SHORT DESCRIPTION

Machine learning for correlation between ultrasonic and X-ray tomography data for composite materials


OTHER DETAILS

Ref. num. Ayudante_de_Investigacion_FS_01

Ayudante Investigacion: ultrasonic and X-ray tomography data for composite materials


IMDEA Materials Institute is a public research organization founded in 2007 by Madrid’s regional government to carry out research of excellence in Material Science and Engineering by attracting talent from all over the world to work in an international and multidisciplinary environment. IMDEA Materials has grown rapidly since its foundation and currently includes more than 120 researchers from 22 nationalities and has become one of the leading research centers in materials in Europe which has received the María de Maeztu seal of excellence from the Spanish government. The research activities have been focused on the areas of materials for transport, energy, and health care and the Institute has state-of-the-art facilities for processing, characterization and simulation of advanced materials. More information can be found at https://materials.imdea.org/

DESCRIPTION

 (English)

The presence of porosity in carbon fibre reinforced composite materials is a common manufacturing defect that could endanger the in-service performance of components. To ensure quality standards, the industry relies on ultrasonic non-destructive testing due to its cost and ease of use. However, to date, ultrasonic methods have not been able to assess porosity levels independently of other attributes such as pores morphology, size, and distribution. One possible solution is to address the problem of relating porosity with ultrasonic propagation utilizing data-driven methodologies. The use of machine learning models could discover the hidden patterns of the interaction between the porosity and the ultrasound wave. A possible solution relates features of the ultrasound wave with porosity characteristics obtained from three-dimensional (3D) reconstructed XCT volumes of carbon fiber composites via machine learning models. We tested some machine learning models that improve the prediction of pore volume fraction by using these data.

In this project, the candidate will test and optimize new machine learning models including (but not restricted to) k-neighbours, support vector machines, ensemblings, and neural networks.

On the other hand, X-ray tomography is by far the best technique non-destructive damage assessment in composite materials, being able to identify in 3D manufacturing defects as well as damage generated upon external forces. Thus, part of the job is to take advantage of this non-destructive technique for the determination of damage evolution (cracks, delamination, fibre fracture, etc.) in sequential and in situ testing using XCT. For this purpose, the project will use both laboratory and synchrotron X-ray tomography techniques. Machine learning techniques will be used for the identification of damage.

The work involves material testing, ultrasonic inspection, X-ray characterization, data analysis and programming. Therefore, a high interest in programming is mandatory. Some programming knowledge (preferable in python) is desirable, as well as in machine learning techniques, data visualization, image analysis.

(Español)

La presencia de porosidad en los materiales compuestos reforzados con fibra de carbono es un defecto de fabricación común que podría poner en peligro el rendimiento en servicio de los componentes. Para garantizar los estándares de calidad, la industria se basa en pruebas ultrasónicas no destructivas debido a su costo y facilidad de uso. Sin embargo, hasta la fecha, los métodos ultrasónicos no han podido evaluar los niveles de porosidad independientemente de otros atributos como la morfología, el tamaño y la distribución de los poros. Una posible solución es abordar el problema de relacionar la porosidad con la propagación ultrasónica utilizando metodologías basadas en datos. El uso de modelos de aprendizaje automático podría descubrir los patrones ocultos de la interacción entre la porosidad y la onda de ultrasonido. Una posible solución relaciona las características de la onda de ultrasonido con las características de porosidad obtenidas a partir de volúmenes XCT reconstruidos tridimensionales (3D) de compuestos de fibra de carbono a través de modelos de aprendizaje automático. Hemos ya probado algunos modelos de aprendizaje automático que mejoran la predicción de la fracción de volumen de poros mediante el uso de estos datos.

En este proyecto, el candidato probará y optimizará nuevos modelos de aprendizaje automático que incluyen (pero no se limitan a) k-vecinos, máquinas de vectores de soporte, conjuntos, y redes neuronales.

Por otro lado, la tomografía de rayos X (XCT) es, con diferencia, la mejor técnica no destructiva de evaluación de daños en materiales compuestos, pudiendo identificar en 3D defectos de fabricación como así también daños generados por fuerzas externas. En este contexto, parte del trabajo es usar las ventajas de esta técnica no destructiva para la determinación de la evolución del daño (fisuras, delaminación, fractura de fibras, etc.) en ensayos secuenciales e in situ mediante XCT. Para ello, el proyecto utilizará técnicas de tomografía de rayos X tanto de laboratorio como de sincrotrón. Se usarán técnicas de inteligencia artificial para la detección de daño.

El trabajo involucra ensayos de materiales, inspección ultrasónica, caracterización de rayos X, análisis de datos y programación. Por lo tanto, un alto interés en la programación es obligatorio. Es deseable algún conocimiento de programación (preferiblemente en python), así como en técnicas de aprendizaje automático, visualización de datos, análisis de imágenes.

REQUIREMENTS


M.Sc., BSc. degree in material science, computer science, electronics, telecommunications, applied mathematics, or related field.

Fluent in English.

Interest, not knowledge, in programming, as it is a strong part of the project.

Desirable requirements:

  • Programming knowledge in any language. Preferably Python for compatibility with already developed work.
  • Notions of machine learning and data analysis.
  • Image and signal treatment.
  • Concern about technology, mainly digital.

Candidates must be:

  • Be over 16 years of age and under 30 years of age.
  • Be registered as an unemployed job seeker in one of the employment offices of the Community of Madrid.

The curriculum assessment will be carried out using the blind curriculum vitae procedure, ensuring the principle of non-discrimination. Candidates must ONLY submit their CV (in English) following precisely these indications:

  • It must not include any personal data except (1) initials of their first and last name(s), (2) year of birth, (3) DNI/NIE number, (4) telephone number and (5) e-mail address that does not contain the first name, surname or any other personal data.
  • It must not contain a photograph.

Candidatures that do not comply with this requirement will not be considered.


CONDITIONS

This contract will be funded by the Call for grants for implementing the INVESTIGO programme within the framework of the Recovery, Transformation and Resilience Plan - funded by the European Union - NextGenerationEU.

Este contrato estará financiado mediante la Convocatoria de subvenciones para la realización del programa INVESTIGO, en el marco del plan de recuperación, transformación y resiliencia – financiado por la Unión Europea – NextGenerationEU.

ADDITIONAL INFORMATION

The working language of the Institute is English. Full command of the English language is required in all positions.

IMDEA Materials Institute is committed to equal opportunities, diversity and the promotion of a healthy work environment and work-life balance. Female applicants are encouraged to apply to our research and technical positions. See our Gender Equality Plan here.

Applications are processed upon reception. The position might be closed once the minimum publication days have passed, so we encourage early application.

Besides on-the-job technical training, IMDEA Materials Institute is committed to training the Institute’s scientists and staff in “soft” or transversal skills. See the available training here.

Meet some of our alumni.