduction of the training time (−36%) and prediction time (−17%) when compared to FCDenseNet.

show abstract

Section: Cnn Hyperparameters Selectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

The effect of convolutional neural network architectures on phase segmentation of composite material X-ray micrographs

Galvez-Hernandez

Kratz

2023

Journal of Composite Materials

View full text Add to dashboard Cite

show abstract

“…improved fatigue performance (Shanmugam et al , 2021). The combination of both lightweight design and materials in composite cellular structures intrigued wide interests in fields such as aerospace and nuclear engineering (Djavadifar et al , 2021; Giusto et al , 2021; Liu et al , 2020). Thus, the past decades have witnessed an increasing number of studies exploring the large design space of composite cellular structures.…”

Section: Introductionmentioning

confidence: 99%

A graph-based path planning method for additive manufacturing of continuous fiber-reinforced planar thin-walled cellular structures

Zhang

Wang

et al. 2022

RPJ

View full text Add to dashboard Cite

Purpose Composite cellular structures have wide application in advanced engineering fields due to their high specific stiffness and strength. As an emerging technology, continuous fiber-reinforced polymer additive manufacturing provides a cost-effective solution for fabricating composite cellular structures with complex designs. However, the corresponding path planning methods are case-specific and have not considered any manufacturing constraints. This study aims to develop a generally applicable path planning method to fill the above research gap. Design/methodology/approach This study proposes a path planning method based on the graph theory, yielding an infill toolpath with a minimum fiber cutting frequency, printing time and total turning angle. More specifically, the cellular structure design is converted to a graph first. Then, the graph is modified to search an Eulerian path by adding an optimal set of extra edges determined through the integer linear programming method. Finally, the toolpath with minimum total turning angle is obtained with a constrained Euler path search algorithm. Findings The effectiveness of the proposed method is validated through the fabrication of both periodic and nonperiodic composite cellular structures, i.e. triangular unit cell-based, Voronoi diagram-based and topology optimized structures. The proposed method provides the basis for manufacturing planar thin-walled cellular structures of continuous fiber-reinforced polymer (CFRP). Moreover, the proposed method shows a notable improvement compared with the existing method. The fiber cutting frequency, printing time and total turning angle have been reduced up to 88.7%, 52.6% and 65.5%, respectively. Originality/value A generally applicable path planning method is developed to generate continuous toolpaths for fabricating cellular structures in CFRP-additive manufacturing, which is an emerging technology. More importantly, manufacturing constraints such as fiber cutting frequency, printing time and total turning angle of fibers are considered within the process planning for the first time.

show abstract

On the Use of Machine Learning for Damage Assessment in Composite Structures: A Review

Ribeiro Junior,

Gomes

2023

Appl Compos Mater

View full text Add to dashboard Cite

Automated visual detection of geometrical defects in composite manufacturing processes using deep convolutional neural networks

Cited by 19 publications

References 48 publications

The effect of convolutional neural network architectures on phase segmentation of composite material X-ray micrographs

The effect of convolutional neural network architectures on phase segmentation of composite material X-ray micrographs

A graph-based path planning method for additive manufacturing of continuous fiber-reinforced planar thin-walled cellular structures

On the Use of Machine Learning for Damage Assessment in Composite Structures: A Review

Contact Info

Product

Resources

About