Self-learning Model Predictive Control using Industry 4.0 Principles

This project offers an exciting opportunity to work at the interface of Academia and Industry in order to develop
an advanced self-learning automated control configuration based on model predictive control (MPC) using the Industry 4.0 framework, incorporating, machine learning methodologies, cloud computing, model reduction techniques and controller robustness principles. In particular, Artificial Neural Networks (ANNs) as well as “equation-free” methodologies will be exploited.

The MPC configuration developed will have predictive capabilities and will be able to ideally address disturbances, before they hit the production line. In addition, the control actions will be “exactly” (as opposed to empirically) calculated to optimise the process performance and stability. The project will develop “smart”, modular controllers that use directly existing process timeline data (rather than actual process models) to predict future process data, to decide control actions for optimal performance (e.g. optimal thickness, softenss, and moisture) and importantly to update process timelines and control actions accordingly. In addition, economic criteria will be built into the optimal process performance evaluation, such as minimum cost, maximum economic potential etc.

The successful candidate will work under the supervision of Prof. C. Theodoropoulos at the University of Manchester, School of Chemical Engineering and Analytical Science and he/she will be expected to work closely with the plant manager and the technical manager of the industrial sponsor. In addition, visits and collaborations with several plants of the industrial sponsor in Europe are envisaged.

Candidates should have (or be about to obtain) an excellent degree in Chemical Engineering or a related field (e.g. Electrical Engineering, Applied Mathematics) evidenced by marks, position in class, if available, etc. They should also have excellent computational/programming skills as well as communication skills, and should value research impact and industrial input.