ML-based porosity inspection method for composite materials

Patent details


Spanish patent granted

Concession number

ES2947809 A1


Federico Sket, Juan Ignacio Caballero, Carlos Daniel González , Ernestina Menasalvas Ruiz y Consuelo Gonzalo Martín 

Priority date

9 May 2022


IMDEA Materials Institute, Universidad Politécnica de Madrid

Transfer opportunity

License of technology


Non-destructive machine learning-based methodology capable of improving porosity estimation and types of porosity on composite materials.


Composite materials are usually formed by two elements: a fabric and a polymer resin. During the manufacturing process of composite components, tiny bubbles and voids appear accidentally. This effect called porosity is the main defect in the production lines of the aerospace industry, and in a less term in the eolic power generation and automotive.

The problem is that the estimation of the amount of porosity is performed by techniques that are not accurate enough and do not provide other relevant information about its impact on the components.

IMDEA Materials and UPM have developed techniques based on signal analysis and machine learning that enable a better estimation of the amount of porosity and other technical details.

Advantages and innovations

  • The ability to establish different types of porosity.
  • The improvement of the estimation of porosity thanks to a new methodology based on using X-rays, the types of porosity and machine learning.


Knowledge & Technology Transfer Department, IMDEA Materials Institute


telephone: +34 91 5493422