FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3208368997> ?p ?o ?g. }

• W3208368997 startingPage "1" @default.
• W3208368997 abstract "The request in the aerospace industry for novel methods and tools that enable the optimization of the industrial operation is continuous. Considering the rapid growth in global air traffic passenger demand over the last years, the aerospace industry is facing great challenges in the production and maintenance of its products. The optimization of the industrial operations can significantly lower the production costs, accelerate the manufacturing processes and achieve most efficient maintenance of the products, while in parallel comply with the strict safety regulations. The objective of this doctoral thesis is to introduce new metrology tools that can perform inspections in large aerospace components and optimize the material assessment processes. These novel tools implement the technology that is developed in Computer Vision and Machine Learning. The last two decades these two areas of computer science enjoy rapid progress and they have introduced new methods with very broad applications. This technology that is already mature in these field is adapted to the requirements of the aerospace industry and combined with existed methods, in order to produce new metrology tools especially designed for industrial applications. Since this is an industrial thesis and most of the research work has been carried out in an SME, the primary focus is the development of metrology tools that can be integrated to the production line or the assessment analysis processes of the industrial sector. The developed metrology tools can perform inspections and quantitative assessment of material damage in two different scales: Macroscopic level: the aim is to develop tools that can perform inspections in large scale aerospace components. More specifically, the main goal is the design and assembly of a metrology system that performs inspections in specific locations along the large surface of an aerospace component and produce accurate digital representations of the inspected areas. The metrology system is composed of two parts: the translation sub-system and the vision sub-system. The role of the translation sub-system is performed by a robotic arm. The development of a vision sub-system, able to produce accurate three-dimensional models of the component’s surface areas, the integration to the robotic device and finally the definition of the operating methodology of the created metrology system is the focus of this work. Subsequently, the developed metrology system was used to perform measurements in carbon composite components. These measurements were performed in the framework of a collaboration with AIRBUS, in a project that was aiming to improve the composite repairing process followed by the company. Although, the individual elements of the metrology system exist and are documented in the scientific bibliography, the integration of all these elements in one complete metrology system and the definition of a measuring methodology that can be incorporated in the industrial processes is the original contribution of this work. The uniqueness of this system is that it can perform inspections in targeted locations along the surface of large aerospace components in a fully automated manner. Microscopic level: the focus is on the development of tools based on Machine Learning that aim to optimize the techniques currently used in Quantitative Fractography. The Machine Learning methods are divided in supervised, where a pre-trained model is necessary for performing predictions, and unsupervised, where the algorithm can perform predictions without prior training. The supervised methods are implemented in the topographic characterization of the fracture surfaces, while the unsupervised methods are used to cluster fracture surfaces of different samples produced under different experimental conditions. Currently, the topographic characterization of the fracture surfaces relies on manual inspections by trained experts that can identify the different micro-structure modes on the fracture surface and then quantify the area that each mode occupies in the SEM fracture image. The machine learning algorithm developed in this doctoral thesis is based on Convolutional Neural Networks, after being trained achieves to classify the micro-structure modes in fracture surface and quantify their relative appearance with very high accuracy. These methods require human input only during the training process. After the training process, the neural network can make predictions in any new image in a fully automated manner. Thus, these Machine Learning methods can replace the human-based current processes with automated tools that can be easily integrated in the material assessment analysis of the industry. Additionally, an unsupervised machine learning pipeline is introduced for the creation of a tool capable of clustering a dataset of fracture images, produced with different experimental conditions. The tool is implemented on a fracture images dataset of tungsten heavy alloys, with different tungsten mass percentage and manages to accurately group the images that belong to the same sample. This tool does not require any pre-training and it can operate directly on any dataset of fracture images. The clustering accuracy is very high on a dataset of fracture images with very similar textural characteristics. Adding a supervised algorithm that requires minimal training to the initial pipeline of the proposed tool extends its capabilities in classifying the different clusters of the fracture images. The classification accuracy remains very high, while other methods currently used by the industry when applied in the specific dataset are proven to fail in classifying the images according to the different experimental conditions." @default.
