본문 시작


Online Conference Website [link]

** You can log-in with your ID and PWD used for conference registration site ([link]).

** Here is a detailed instruction for oral presentation and the online platform ([link])

Plenary Speakers

 Plenary Speaker 1


 Feedback control for optimal process operation revisited: Models, data, and robustness

Sebastian Engell, Department of Biochemical and Chemical Engineering, TU Dortmund, Dortmund, Germany


In my ADCHEM 2006 Plenary Talk, I advertised to consider feedback control as a means to achieve optimal process operation, beyond regulating key variables of a process to the desired setpoints. Specifically, I pointed out that model predictive control with rigorous models has the potential to achieve economically optimal dynamic operation and that it is computationally feasible [1]. This idea has since been pursued in many works, under different names, e.g. eNMPC, DRTO. Tailored solutions based on this principle are applied industrially, e.g. in the minimization of the batch time of polymerizations. Other proposals aim at establishing optimality without employing models explicitly as e.g. “self-optimizing control” and “extremum-seeking control.” Still, in the design of such schemes, models are also used.

To build good models of complex processes is a demanding task. Even if large efforts have been invested into modelling, the behavior of a real process will always deviate from the model predictions. The use of measured data can compensate these deficiencies, that is why feedback is employed, but controllers can also fail due to model errors. There is a broad spectrum of possible combinations of using models and data, from using models only in the design of control algorithms to relying on model predictions and using only simple correction schemes, or adapting models online. 

Generally, one is interested in techniques that provide good performance without requiring huge modelling efforts. In the talk, we discuss two approaches to reducing the negative effects of model errors in optimization and control that represent different generic concepts. For real-time optimization, we outline the so-called modifier adaptation approach, which adds a data-based local model to a global nonlinear model and updates it iteratively to ensure convergence to the true optimum of the real plant. As an example of robust control, the multi-stage MPC approach is presented in which the presence of future information on the realization of the model uncertainty is included in the optimization to reduce conservatism.

Finally, the idea of replacing rigorous models by “machine learning”, possible pitfalls and strategies to handle model deficiencies in this context will be discussed.

[1] Engell, S. (2007): Feedback control for optimal process operation. Journal of Process Control 17, 203-219. 


Sebastian Engell received a Dipl.-Ing degree in Electrical Engineering from Ruhr-Universität Bochum, Germany in 1978 and the Dr.-Ing. Degree and the venia legendi in Automatic Control from Universität Duisburg in 1981 and 1987. 1984/1985 he spent a year as a PostDoc at McGill University, Montréal, Canada. 1986 – 1990 he was the head of an R&D group at the Fraunhofer Institut IITB in Karlsruhe, Germany. 1990 he was appointed as a Full Professor of Process Dynamics and Operations in the Department of Biochemical and Chemical Engineering at TU Dortmund. 2008 he was a Distinguished Visiting Professor at Carnegie Mellon University, Pittsburgh, USA. He was Department Chairman 1996-1999 and 2012-2014 and Vice-Rector for Research of TU Dortmund 2002-2006. He currently is a member of the Research Council of the Alliance of the Universities in the Ruhr Region, UA Ruhr.

Prof. Engell is a Fellow of IFAC, the International Federation of Automatic Control since 2006. He received an IFAC Journal of Process Control Best Paper Award, and is a co-author of the 2014 and 2016 Best Papers in Computers and Chemical Engineering. He gave the Bayer Lecture in Process Systems Engineering at Carnegie Mellon University in 2008 and the Roger Sargent Lecture at Imperial College, London, in 2012. In 2011, he was awarded an ERC Advanced Investigator Grant for the Project MOBOCON: Model-based Optimizing Control – From a Vision to Industrial Reality. He is author or co-author of more than 600 papers that are listed in Scopus and has graduated more than 80 PhD students at TU Dortmund.

His research interests are in the areas of model-based optimizing control, real-time optimization, coordination between locally optimized coupled production systems, and scheduling. In his research, the aspect of uncertainty about the behavior of the system that is controlled or optimized has always been in the focus.

