Engineering empowered by physics-based and data-driven hybrid models: A methodological overview

Abstract

Smart manufacturing implies creating virtual replicas of the processing operations, taking into account the material dimension and its multi-physics transformation when forming processes operate. Performing efficient, that is, online accurate predictions of the induced properties (including potential defects) of the formed part (to optimally control the process parameters) needs moving beyond usual offline simulation based on nominal models, and proceeds by assimilating data. This will serve, from one side, to keep the model calibrated, and from the other, to enrich the model and its associated predictions, to avoid bias, to improve accuracy or for performing online diagnosis, by advertising on preventive maintenance. For all these purposes, a new alliance between physics-based and data-driven modelling approaches seems a very valuable route for empowering engineering in general, and smart manufacturing in particular. The present paper revisits the main methodologies involved in the construction of the component or system Hybrid Twins.

Digital twins and the future of precision mental health

“…Both machine learning approaches have the potential to contribute to the understanding of MDD pathophysiology, they have complementary limitations: supervised methods are limited by the accuracy of the prior knowledge they rely on, and unsupervised methods by the potential for unclear or uninterpretable derivation of data-driven subtypes, which reduces their likelihood of being implemented in clinical practice. Given these complementary limitations, combining the two with mechanistic models into a hybrid model ( 45 ) can maximize their ability to yield meaningful new knowledge on mental health biotypes. Hence, to maximize interpretability and generalizability, the data-driven results are integrated with commonly used conceptual models of psychopathology and psychotherapy.…”

Finite electro-elasticity with physics-augmented neural networks

“…As a consequence, the following relationships between the macroscopic fields P , d 0 and their microscopic counterparts P µ , d 0µ can be established, 13)…”

“…To close, these methods for augmenting or informing ANNs about physical requirements are not restricted to mechanical material modeling, but also find application in other fields of physics and engineering, e.g. [11,13,54,60].…”

Hybrid twin of RTM process at the scarce data limit

“…For this reason, a new type of twin has been proposed in [5], the hybrid twin (HT), which combines a physics-based model (virtual twin) with data, capable of overcoming the limitations of their virtual and digital counterparts [13,14,15]. The main idea is to fit a virtual twin with respect to the data measured by sensors in real-time, allowing the numerical simulation to better predict the experience, by estimating and correcting on the fly the difference between the numerical predictions and reality.…”