During the last years, humanity has witnessed a continuous emergence of applications that contain algorithms based on artificial intelligence. Examples of its everyday use go from a smartphone with integrated facial recognition to devices that allow us to control our cardiovascular health. In the same way, artificial intelligence is strongly permeating the industry and its influence is such that our time can be described as the beginning of the fourth industrial revolution. Industry 4.0 accentuates the idea of a continuous efficient digitization of processes controlling and managing the massive data generated by it, analyzing them and looking for hidden interrelationships while all system devices are digitally interconnected. Definitely, this era is undoubtedly marked by the massive implementation of robotics, the internet of things (IoT), Digital twins and artificial intelligence.

Manufacturing and manufacturing are not fields outside of Industry 4.0. In fact, smart manufacturing is one of the great pillars of this industry and it is one of the areas where artificial intelligence can undoubtedly provide the greatest benefits. The optimization of production processes brings innumerable improvements that range from reducing the carbon footprint to the rational and efficient use of scarce raw materials.

The MAT4.0 project (Intelligent Manufacturing of Advanced Materials for Energy, Transportation and Health), coordinated by IMDEA Materials and financed by the Community of Madrid within the 2018 Technology Call, intends, as one of its main objectives, to unite efforts to introduce artificial intelligence concepts in the field of intelligent manufacturing of structural materials .

In particular, researchers from the MAT4.0 project are focused on the improved ultrasonic inspection capabilities of composite materials manufactured by automatic fiber deposition . Carbon fiber composite materials are used in high value-added applications that require good mechanical performance as well as a significant weight reduction, as occurs in the aeronautical or space industry. That parts, once they are manufactured, are subjected to strict quality controls in order to detect the formation of defects generated during the manufacturing process or damage generated during the life of the material.

Scheme of predictive modeling for ultrasound inspection systems.

Among the usual work methods, we can highlight the ultrasonic inspection for its speed and scalability. This technique is based on the propagation of an ultrasound pulse and the subsequent reception and analysis of the echo produced. However, there are higher resolution techniques such as X-ray tomography that are obviously more complex and difficult to implement in the industrial field. The project works together in the development of models supporting the ultrasonic inspection by incorporating tomographic information through deep learning systems ( Deep learning ).

Within the framework of the collaborative project MAT4.0, the Foundation for Research, Development and Application of Composite Materials (FIDAMC) is in charge of manufacturing composite materials with the latest aeronautical technologies for automatic fiber deposition, IMDEA Materials Institute performs tomographic inspections and ultrasound of these materials and the MIDAS (Data Mining and Simulation) and MMEAN (Mechanics of Advanced Structural Materials and Nanomaterials) groups of the Technical University of Madrid are in charge of incorporating the information into the learning models to allow increasing capacities of the current ultrasonic inspection without the need for complex characterization campaigns.

This innovative collaborative project highlights the great R&D capabilities of Madrid research centers in a strategic field for the region such as the Digitization of the manufacturing industry. The results of MAT4.0 are laying the scientific-technological foundations for the development of innovations that will allow companies manufacturing structural components to be more competitive and sustainable.