Reliable nonlinear dynamic gray-box modeling by regularized training data estimation and sensitivity analysis

Self Cite

Fermentation processes are difficult to describe using purely mechanistic relations as the underlying biochemical phenomena are complex and often not fully understood. In order to cope with this challenge, we developed an approach to augment standard dynamic model equations by data‐based components that are fitted to data using machine learning techniques, which results in dynamic gray‐box models. This methodology is applied here to the batch fermentation process of the sporulating bacterium Bacillus subtilis, using experimental data from a lab‐scale fermenter. The key step in developing the model is the estimation of a training set for the machine learning submodels. The quality of the resulting model is analyzed, and the predictions are compared with real data.

Section: Methodology Of Dynamic Gray-box Modelingmentioning

confidence: 99%

Development of a Dynamic Gray‐Box Model of a Fermentation Process for Spore Production

Winz

Assawajaruwan

Engell

2023

Self Cite

“…The embedding of the kernel function into the mass balance in (1) or ( 2) leads to a serial or embedded gray-box approach, where the output of a data-based model that represents the kernel function is used as an input of the physical mass balance. Winz et al [1] describe a modeling methodology for embedded gray-box systems, which consists of first estimating the outputs of the embedded models from the available data, followed by a decoupled training of the data-based models and finally an integrated parameter estimation step. This decoupling allows for a more informed selection of the embedded model structure since the computational cost of the decoupled estimation and training steps are often lower than repeatedly solving an integrated parameter estimation step with different model candidates.…”

Section: Gray-box Model Trainingmentioning

confidence: 99%

“…Gray-box modeling methods aim at decreasing the reliance on data by incorporating known physical relationships between process variables into the model. By doing so, faster model development on the one hand and a higher level of interpretability on the other hand can be achieved compared to purely white-box or black-box approaches [1]. Two main configurations of combined physical and databased models can be identified from past work [2][3][4].…”

Section: Introductionmentioning

confidence: 99%

Gray‐Box Modeling of the Molecular Weight Distribution in a Batch Polymerization Reactor

Bordas

Kurt

Bamberg

et al. 2023

Self Cite

We introduce a gray‐box approach for modeling the molecular weight distribution in step‐growth polymerization reactions using the aggregation population balance equation. The approach is based on extracting a data‐based kernel function from in‐process measurements of the molecular weight distribution. The method is applied to historical data from an industrial batch polymerization reactor. The resulting model is used for decision support in production by predicting the reaction endpoint corresponding to a target molecular weight. The accuracy of the predictions proved to be sufficient for the deployment of the method.

“…Another approach is to embed data-based sub-models into structured models that represent the basic energy and material balances, leading to grey-box models [32]. This has the advantage that the complexity of the data-based models is significantly smaller than that of the overall model and the validity of the data-based elements and the effect of errors in these can be controlled much better.…”

Section: Operationsmentioning

confidence: 99%

AI in Process Industries – Current Status and Future Prospects

Bortz

Dadhe

Engell

et al. 2023

Self Cite

The chemical industry is one of the key industrial sectors in Germany and at the same time one of the largest consumers of energy and raw materials. A successful energy transition and the development of a circular economy can only succeed if they are actively supported and shaped by the chemical industry – through the redesign of existing production processes and the exploration and implementation of new process routes. The challenge is to realize this transformation within a very short time and for many production processes, whereby a much larger number of process routes must be explored. Digital technologies are key to master this transformation towards more sustainability, climate, and environmental protection. The KEEN project aims to explore and leverage artificial intelligence (AI) opportunities in process industry. The newly developed AI methods are tested wherever possible in real working environments and production plants to prove the economic benefit, applicability, and reliability of the methods and technologies.