Strategies for efficient and robust model predictive control

by Johannes Jäschke and Evren Turan (Norwegian University of Science and Technology, Norway)

Download slides

In a model predictive control (MPC) strategy we solve an optimisation problem to find control inputs such that some objective is minimized and controls are satisfied, using a dynamic system model to predict the short-term response. Using modern techniques and optimisation algorithms this problem is feasible to solve in real-time, with the industrial implementation of linear and non-linear MPC schemes. However, it is still an open challenge to robustly consider uncertainty in the MPC formulation due to the increased computational expense. This workshop will review some current MPC approaches and techniques to reduce the computation cost. An outline of the covered topics is as follows:

  1. Introduction to Model predictive control
  2. Overview of robust MPC methods
  3. Speeding up NMPC, including:
    – Sensitivity-based methods
    – Imitation learning approaches
    – Policy optimization
  4. Outlook and research challenges


Johannes Jäschke is a Professor at the Dept. of Chemical Engineering at the Norwegian University of Science and Technology (NTNU) in Trondheim. He received the degree Dipl.-Ing. (M.S.) in Mechanical Engineering at RWTH Aachen University in 2007, and his PhD in Chemical Engineering at NTNU in 2011. He worked as a postdoc at NTNU and Carnegie Mellon University before being appointed to his current position in 2014. His research interests lie within the field of process systems engineering, with a strong focus on modelling, numerical optimization and control. By linking methods from chemical engineering, optimization and control theory, his goal is to systematically develop practically applicable solutions for operating process systems in a safe, reliable, and economical way.

Evren M. Turan is a third-year PhD candidate at the Norwegian University of Science and Technology (NTNU), focusing on methods that combine operational data, first principles knowledge, and machine learning for use in process systems engineering applications. Before coming to NTNU he completed a master’s and bachelor’s studies in Chemical Engineering at the University of Cape Town.