Stochastic Modeling and Adaptive LQG Control of a Ship

Abstract

The purpose of the paper is to discuss problems associated with variations of process dynamics and disturbances. The paper is based on a well known ship model which is expanded with external disturbances, i.e. second order approximations to wind and waves. It serves as a complement to computer studies using more elaborate models. A selective forgetting method is used for identification, showing tracking ability and limitations on the covariance matrix. Using an explicit LQG design, an adaptive autopilot is formed, showing superior performance relative to a conventional PID controller.