Identification of voids and interlaminar shear strengths of polymer‐matrix composites by optical microscopy experiment and deep learning methodology

Abstract

The internal void defects induced during the manufacturing process of polymer‐matrix composites can significantly degrade the mechanical properties of the composite, particularly the interlaminar shear strength (ILSS). In this study, we developed an innovative integrated methodology based on a deep learning semantic segmentation algorithm, named DeepLabV3+, and a theoretically driven equation to automatically identify voids in optical images and investigate the relationship between the microscopic voids and macroscopic ILSS parameters of the composite laminates. Results suggest that for the best fine‐tuned DeepLabV3+ framework, the corresponding mean pixel accuracy and intersection over union scores on the testing set were 99.84% and 90.82%, respectively, thereby indicating the potential of the generalized trained model. In addition, detailed experiments revealed that the proposed method can successfully obtain the ILSS values of laminates with different void contents. In addition, the ILSS values of the carbon/epoxy laminates decreased by approximately 27% with an increase in the void content from 0.07% to 3.14%.

Automatic void content assessment of composite laminates using a machine-learning approach

“…Nevertheless, the void sizes obtained for all laminate types in our dataset are in agreement with the range of void sizes found in the literature [21,23]. Moreover, plotting the frequency measures of void sizes into an histogram, for each laminate type, it can be observed that the distribution of void sizes follows a Weibull distribution ( Figures 10,11,12), which is also consistent with the reported literature [47].…”

“…Usually, microscopy and X-ray micro-CT techniques are reported to provide a good level of detail, which enables the accurate measurement and parametrization of void characteristics on the smaller length scales [20,22,23]. Due to its simplicity, lower cost and reasonable accuracy and detail, optical microscopy is still a commonly employed imaging technique to conduct void content analyses [1,4,[12][13][14][15][16][17].…”

“…The adoption of machine-learning algorithms to do automatic detection of voids has been reported in the literature for several void assessment techniques, such as X-ray micro-CT [24][25][26], thermography [27] and ultrasound testing [28]. Luo et al used a deep learning framework based on DeepLabV3+, which achieved good void segmentation results in optical microscopy images [23]. However, their results show that the segmentation accuracy of a thresholding algorithm is very close to the one obtained by the deep learning algorithm.…”

Effect Mechanism and Simulation of Voids on Hygrothermal Performances of Composites

“…Voids are the most common type of defects induced by the residual air during the manufacturing process of composites [ 1 ], which significantly affect the hygrothermal aging performances of composites by altering the stress field and moisture field [ 2 , 3 ]. Particularly, void defects are highly sensitive to the moisture under the hygrothermal environment, and they can further decrease the matrix-dominated properties, which can ultimately reduce the service life of composite structures [ 4 , 5 ].…”

Machine learning in electron microscopy for advanced nanocharacterization: current developments, available tools and future outlook

“…As a result, supervised (mostly CNN) approaches are excelling in this job in contrast, fluorescence, and super-resolution microscopy, among others. 502–510 For that, the idea we have widely reviewed on simulating data that resembles the variability of our target micrographs is also valid here. For instance, A. Sekh et al followed the idea of physics-based labelling by simulating fluorescence microscopy images out of the simulated 3D geometries of organelles (Fig.…”