• W3208368997 created "2021-11-08" @default.
• W3208368997 creator A5053927457 @default.
• W3208368997 date "2020-01-01" @default.
• W3208368997 modified "2023-09-27" @default.
• W3208368997 title "Novel metrology tools based on artificial intelligence to study damage and fracture in aerospace structures" @default.
• W3208368997 hasPublicationYear "2020" @default.
• W3208368997 type Work @default.
• W3208368997 sameAs 3208368997 @default.
• W3208368997 citedByCount "0" @default.
• W3208368997 crossrefType "journal-article" @default.
• W3208368997 hasAuthorship W3208368997A5053927457 @default.
• W3208368997 hasConcept C105795698 @default.
• W3208368997 hasConcept C117671659 @default.
• W3208368997 hasConcept C127413603 @default.
• W3208368997 hasConcept C146978453 @default.
• W3208368997 hasConcept C162324750 @default.
• W3208368997 hasConcept C165556158 @default.
• W3208368997 hasConcept C167740415 @default.
• W3208368997 hasConcept C195766429 @default.
• W3208368997 hasConcept C201995342 @default.
• W3208368997 hasConcept C33923547 @default.
• W3208368997 hasConcept C41008148 @default.
• W3208368997 hasConcept C526921623 @default.
• W3208368997 hasConcept C78519656 @default.
• W3208368997 hasConcept C82753439 @default.
• W3208368997 hasConcept C99862985 @default.
• W3208368997 hasConceptScore W3208368997C105795698 @default.
• W3208368997 hasConceptScore W3208368997C117671659 @default.
• W3208368997 hasConceptScore W3208368997C127413603 @default.
• W3208368997 hasConceptScore W3208368997C146978453 @default.
• W3208368997 hasConceptScore W3208368997C162324750 @default.
• W3208368997 hasConceptScore W3208368997C165556158 @default.
• W3208368997 hasConceptScore W3208368997C167740415 @default.
• W3208368997 hasConceptScore W3208368997C195766429 @default.
• W3208368997 hasConceptScore W3208368997C201995342 @default.
• W3208368997 hasConceptScore W3208368997C33923547 @default.
• W3208368997 hasConceptScore W3208368997C41008148 @default.
• W3208368997 hasConceptScore W3208368997C526921623 @default.
• W3208368997 hasConceptScore W3208368997C78519656 @default.
• W3208368997 hasConceptScore W3208368997C82753439 @default.
• W3208368997 hasConceptScore W3208368997C99862985 @default.
• W3208368997 hasLocation W32083689971 @default.
• W3208368997 hasOpenAccess W3208368997 @default.
• W3208368997 hasPrimaryLocation W32083689971 @default.
• W3208368997 hasRelatedWork W2108987700 @default.
• W3208368997 hasRelatedWork W2502319666 @default.
• W3208368997 hasRelatedWork W2536228246 @default.
• W3208368997 hasRelatedWork W2568102700 @default.
• W3208368997 hasRelatedWork W2609898049 @default.
• W3208368997 hasRelatedWork W2783336473 @default.
• W3208368997 hasRelatedWork W2975372179 @default.
• W3208368997 hasRelatedWork W2979282337 @default.
• W3208368997 hasRelatedWork W3007934944 @default.
• W3208368997 hasRelatedWork W3027092765 @default.
• W3208368997 hasRelatedWork W3028286648 @default.
• W3208368997 hasRelatedWork W3033962174 @default.
• W3208368997 hasRelatedWork W3039792520 @default.
• W3208368997 hasRelatedWork W3088170195 @default.
• W3208368997 hasRelatedWork W3092912397 @default.
• W3208368997 hasRelatedWork W3165560739 @default.
• W3208368997 hasRelatedWork W3183276810 @default.
• W3208368997 hasRelatedWork W3203127079 @default.
• W3208368997 hasRelatedWork W83351448 @default.
• W3208368997 hasRelatedWork W856056358 @default.
• W3208368997 isParatext "false" @default.
• W3208368997 isRetracted "false" @default.
• W3208368997 magId "3208368997" @default.
• W3208368997 workType "article" @default.