Process Control Research Group

Dynamic process models are derived form the 1st law of thermodynamics constrained by equalities and optimality-type inequalities derived from the 2nd law. With this understanding it is possible to use process knowledge to determine the dynamic system properties of process systems and to design better controllers therefrom.

Process Modelling

A synergetic grey-box approach is applied to solve research problems in process control integrating process systems engineering with systems and control theory. The basis of the approach is process modelling and model analysis using first engineering principles. Formal methods of computer science and artificial intelligence are applied to construct, verify, analyse and simplify process models in a rigorous and automated way. These methods are implemented in intelligent computer-aided modelling tools to support the modeller in process model building and analysis.

Our interest also includes the investigation of the effect of algebraic, model building and model simplification transformations on the computational and dynamic properties of process models. The approaches and methods of model reduction are also investigated and applied to define the notion of minimal process models and analyse their properties.

Multi-scale models form a challenging modern area of process modelling. The integration frameworks of multiscale models have been investigated together with the application of model metrics to evaluate the computational properties of multi-scale models. Intelligent diagnostic systems that utilize the structure of multi-scale process models are also in the focus of our research.

Work in this area, a joint effort with Prof. Ian Cameron, CAPE Centre, Dept. of Chemical Engineering, The University of Queensland, Brisbane (Australia).

Nonlinear Process Systems

Nonlinear system theory, nonlinear control and diagnosis belong to the most challenging and developing areas in post-modern systems and control theory. Process systems are known to be highly nonlinear and are governed by the basic laws of thermodynamics. Therefore, the grey-box modelling and control of these systems are based on the joint understanding of modern nonlinear system analysis and control methods and the fundamentals of process systems engineering.

Our interest in nonlinear process systems includes: nonlinear reachability and stability analysis Hamiltonian process systems passivation and loop-shaping controllers nonlinear controller structure selection based on the analysis of input-output behaviour.

Quasi-polynomial systems for a wide class of nonlinear systems with smooth nonlinearities. Besides of a well-studied algebraic structure and a canonical Lotka-Volterra representation of this class, an entropy-analogue Lyapunov-function candidate is also available. It was shown that a Lotka-Volterra system is globally stable with a formerly known entropy-like Lyapunov function candidate if and only if there exists a local dissipative-Hamiltonian description of the system in the neighbourhood of the equilibrium point with a quadratic Hamiltonian function. Furthermore, a method for the estimation of guaranteed quadratic stability neighborhood has been developed using linear matrix inequalities. Globally stabilizing feedback controller design was shown to be solved by using bilinear matrix inequalities.

Simple but industrially important process systems, such as heat exchangers and fermentation processes are used as case-studies. Mechanical-process systems, such as gas turbines and electro-mechanical brake systems are also investigated for the purpose of designing model-based nonlinear controllers.

Intelligent discrete process control and diagnosis

Intelligent discrete process control and diagnosis methods apply discrete event system models. Based on our experience in coloured Petri net models, the aim is to find efficient methods and algorithms for automatic generation, verification and hierarchical decomposition of operating (control), safety, diagnostic and maintenance scheduling procedures for process systems. The underlying engineering knowledge is expressed in terms of discrete event dynamic process models.