SystemX, the technological research institute (IRT) dedicated to the digital engineering of future systems, announced on September 2 the launch of the 4th project of its AI2 program, which aims to combine AI, business knowledge and physical models. The new collaborative R&D project led by SystemX, “Agility and Fidelity of Simulations” (AFS), which brings together four industrial partners (ESI Group, Plastic Omnium, Groupe Renault and Stellantis), aims to develop new methodologies to improve the performance of design cycles for complex physical systems using digital simulation.
The AFS project is part of the ” Artificial Intelligence and Augmented Engineering (AI2)” program launched by IRT SystemX in February 2020.
The roadmap of the IA2 program includes 6 collaborative R&D projects, not to mention the upstream scientific coordination project, led by Marc Schoenauer, Director of Research at Inria Saclay, which aims to pool work. IRT SystemX proposes to develop solutions that hybridize three approaches to modeling and simulation:
- Physical modeling of systems,
- The use of business knowledge or behavioral models expressed by experts,
- Learning models based on real data.
AFS is the fourth AI2 project, after the Hybridization Simulation and Learning project, HSA (April 2020), the Cockpit and Bidirectional Assistant project, CAB (June 2020) and the Business Semantics for Multi-Source Data Mining project, SMD (June 2021).
With the start of the AFS project, the consortium of the IA2 program reaches the number of 16 industrial partners and 5 academic partners.
The “Agility and Fidelity of Simulations” AFS project
Although only recently announced, the AFS project was launched last June and is based on two main axes:
- Improving the agility of simulation processes in the design of complex systems;
- Controlling the fidelity of simulation models through the implementation of new metrics for better decision making.
Mouadh Yagoubi, AFS project leader at SystemX, explains:
“Today, digital simulation represents an indispensable tool in the design phase of complex systems in order to apprehend behaviors and situations that are difficult to predict. In a context where several disciplines and players are involved, simulation processes must deal with multi-disciplinary issues and ensure that the various players in a project are well coordinated to formulate a coherent response. This is precisely the ambition of the AFS project, which is particularly interested in multidisciplinary simulations, based on a model that can describe several levels of fidelity to the physics.
The objectives of the project are fivefold:
- Develop a methodology to strengthen the link between system architecture and digital simulation and improve the agility of the design of complex systems;
- Develop a methodology to improve the collaboration between manufacturers and equipment suppliers in the design of complex systems;
- Standardize the exchange of information in simulation-based design processes;
- Define a methodology for managing the fidelity of simulation models to control the credibility of simulation results;
- Develop new approaches based on artificial intelligence to improve the fidelity of numerical simulation models by physical results (modeling of error by AI, model registration … etc.).
This work will be implemented on an automotive use case dealing with thermal control, in order to illustrate the two major research axes of the project: agility and fidelity of simulation models.
In this objective, the project partners will have to overcome together several technological barriers:
- Formalization of design processes in an extended enterprise, with a focus on information exchange between stakeholders.
- Specification and development of software prototypes to accelerate simulation-based design processes.
- Measuring the fidelity of simulation models and assessing the credibility of simulations.
- Enhancing the fidelity of simulation models using AI.
