Gerado Brand Rihm, M.Sc.

In the course of my research, the following topics crystallize:

• Machine Learning (ML) in Process Engineering
• Dynamic process simulation
• Dynamic sampling methods
• Data-driven modeling / Nonlinear system identification
• Nonlinear model predictive control (NMPC)
• Trajectory optimization
• Multivariate causality analysis in time series

Among the most important goals is the data-driven modeling of complex procedural processes, which are characterized by multivariate, nonlinear and partially discrete dynamics. Be it models identified from real plant data or surrogate models of first-principles models, the complexity of these is crucial for their application in real-time optimization. Here, machine learning (ML) methods find promising applicability, even though dependence on process knowledge is unavoidable. Identifying causal relationships between process variables could reduce these dependencies.