Automatic weight selection for H ∞ controller synthesis

Abstract

This article is concerned with the development of a procedure to calculate the parameters of weighting functions used in H∞ controller designs to achieve a desired system performance. The H∞ synthesis problem is formulated as a multi-criterion optimization problem. A Genetic Algorithm is then employed to search for suitable solutions. To cope with the imprecision and vagueness that arises in the description of objective functions and constraints of the process and actuators, concepts from the fuzzy logic are incorporated into the solution. A multi-objective fuzzy optimization is stated and a fuzzy convex decision-making is established. To demonstrate the advantages of the proposed method, the tuning of SISO and MIMO controllers are investigated. The method is applied to the tuning of controllers of a ship autopilot and of a dynamic ship positioning system. Results are presented, analyzed and compared to former works.