https://sam.ensam.eu:443 The DSpace digital repository system captures, stores, indexes, preserves, and distributes digital research material. Mon, 20 Apr 2026 00:12:23 GMT 2026-04-20T00:12:23Z http://hdl.handle.net/10985/26881 Automated classification of subsurface impact damage in thermoplastic composites using depth-resolved terahertz imaging and deep learning SILITONGA, Dicky J.; POMARÈDE, Pascal; BAWANA, Niyem M.; SHI, Haolian; DECLERCQ, Nico F.; CITRIN, D. S.; MERAGHNI, Fodil; LOCQUET, Alexandre Reliable detection of barely visible impact damage is critical to ensure the structural integrity of composite components in service, particularly in safety-critical applications such as pressure vessels and transportation systems. This study presents a solution for detecting such damage in woven glass fiber-reinforced thermoplastic composites using terahertz (THz) time-of-flight tomography and convolutional neural networks. THz provides non-contact, non-ionizing, high-axial-resolution imaging of subsurface and back-surface damage, addressing key limitations of surface-based inspection methods. While THz imaging alone may not always permit conclusive damage identification, we bridge this gap by training neural network classifiers on depth-resolved THz B-scan images using ground truth from co-located X-ray micro-computed tomography. Among several pretrained architectures tested via transfer learning, DenseNet-121 exhibits the highest accuracy. The model remains robust even when trained on truncated B-scans excluding surface indentation features, confirming its ability to detect structural anomalies located internally or on the back surface. This is particularly relevant for applications where back-side access is not feasible. Experimental validation is performed on impacted glass-fiber-reinforced thermoplastic coupons prepared in accordance with ASTM D7136, with damage severity quantified through force–displacement data and micro-tomographic analysis. Labeling for supervised learning conforms to acceptance criteria from industrial standards for composite pressure vessels (ASME BPVC Section X, CGA C-6.2), ensuring regulatory alignment and enabling deployment in quality control workflows. The proposed method minimizes the need for expert interpretation or secondary validation and offers direct applicability to in-service inspection and manufacturing quality control. Thu, 01 Jan 2026 00:00:00 GMT http://hdl.handle.net/10985/26881 2026-01-01T00:00:00Z SILITONGA, Dicky J. POMARÈDE, Pascal BAWANA, Niyem M. SHI, Haolian DECLERCQ, Nico F. CITRIN, D. S. MERAGHNI, Fodil LOCQUET, Alexandre Reliable detection of barely visible impact damage is critical to ensure the structural integrity of composite components in service, particularly in safety-critical applications such as pressure vessels and transportation systems. This study presents a solution for detecting such damage in woven glass fiber-reinforced thermoplastic composites using terahertz (THz) time-of-flight tomography and convolutional neural networks. THz provides non-contact, non-ionizing, high-axial-resolution imaging of subsurface and back-surface damage, addressing key limitations of surface-based inspection methods. While THz imaging alone may not always permit conclusive damage identification, we bridge this gap by training neural network classifiers on depth-resolved THz B-scan images using ground truth from co-located X-ray micro-computed tomography. Among several pretrained architectures tested via transfer learning, DenseNet-121 exhibits the highest accuracy. The model remains robust even when trained on truncated B-scans excluding surface indentation features, confirming its ability to detect structural anomalies located internally or on the back surface. This is particularly relevant for applications where back-side access is not feasible. Experimental validation is performed on impacted glass-fiber-reinforced thermoplastic coupons prepared in accordance with ASTM D7136, with damage severity quantified through force–displacement data and micro-tomographic analysis. Labeling for supervised learning conforms to acceptance criteria from industrial standards for composite pressure vessels (ASME BPVC Section X, CGA C-6.2), ensuring regulatory alignment and enabling deployment in quality control workflows. The proposed method minimizes the need for expert interpretation or secondary validation and offers direct applicability to in-service inspection and manufacturing quality control. http://hdl.handle.net/10985/26961 Detection of Low-Velocity Impact Damage in Woven-Fabric Reinforced Thermoplastic Composite Laminates by Deep-Learning Classification Trained on Terahertz-Imaging Data SILITONGA, Dicky J.; POMAREDE, Pascal; BAWANA, Niyem M.; SHI, Haolian; DECLERCQ, Nico F.; CITRIN, D.S.; MERAGHNI, Fodil; LOCQUET, Alexandre Terahertz (THz) imaging is gaining attention as a nondestructive testing technique for assessing damage due to its high axial resolution and nonionizing nature, presenting a promising alternative to conventional methods such as ultrasound and X-ray imaging. Its practical implementation, however, remains limited by the reliance on expert interpretation and the frequent need for validation using supplementary techniques such as X-ray microcomputed tomography (µCT), particularly for complex damage modes. This study focuses on woven-fabric-reinforced thermoplastic composites subjected to low-velocity impact, which typically causes barely visible impact damage (BVID). The damage is subtle yet critical, potentially leading to failure under subsequent loading. The multilayered and spatially distributed characteristics of BVID make it especially challenging to identify. To overcome these challenges, this work integrates deep learning with pulsed THz time-of-flight tomography (TOFT) imaging to enable automated damage detection in composite laminates. In contrast to existing research that mainly targets delamination using A- or C-scan data, this study emphasizes the detection of low-velocity impact damage by leveraging THz B-scans, which offer nondestructive depth-resolved cross-sectional imaging. The training dataset is labeled by correlating THz TOFT scans with X-ray CT images used as ground truth. A transfer learning approach, based on convolutional neural network (CNN) architectures, is employed for binary classification to distinguish damaged from undamaged regions. The resulting classifier achieves over 95 % accuracy, demonstrating the viability of this method for industrial applications such as quality assurance and in-service inspection of composite structures. Fri, 01 Aug 2025 00:00:00 GMT http://hdl.handle.net/10985/26961 2025-08-01T00:00:00Z SILITONGA, Dicky J. POMAREDE, Pascal BAWANA, Niyem M. SHI, Haolian DECLERCQ, Nico F. CITRIN, D.S. MERAGHNI, Fodil LOCQUET, Alexandre Terahertz (THz) imaging is gaining attention as a nondestructive testing technique for assessing damage due to its high axial resolution and nonionizing nature, presenting a promising alternative to conventional methods such as ultrasound and X-ray imaging. Its practical implementation, however, remains limited by the reliance on expert interpretation and the frequent need for validation using supplementary techniques such as X-ray microcomputed tomography (µCT), particularly for complex damage modes. This study focuses on woven-fabric-reinforced thermoplastic composites subjected to low-velocity impact, which typically causes barely visible impact damage (BVID). The damage is subtle yet critical, potentially leading to failure under subsequent loading. The multilayered and spatially distributed characteristics of BVID make it especially challenging to identify. To overcome these challenges, this work integrates deep learning with pulsed THz time-of-flight tomography (TOFT) imaging to enable automated damage detection in composite laminates. In contrast to existing research that mainly targets delamination using A- or C-scan data, this study emphasizes the detection of low-velocity impact damage by leveraging THz B-scans, which offer nondestructive depth-resolved cross-sectional imaging. The training dataset is labeled by correlating THz TOFT scans with X-ray CT images used as ground truth. A transfer learning approach, based on convolutional neural network (CNN) architectures, is employed for binary classification to distinguish damaged from undamaged regions. The resulting classifier achieves over 95 % accuracy, demonstrating the viability of this method for industrial applications such as quality assurance and in-service inspection of composite structures. http://hdl.handle.net/10985/27143 Automated classification of subsurface impact damage in thermoplastic composites using depth-resolved terahertz imaging and deep learning SILITONGA, Dicky Januarizky; POMAREDE, Pascal; BAWANA, Niyem Mawenbe; SHI, Haolian; DECLERCQ, Nico; CITRIN, David; MERAGHNI, Fodil; LOCQUET, Alexandre Reliable detection of barely visible impact damage is critical to ensure the structural integrity of composite components in service, particularly in safety-critical applications such as pressure vessels and transportation systems. This study presents a solution for detecting such damage in woven glass fiber-reinforced thermoplastic composites using terahertz (THz) time-of-flight tomography and convolutional neural networks. THz provides non-contact, non-ionizing, high-axial-resolution imaging of subsurface and back-surface damage, addressing key limitations of surface-based inspection methods. While THz imaging alone may not always permit conclusive damage identification, we bridge this gap by training neural network classifiers on depth-resolved THz B-scan images using ground truth from co-located X-ray micro-computed tomography. Among several pretrained architectures tested via transfer learning, DenseNet-121 exhibits the highest accuracy. The model remains robust even when trained on truncated B-scans excluding surface indentation features, confirming its ability to detect structural anomalies located internally or on the back surface. This is particularly relevant for applications where back-side access is not feasible. Experimental validation is performed on impacted glass-fiber-reinforced thermoplastic coupons prepared in accordance with ASTM D7136, with damage severity quantified through force–displacement data and micro-tomographic analysis. Labeling for supervised learning conforms to acceptance criteria from industrial standards for composite pressure vessels (ASME BPVC Section X, CGA C-6.2), ensuring regulatory alignment and enabling deployment in quality control workflows. The proposed method minimizes the need for expert interpretation or secondary validation and offers direct applicability to in-service inspection and manufacturing quality control. Thu, 01 Jan 2026 00:00:00 GMT http://hdl.handle.net/10985/27143 2026-01-01T00:00:00Z SILITONGA, Dicky Januarizky POMAREDE, Pascal BAWANA, Niyem Mawenbe SHI, Haolian DECLERCQ, Nico CITRIN, David MERAGHNI, Fodil LOCQUET, Alexandre Reliable detection of barely visible impact damage is critical to ensure the structural integrity of composite components in service, particularly in safety-critical applications such as pressure vessels and transportation systems. This study presents a solution for detecting such damage in woven glass fiber-reinforced thermoplastic composites using terahertz (THz) time-of-flight tomography and convolutional neural networks. THz provides non-contact, non-ionizing, high-axial-resolution imaging of subsurface and back-surface damage, addressing key limitations of surface-based inspection methods. While THz imaging alone may not always permit conclusive damage identification, we bridge this gap by training neural network classifiers on depth-resolved THz B-scan images using ground truth from co-located X-ray micro-computed tomography. Among several pretrained architectures tested via transfer learning, DenseNet-121 exhibits the highest accuracy. The model remains robust even when trained on truncated B-scans excluding surface indentation features, confirming its ability to detect structural anomalies located internally or on the back surface. This is particularly relevant for applications where back-side access is not feasible. Experimental validation is performed on impacted glass-fiber-reinforced thermoplastic coupons prepared in accordance with ASTM D7136, with damage severity quantified through force–displacement data and micro-tomographic analysis. Labeling for supervised learning conforms to acceptance criteria from industrial standards for composite pressure vessels (ASME BPVC Section X, CGA C-6.2), ensuring regulatory alignment and enabling deployment in quality control workflows. The proposed method minimizes the need for expert interpretation or secondary validation and offers direct applicability to in-service inspection and manufacturing quality control.