Basf Se Corporation

Technische Universitaet Berlin

Simeon Sauer

Klaus-Robert Mueller

Eric Jenne

Hendrik De Winne

Miriam Angela Anna Hahkala

Alexander Badinski

Mihail Bogojeski

Britta Carolin Buck

Nataliya Yakut

Bart Blankers

Franziska Horn

Daniel Keck

In order to predict the future evolution of a health-state of an equipment and/or a processing unit of a chemical production plant, e.g., a steam cracker, a computer-implemented method is provided, which builds a data-driven model for the future key performance indicator based on the key performance indicator of today, the processing condition of today, and the processing condition over a prediction horizon.

Forecasting the progress of coking and fouling for improved production planning in chemical production plants

By accurately predicting industrial aging processes (IAP), such as the slow deactivation of a catalyst in a chemical plant, it is possible to schedule maintenance events further in advance, thereby ensuring a cost-efficient and reliable operation of the plant. So far, these degradation processes were usually described by mechanistic models or simple empirical prediction models. In order to accurately predict IAP, data-driven models are proposed, comparing some traditional stateless models (linear and kernel ridge regression, as well as feed-forward neural networks) to more complex stateful recurrent neural networks (echo state networks and long short-term memory networks). Additionally, variations of the stateful models are discussed. In particular, stateful models using mechanistical pre-knowledge about the degradation dynamics (hybrid models). Stateful models and their variations may be more suitable for generating near perfect predictions when they are trained on a large enough dataset, while hybrid models may be more suitable for generalizing better given smaller datasets with changing conditions.

Forecasting industrial aging processes with machine learning methods

Growing community of inventors

Heidelberg, Germany

Simeon Sauer