Algorithm development and analysis for on-line optimising control of large scale industrial processes

The work presented in the thesis is concerned with on-line optimising control of large scale industrial processes. Theoretical analysis has been carried out to investigate optimality and convergence features of various optimising control algorithms in both centralised and hierarchical forms, providing a basis for algorithm design and assessment. Important issues, such as iterative strategies, coordination methods and feedback structures, concerning the improvement of algorithm efficiency are explored. An improved price updating formula is proposed and implemented in the single iterative loop Integrated Optimisation and Parameter Estimation (ISOPE) structure with global feedback to further improve the convergence features of the algorithm. A new coordination technique — Modifier Coordination (MC) method is proposed and implemented in both single and double iterative ISOPE structures. Approaches for coping with output dependent constraints are examined and the Penalty Relaxation (PR) technique is integrated into the ISOPE structure to extend the existing ISOPE algorithms so as to cover many output dependent cases. Comparative studies of some newly developed algorithms, techniques and methods based on substantial computer simulations are also provided. Issues concerning software implementation of optimising control algorithms are discussed providing a general guideline to such practices. Suggestions for future research as a continuation of the work presented in this thesis are also made