Trajectory Controller Network and Its Design Automation through Evolutionary Computing

Abstract

Classical controllers are highly popular in industrial applications. However, most controllers are tuned manually in a trial and error process though computer simulation. This is particularly difficult when the system to be controlled is nonlinear. To address this problem and help design of industrial controllers for a wider range of operating trajectory, this paper proposes a trajectory controller network (TCN) technique based on linear approximation model (LAM) technique. In a TCN, each controller can be of a simple form, which may be obtained straightforwardly via classical design or evolutionary means. To co-ordinate the overall controller performance, the scheduling of the TCN is evolved through the entire operating envelope. Since plant step response data are often readily available in engineering practice, the design of such TCN is fully automated using an evolutionary algorithm without the need of model identification. This is illustrated and validated through a nonlinear control example.