Model predictive control and stabilisation of interconnected systems

Publication Type:
Issue Date:
Full metadata record
The attraction of having higher efficiency and quality, as well as increasing reliability and flexibility for industrial plants and network systems has created opportunities for new research in the control and optimisation fields. Among various design methods, model predictive control (MPC) strategies have proved to be effective in industrial applications. Whilst found widespread used with stand-alone controllers in the refining and many other industries, the field of orchestrating non-centralised MPCs and distributed MPCs is evaluated as still in its infancy. The work in this thesis is concerned with stabilising methods for the control of complex interconnected systems with mixed connection configurations employing distributed and decentralised model predictive control schemes. Inheriting the advantage of the MPC strategy, the control and state constraints are naturally dealt with by the employed methods. As a result, the novel concept of asymptotically positive realness constraint (APRC) and the segregation and integration constructive methods for the constrained stabilisation of interconnected systems are introduced and developed. The MPC is formulated with state space models and stabilising constraints within the open-loop paradigm in this thesis. By having the control inputs entirely decoupled between subsystems and no additional constraints imposed on the interactive variables rather than the coupling constraint itself, the proposed approaches outreach various types of systems and applications. For parallel connections that emulate parallel redundant structures and have unknown splitting ratios, a fully decentralised control strategy is developed as an alternative to the hybrid approaches. For the semi-automatic control systems, which is involved with both closed-loop and humanin- the-loop regulatory controls, the stability-guaranteed method of decentralised stabilising agents which are interoperable with different control algorithms is germinated and implemented for each single subsystem.
Please use this identifier to cite or link to this item: