Abstract The paper describes an adaptive autopilot for ships which has been put into commercial production. It uses a self tuning adaptive scheme combined with lqq-controllers. The Kalman filter for state estimation includes high frequency states to model wave induced oscillatory yaw motion, For reasons of robustness the basic ship model is of second order only. Course keeping controllers for either precise or energy economic steering are designed to minimize different quadratic criteria. Course changing is done via a phase of turnrate control. For estimation of ship parameters a stochastic gradient algorithm is applied in order to eliminate the problem of temporarily poor excitation. The frequency of the osci1latory yaw motion is estimated with a recursive least square scheme fitting an ARMA–model to the high pass filtered yaw signal. Results from full scale sea trials with the described autopilot show the improved oerformance of the autopilot.
[1]
Karl Johan Åström,et al.
Theory and applications of adaptive control - A survey
,
1983,
Autom..
[2]
Robert E. Reid,et al.
The use of wave filter design in Kalman filter state estimation of the automatic steering problem of a tanker in a seaway
,
1984
.
[3]
Job van Amerongen,et al.
Adaptive steering of ships - A model reference approach
,
1982,
Autom..
[4]
N. A. Jenssen,et al.
Adaptive Autopilot with Wave Filter
,
1982
.
[5]
Lennart Ljung.
Analysis of stochastic gradient algorithms for linear regression problems
,
1984,
IEEE Trans. Inf. Theory.
[6]
Karl Johan Åström,et al.
Identification of ship steering dynamics
,
1976,
Autom..