DataProMat · A Diffusion Maps workflow enabled by Neural Networks and Equation-free calculations for multi scale material and process modelling
Horizon 2020 — Marie Skłodowska-Curie Actions
- Duration
- 2021-09-01 → 2023-08-31
- EU contribution
- €178,320
- Participants
- 1
- Scheme
- MSCA-IF
Lines connect the coordinator with its partners.
Results in brief
A Diffusion Maps workflow enabled by Neural Networks and Equation-free calculations for multi scale material and process modelling
The goal is project is to develop an integrated machine learning framework for efficient material and process modeling that will help accelerate the design of new materials and their large-scale production process, by leveraging data from simulations and experiments. Specifically: 1. To achieve Reduction of size and complexity of reactive flows through model order reduction of the reactor dynamics 2. To develop, implement, and validate a machine learning framework for the prediction of material properties. A key component of the proposed approach is efficient dimensionality reduction using manifold learning, here by implementing Diffusion Maps. The latter can be thought of as the nonlinear counterpart of Proper Orthogonal Decomposition appropriate for data belonging to a curved manifold. This method yields a parsimonious parametrisation of data manifold, leading to improved accuracy and efficiency, in comparison to POD. Through appropriate interpolation technology, also based on Diffusion Maps, the so-called Geometric Harmonics, it is possible to map from the input to the output space (here the distribution of mass, momentum and temperature inside a chemical vapour deposition reactor) and vice versa, as well as from partial observations to other partial observations or even to the full solution space.
Data: CORDIS, © European Union
Project objective
The ambition of this fellowship, hosted by Professor S.P.A. Bordas (UL), is to propose a nonlinear manifold learning framework, in particular to implement the Diffusion Maps methodology, enabled by “equation-free” calculations and Artificial Neural Networks, in the context of multi-scale materials and process modeling and design. The goal is to push the boundaries of the “Digital Twins” paradigm beyond the current-state-of-the-art and to establish a methodological framework that links macro-scale process parameters and conditions to properties of complex materials, in an effort to meet the current market-driven demands for efficiency, scalability, safety, sustainability and innovation. The proposed approach is based on the current trends in materials and process modeling, on which the host is in the best possible position to advise as a world leading expert. Effectively, the fellowship sets the stage for interdisciplinary integration: Starting from the requirement for a specific set of properties, we must be able to predict the appropriate material structure, its capabilities and limitations and to propose ideal processing steps that will enable large-scale production. In this context, machine learning in the form of Diffusion Maps will be implemented for dimensionality reduction aiming to the reach the maximum possible size compression. The equation-free approach will be integrated with Diffusion Maps, in order to efficiently explore the, typically large, parameter space and Artificial Neural Networks will be applied as a means of leveraging the abundantly available digitized images to “learn” the long-term dynamics of the material behavior.
Original text from CORDIS.
Participants
- UNIVERSITE DU LUXEMBOURG · ESCH-SUR-ALZETTECoordinatorLuxembourg
Links
Data: CORDIS, © European Union