Detecting defects within thermal protection materials, which are integral for spacecraft safety during extreme temperature conditions, has posed significant challenges due to the materials’ complex textures and similarities to background features. A groundbreaking approach, the Texture-Enhanced Attention Defect Detection (TADD) model, has been developed by researchers to streamline this process, achieving impressive results and enhancing the safety standards for aerospace structures.
Thermal protection materials are utilized extensively within aerospace engineering, especially for solid rocket motors and thermal insulation components. Maintaining structural integrity is pivotal, as defects such as cracking and delamination could jeopardize safety. Existing detection methods have demonstrated limitations; they typically require the expertise of seasoned operators, making the processes not only labor-intensive but time-consuming. These traditional methods struggle particularly with identifying concealed defects due to the materials’ similar texture characteristics.
To counter these challenges, the research team constructed the Thermal Protection Material Digital Radiographic (TPMDR) dataset, comprising 670 images and annotated with 6,269 defect instances across six categories, including lamination and crazing. Through this extensive and detailed dataset, the team has set the stage for advanced automated defect detection processes.
The TADD model aims to improve the accuracy and efficiency of defect detection by incorporating innovative techniques. It employs two key components: a texture enhancement module and a non-local dual attention mechanism. The texture enhancement module enhances the intrinsic features of concealed defects, increasing the visibility of their edges against the material background. Meanwhile, the non-local dual attention mechanism focuses on integrating relevant features, effectively reducing feature loss and significantly improving the model's ability to detect tiny and multi-scale defects.
Evaluation results demonstrate the model's capabilities; the TADD achieves 54.74% mean Average Precision (mAP) at 25 frames per second, surpassing previous methods by 11.05%. This remarkable performance highlights both its accuracy and its practicality for real-time applications—a necessity for effective monitoring of aerospace components. "To avoid damage to the structure of special parts due to breakage of thermal protection materials, it is important to perform efficient and real-time monitoring," the authors stated.
This improved detection method holds potential not just for enhancing aerospace safety but also for broader applications within industrial quality control processes. By validating the TADD model against both the TPMDR and public datasets, researchers have underscored its versatility and effectiveness across various contexts.
Overall, the study not only sheds light on the pressing need for automated defect detection technologies but also establishes the TADD model as a significant step forward within this specialized field. With the foundation laid by the TPMDR dataset and the innovations brought forth by the TADD model, future endeavors may focus on refining these techniques and extending their applicability, thereby ensuring higher safety standards across all domains utilizing thermal protection materials.