CIMCIP – Computational Intelligence Methodologies for Control of Industrial Processes [2010-2014]
My Role: Principal Investigator (PhD Student)
Host institutions: University of Coimbra (UC), and Acontrol.
Financing: Foundation of Science and Technology (FCT);
Reference number: SFRH/BD/63383/2009.
Abstract (original summary proposal): This thesis is devoted to research on adaptive fuzzy controllers, predictive control, and intelligent control methodologies such as neural and neuro-fuzzy control for industrial nonlinear andor time-varying plants. Nonlinear andor time varying processes are difficult to control due to their complexity. The issues of varying parameters, presence of disturbances, non-modeled dynamics, robustness, and stability will be addressed. The developed methodologies will be validated on a main case study concerning the control of NOx and SOx emissions on a cement kiln. The process is nonlinear time-varying exhibiting the above mentioned problems. Presently, there are no automatic methodologies to control these emissions under the legal limits. We intend to research control methodologies, integrating human knowledge and/or adaptivity towards improved solutions, for the kiln and treatment systems, having large impact on the amount emission removal chemicals. These chemicals have large economic costs comparable to the maintenance costs of all the kiln electrical systems.