Summary
- Profile Type
- Technology offer
- POD Reference
- TOES20240717011
- Term of Validity
- 17 July 2024 - 17 July 2025
- Company's Country
- Spain
- Type of partnership
- Commercial agreement with technical assistance
- Targeted Countries
- All countries
Contact the EEN partner nearest to you for more information.
Find my local partner
General information
- Short Summary
- Hybrid materials modelling is a powerful combination of Computational Materials Science (CMS) and Artificial Intelligence(AI). CMS predicts the properties and behavior of materials, while AI and data-driven models are one of the most promising and disruptive research areas. A Basque (Spain) startup is looking for partners to test a technology based on the combination of both, opening new fields in material modelling and addressing the challenges of characterization complex and advanced materials
- Full Description
-
Research and innovation in materials science are fundamental to sustainable technological and industrial development in Europe. New materials and manufacturing processes lead to new industries, technologies, business models, and challenges. Today, materials science must deal with complex materials and advanced manufacturing processes, requiring new methodologies for material modelling and characterization. Materials science, specifically the characterization of materials, faces significant challenges today. The development of new materials such as nanomaterials or biomaterials hydrogels requires a thorough review of conventional characterization techniques. New manufacturing technologies, such as additive manufacturing, or advanced coating, introduce complexity by generating gradients, residual stresses, and general heterogeneity. The emergence of new experimental and modelling techniques tries to partially solve these drawbacks. Advances in tomography, micro and nano testing, digital image correlation, microscopy, thermography, and acoustics increase the quantity and type of available data and information. Developments in materials modelling, such as Multiphysics or multiscale analysis, are closely linked to characterization. New models require new properties and provide new methods through inverse analysis.
Integrating new experimental and modelling techniques, including artificial intelligence, offers a powerful alternative to standard material characterization by determining material properties through inverse analysis. The inverse problem involves determining the material properties and their uncertainty, which, when introduced into simulations, better predict experimental results. Beyond common machine learning algorithms, some neural network architectures have been identified as potentially useful for hybrid modelling in materials characterization. These include Gaussian processes, Bayesian neural networks, and invertible neural networks (INNs). Probabilistic neural networks establish a proper framework for inverse analysis to determine material properties from experimental data. INNs introduce a bijective mapping between properties and measurements. Given a specific measurement the inverse process of the INN provides the corresponding material properties. The final output is the full posterior distribution for each material property, conditioned by the observed measurements.
This methodology has been applied successfully in specific cases, such as determining the poroacoustic properties of cellular concrete or obtaining the plastic stress-strain curve in metal alloys from non-conventional testing. It can be extended to other types of materials and both laboratory and industrial tests. Industrial quality control is a particularly interesting field where this methodology can significantly improve the accuracy and depth of material characterization.
The application of this methodology at industrial level requires to analyse previously some points as which is the optimum structure of the invertible neural network (INN) depending on the available data and their type, as well as the specific problems to which it is applied, how to extend the preliminary results to new cases.
By integrating experimental data with advanced modelling techniques, industries can achieve a more comprehensive understanding of material properties and behaviours. This integration allows for the prediction of material performance under various conditions, leading to better quality control and more efficient production processes. By combining the capabilities of industrial partners, research centres, IT partners, and AI partners, the collaboration aims to enhance the accuracy, efficiency, and applicability of material characterization validated in industrial scenarios. The desired outcomes include improved accuracy and predictive capabilities through the integration of hybrid modelling techniques with real-world data and advanced experimental testing. - Advantages and Innovations
- By integrating experimental data with advanced modelling techniques, industries can achieve a more comprehensive understanding of material properties and behaviours. This integration allows for the prediction of material performance under various conditions, leading to better quality control and more efficient production processes. By combining the capabilities of industrial partners, research centres, IT partners, and AI partners, the collaboration aims to enhance the accuracy, efficiency, and applicability of material characterization validated in industrial scenarios. The desired outcomes include improved accuracy and predictive capabilities through the integration of hybrid modelling techniques with real-world data and advanced experimental testing.
- Technical Specification or Expertise Sought
- The incorporation of industrial partners is a crucial role for integrating hybrid modelling techniques into their existing workflows. Industry provides the practical environment for real-world testing and validation, using their facilities and production lines to ensure the models work under actual operating conditions. Continuous feedback from industrial partners is fundamental key allowing developers to refine and improve their approaches.
- Stage of Development
- Available for demonstration
- Sustainable Development Goals
- Goal 9: Industry, Innovation and Infrastructure
- IPR status
- Secret know-how
Partner Sought
- Expected Role of a Partner
-
The incorporation of industrial partners is a crucial role for integrating hybrid modelling techniques into their existing workflows. Industry provides the practical environment for real-world testing and validation, using their facilities and production lines to ensure the models work under actual operating conditions. Continuous feedback from industrial partners is fundamental key allowing developers to refine and improve their approaches.
Other potential partners could be:
• Research centres and Universities can contribute by conducting experimental testing and validation, including designing and executing experiments to gather data on material properties and behaviours using advanced testing facilities.
• IT partner for developing the front-end interface that users interact with, including designing intuitive and user-friendly interfaces that allow effective interaction with hybrid models and result analysis.
• AI partner can bring expertise in analysing data and optimizing the performance of the models implementing the optimal neural network for each particular experimental problem. - Type and Size of Partner
- Big company
- SME 11-49
- SME <=10
- SME 50 - 249
- R&D Institution
- University
- Type of partnership
- Commercial agreement with technical assistance
Dissemination
- Technology keywords
- 01003008 - Data Processing / Data Interchange, Middleware
- 01003016 - Simulation
- 01003003 - Artificial Intelligence (AI)
- 01003007 - Computer Technology/Graphics, Meta Computing
- Market keywords
- 09007004 - Engineering and consulting services related to construction
- 02006004 - Data processing, analysis and input services
- 02002003 - Graphics software
- Sector Groups Involved
- Aerospace and Defence
- Targeted countries
- All countries