  Plenary Speaker 2


 Batch Process Automation Methodology Development -- a Thermoplastic Injection Molding Example

Furong GAO, Department of Chemical & Biological Engineering, and Center for Polymer Processing and Systems, Hong Kong University of Science and Technology, Hong Kong SAR, CHINA


Batch process is the preferred choice for the manufacturing of specialty chemicals or other high value-added products.   The differences between a continuous process and a batch process call for the need of modeling, monitoring, control, and optimization methods to be developed in harmony with the batch process nature. In this talk, the speaker will firstly take a thermoplastic injection molding, a process converting plastic granules into various molded parts, as an illustrating process to highlight the batch process nature. Secondly, the speaker will present progresses of these automation methodology developments by exploring or addressing the batch nature. Thirdly, the application of these methods to injection molding will be demonstrated. Finally, some personal views of batch process automation future development particularly on injection molding will be given.  


Dr Furong Gao is a Chair Professor of Chemical and Biological Engineering, Hong Kong University of Science and Technology (HKUST), he also serves HKUST as the Founding Director, Center for Polymer Processing and Systems (CPPS) since 2007 and international molding communities as the President, Society of Advanced Molding Technology (SAMT) since 2017.


Professor Gao received his BEng in Automation from China University of Petroleum in 1985, MEng and PhD, both in Chemical Engineering, from McGill University in 1989 and 1993, respectively. Before joining HKUST in 1995, he worked as a Senior Research Engineer from 1993 to 1995 in charge of Intelligent Process Control (IPC) project at Moldflow Pty Ltd (now, a part of Autodesk Inc), Melbourne, Australia. His research interest is in batch process automation method developments and their applications to polymer processing and energy management areas. To date, he has authored more than 500 journal and conference papers, seven books, and 80 patents. Some of his technologies have been licensed to several major molding and control companies and he founded two start-up companies.

 Plenary Speaker 3 


Heterogeneous cell population models uncover “rules of life”, inform experimental design, and enable control of cell population dynamics. 

Prof. Neda Bagheri, Chemical & Biological Engineering, Northwestern University


 Computational models are essential tools that can be used to simultaneously explain and guide biological intuition. My lab employs machine learning, dynamical systems, and agent-based modeling strategies to explain biological observations and uncover fundamental principles that drive both individual cellular decisions and cell population dynamics. We are interested in the inherent multiscale nature of biology, with a specific focus on system-level dynamics that emerge from interactions of simpler individual-level modules.

 In this presentation, I introduce a multiscale agent-based model of a cell population that integrates subcellular signaling and metabolism, cellular level decision processes, and dynamic vascular architecture and function to interrogate multilateral regulation among heterogeneous cell agents. I also introduce a reduced phase model of heterogeneous circadian oscillators and use this model to control phase resetting. Both modeling frameworks are flexible and can be adapted to represent, analyze, and control a wide variety of biological systems.  


 Neda Bagheri earned her doctorate in electrical engineering from the University of California Santa Barbara. Her emphasis in control theory reinforced a deep interest in biology and set the course for her research trajectory. After completing a postdoctorate in biological engineering at MIT, she joined the chemical & biological engineering faculty at Northwestern University and launched the Modeling Dynamics of Living Systems (MoDyLS) Laboratory. The Bagheri Lab develops computational models to help explain experimental observations and elucidate fundamental properties governing cellular dynamics, regulation, and function. In 2019, she joined the Allen Institute for Cell Science in an advisory role and moved her lab to the University of Washington Seattle where she holds a joint position in biology and chemical engineering. 




Araujo Pimentel, G., Dewasme, L.,  Vande Wouwer, A.

 Data-driven Linear Predictor based on Maximum Likelihood Nonnegative Matrix Decomposition for Batch Cultures of Hybridoma Cells

 A Yousef, I., Shah, S. L., Gopaluni, B.

 Visual Analytics: A New Paradigm for Process Monitoring

 Lu, C., Paulson, J.

 No-Regret Bayesian Optimization with Unknown Equality and Inequality Constraints using Exact Penalty Functions

 Turan, E. M., Jäschke, J.

 Designing neural network control policies under parametric uncertainty: A Koopman operator approach

 Busschaert, M., Waldherr, S.

 Inference of Kinetics in Population Balance Models using Gaussian Process Regression

 Bagla, G., Valluru, J., Deshpande, A., Patwardhan, S.C.

 Intelligent State Estimation for Online Optimizing Control of a Reactor System exhibiting Input Multiplicity

 Espinel-Ríos, S., Morabito, B., Pohlodek, J,. Bettenbrock, K., Klamt, S., Findeisen, R.

 Optimal Control and Dynamic Modulation of the ATPase Gene Expression for Enforced ATP Wasting in Batch Fermentations

 Liu, Y., QIN, S. J.

 A Novel Two-step Lasso  Approach for  Inferential Sensor Variable Selection and Estimation