Design of self-tuning controllers for highly non-linear mechatronic systems

Abstract

A new procedure for design of self-tuning adaptive controllers (e.g., PID, lead-lag and in general n th order linear controllers) for use with highly non-linear mechatronic systems of the sort encountered in robotic and servomechanism control is developed. The procedure is presented in an algorithmic format. Describing function models, using analytical or experimental techniques, are used to determine the structure of the approximating linear plant at various operating regimes. The parameters of this linear structure are identified (using any available identification scheme) followed by obtaining the parameters of the linear controller by solving a set of simultaneous linear algebraic equations in real time. The procedure is demonstrated for design of self-tuning PID and lead-lag controllers for a non-linear plant of the sort encountered in servomechanism. It is shown that results are better than a fixed gain linear PID controller, and they compete with designs involving a non-linear PID and a non-linear H ∞ controller that were previously reported in the open literature.