The Ile-de-France region and Dassault Aviation join forces to organize an Artificial Intelligence Challenge.
Commercial aircraft design is based on physical models and digital simulation tools that model the stress exerted on an aircraft and ensure right-sizing. This fine modeling is proven and has allowed sizing and certification of a number of Dassault Aviation aircraft, but it is less suited for real-time flight analysis. To achieve this objective, a new approach is being considered. It would use data science and artificial intelligence.
To ensure aircraft modeling quality, our test aircraft fly several hundred flights. These aircraft are specially equipped with thousands of sensors that measure a large number of flight parameters.
The promise of artificial intelligence is to process these terabytes of data generated and virtually create the test instrumentation on customer aircraft, and therefore estimate aircraft fatigue and the maintenance required. Learning from time series data such as flight sequences requires cutting-edge techniques which are rarely used in aviation. With this in mind, Dassault Aviation sought out the fresh minds of startups at the Greater Paris AI Challenge for Industry.
On October 15, 2018, the Ile-de-France region (ie. Greater Paris) adopted its AI2021 plan, which seeks to reward the AI excellence of local companies, and to give them the resources to position themselves within the international competition while contributing to advances in economic, social and societal issues. At the Paris Air Show 2019, Dassault Aviation and the Ile-de-France region announced their intent to organize an artificial intelligence “challenge”.
The purpose of the Challenge is to select the startups with the most relevant ideas for developing and validating an algorithm that predicts the in-flight time response of stress gages only found on test aircraft. This prediction must be established based on time signals generated by a set of functional sensors found on all aircraft in the fleet. The algorithm will contribute to the predictive maintenance objectives of Dassault Aviation, in particular concerning the structural elements of its Falcon business jets line.
This algorithm will contribute to the Dassault Aviation teams’ search for the best design compromise between robustness and weight of future products to help reduce the carbon emissions of its aircraft.
The Challenge is organized with the support of the Astech and Systematic competitiveness clusters and the nonprofit Startup Inside. It will be held on the Codalab Challenge platform designed by Professor Isabelle Guyon’s team and hosted by Paris-Saclay University, which provides the technical support for the Challenge.