On the design of simple structure control laws for complex processes

The problem of designing a control law for a large scale multivariable process, which can be made of a number of subsystems interacting with each other, involves many difficulties and has received a great of attention during the last decade [1]. It can be expressed as finding a control structure as simple as possible in order to obtain some prespecified static and dynamical performances while minimizing the closed loop interactions between subsystems. Starting from the classical optimal control theory, different suboptimal approaches have been proposed, such as optimization of structure-constrained regulators [2], decomposition by means of singular perturbations [3], use of reduced-order models [4], hierarchical control and decomposition-coordination techniques [5, 8, 9]. The aim of this communication is to propose further developments of a new method [6], which belongs to this latest class, in order to precise its possibility of application while improving its efficiency.