Research areas
Research areas of the Chair of Process Control Systems and Research Group of Process System Engineering include following topics:
Machine Learning
The ongoing digital transformation in the process industry creates a solid data basis for the comprehensive evaluation of process data. Machine Learning solution has a potential to energy efficiency of the process, e.g., by reducing the overall batch duration and required maintenance effort.
Specifically, we use computer vision techniques for online process analysis of multiphase processes such as aeration in bioreactors. Another focus is on process prediction based on historical data, supervised and unsupervised phase detection, and change point analysis in multidimensional time series from production.
To extract valuable information from data and identify and model complex relationships in plant behavior, expertise in data engineering, data science and machine learning are essential.
More about Machine Learning
Smart Architectures
Heterogeneous and distributed devices and services are one of the core components of smart manufacturing and cyber-physical production systems. Such systems mainly depend on applications and/or services that provide a solution to achieve the integration and connectivity of diverse and heterogeneous sources. From the smart manufacturing point of view, developing architectural patterns and concepts is crucial for performance.
The main principles that form the Industry 4.0 concept include the research areas like flexibility and the modularity of production systems. As a result, process orchestration plays a significant role in modular automation. One of the critical research areas of the team is to provide modular and flexible solutions for the orchestration problem.
AI/ML stands as a promising solution for several issues and subproblems that SMEs face today. However, several technical issues must be appropriately resolved to reduce the industry's barriers to data and ML applications. As a result, another focus of the team is to ease the integration of AI/ML-based solutions in the industry.
More about Smart Architectures
Hybrid Modeling, Simulation and Optimization (HybridMSO)
Modularization and Digital Transformation are intended to increase the flexibility and efficiency of processes and plants. Methods and tools are therefore required for process development and plant operation in modular systems with distributed knowledge. Digital twins are an important solution approach for which, however, the following research questions arise, among others:
- How can digital twins be exchanged as assets reliably and vendor-independently?
- How can prior knowledge and plant data be optimally integrated into digital twins?
Open standards and quality assurance mechanisms are necessary for the vendor-independent exchange of digital twins. This concerns open standards for structural models (e.g.: DEXPI), operational models (e.g.: MTP) and behavioral models (e.g.: FMI). For quality assurance, methods from quality assurance for software (e.g.: ISO/IEC 25000) can be adapted and integrated with the Verfication & Validation(V&V) methods, so that prescriptive quality assurance is also possible for behavioral models.
The behavioral models (or simulation models) in the digital twin can be knowledge-based, data-based or hybrid. Prior knowledge about physical, chemical or even biological behavior of material systems and apparatuses serves to structure the solution space and reduces the quantity and quality of the required data. With the help of data obtained through experimental design and the plant history, models are adapted to the real system through system identification.
More about Hybrid Modeling, Simulation and Optimization (HybridMSO)
Safe and Modular Automation
The modularization of the process industry enables not only the integration of process equipment assemblies (PEA) across manufacturers, but also a variety of starting points for engineering processes. In the Safe and Modular Automation group, we focus on how efficient processes can be used for flexible, adaptable and resilient operation of modular plants.
The core topic is MTP, the associated service-oriented automation of process modules and the orchestration of process plants.
More about Safe and Modular Automation