Automatic design of multivariable QFT control system via evolutionary computation

Abstract

This paper proposes a multi-objective evolutionary automated design methodology for multivariable QFT control systems. Unlike existing manual or convex optimisation based QFT design approaches, the’intelligent’ evolutionary technique is capable of automatically evolving both the nominal controller and pre-filter simultaneously to meet all performance requirements in QFT, without going through the conservative and sequential design stages for each of the multivariable sub-systems. In addition, it avoids the need of manual QFT bound computation and trial-and-error loop-shaping design procedures, which is particularly useful for unstable or non-minimum phase plants for which stabilising controllers maybe difficult to be synthesised. Effectiveness of the proposed QFT design methodology is validated upon a benchmark multivariable system, which offers a set of low-order Pareto optimal controllers that satisfy all the required closed-loop performances under practical constraints.