Descripción del proyecto
Los modelos automatizados impulsan la investigación de la materia blanda
Los materiales blandos, que se pueden deformar o alterar estructuralmente con facilidad por tensiones térmicas o mecánicas, son esenciales en la vida moderna, pues afectan a la autonomía, la sostenibilidad y la salud. Sin embargo, modelizar estos materiales con precisión es complejo y suele estar limitado a unos pocos expertos bien formados. El equipo del proyecto DISCOVER, financiado por el CEI, pretende hacer más accesible la modelización constitutiva mediante el descubrimiento automatizado de modelos. Los objetivos incluyen el desarrollo de redes neuronales que encuentren de forma autónoma los mejores modelos, parámetros y experimentos para los diversos sistemas de materia blanda. Además, los investigadores evaluarán el rendimiento del modelo en diferentes experimentos y utilizarán el análisis bayesiano para medir las incertidumbres. El descubrimiento automatizado de modelos debería permitir la exploración de una amplia gama de parámetros de modelos, lo que ofrece una visión de los sistemas de materia blanda que los métodos tradicionales no pueden alcanzar.
Objetivo
Soft materials play an integral part in many aspects of modern life including autonomy, sustainability, and human health, and their accurate modeling is critical to understand their unique properties and functions. However, the criteria for model selection remain elusive and successful modeling is limited to a few well-trained specialists in the field. My goal is to democratize constitutive modeling through automated model discovery and make it accessible to a more inclusive and diverse community to accelerate scientific innovation. My overall objectives are: i) Establish a new family of constitutive neural networks that simultaneously and fully autonomously discover the model, parameters, and experiment that best explain a wide variety of soft matter systems; ii) Quantify the performance of our discovered models on tension, compression, and shear experiments for the heart, arteries, muscle, lung, liver, skin, brain, hydrogels, silicone, artificial meat, foams, and rubber; and iii) Quantify the uncertainty of our models, parameters, and experiments using a Bayesian analysis. My hypothesis is that automated model discovery will facilitate the exploration of a large parameter space of models and provide unprecedented insights into soft matter systems that are out of reach with conventional theoretical and numerical approaches today. My immediate deliverable is a fully documented open source scientific discovery platform that includes our new neural networks, experimental data, benchmarks, models, and parameters. This discovery platform has the potential to induce a ground-breaking change in constitutive modeling and will forever change how we simulate materials and structures. This project will democratize constitutive modeling; stimulate discovery in soft matter systems; provide deep-learning based tools to characterize, create, and functionalize soft matter; and train the next generation of scientists and engineers to adopt and promote these innovative technologies.
Palabras clave
Programa(s)
- HORIZON.1.1 - European Research Council (ERC) Main Programme
Régimen de financiación
HORIZON-ERC - HORIZON ERC GrantsInstitución de acogida
91058 ERLANGEN
Alemania