On October 15, we organized a scientific seminar titled “Advanced Control Approaches for Optimal Operation of Complex Processes”. Johannes Jäschke, a professor at Norwegian University of Science and Technology, led the seminar, which was organized within the framework of the FrontSeat project as part of the “Research Seminar on Smart Cybernetics” series.
Abstract:
The first part of the seminar focuses on flexible operation of electrolyzers. We consider an alkaline water electrolyser which separates water into a hydrogen and an oxygen stream. In practice, there will always be some impurities in the product streams. When subjecting the electrolyzer to fluctuating power, it is very important to ensure that the hydrogen content in the oxygen stream is below a safety critical threshold. If that is not given, and the hydrogen content in the oxygen stream exceeds ca 4%, the gas mixture may self-ignite and explode. We present a control structure that is designed to maximize hydrogen production under fluctuating power conditions, while ensuring that the hydrogen-to-oxygen ratio is below a given value.
In the second part, we present some approaches for robust nonlinear model-predictive control (NMPC) with a focus on limiting the online computational cost. The first approach represents a heuristic simplification of the Multi-stage NMPC approach, in which the uncertainty is considered only for the first few sample times, while a nominal scenario is considered for the sample times further into the future. The second approach aims to reduce the online-cost of the NMPC by training a neural network controller offline in using a model closed-loop. When deployed on the process, the online cost of this controller is reduced to only evaluating a neural network.
Photo: Juraj Oravec
This project has received funding from the European Union’s Horizon under grant no. 101079342 (Fostering Opportunities Towards Slovak Excellence in Advanced Control for Smart Industries).
0 Comments