Dresden 2020 – scientific programme

The DPG Spring Meeting in Dresden had to be cancelled! Read more ...

Parts | Days | Selection | Search | Updates | Downloads | Help

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 35: Topical Session: Data Driven Materials Science - Machine Learning for Production (joint session MM/CPP)

CPP 35.1: Topical Talk

Monday, March 16, 2020, 17:00–17:30, BAR 205

First-principle infused machine learning models allowing digital twins to self-organise production processes — •Marcus Neuer — Sohnstr. 65, 40237 Düsseldorf

For European process industries, optimization of the production route plays an increasingly important role for keeping a competitive edge in a tough market. New concepts like digital twins arrived recently in real-world applications. They allow an agent-based, active self-organisation of the route and introduced the ability to apply models live, during the processing. These models may be analytically derived, data-based or a combination of both: first-principle infused machine learning models. Herein, the stochasticity of the process is modelled by the machine learning approach, while an analytical first-principle backbone acts as basis. With the ability to forecast its potential future, materials and products have new degrees of freedom. They optimize their future path with respect to energy consumption, order matching and material homogeinety. The presented concepts are shown in real-world production environments, where they have already reached technical readiness to sustain continuous production.