M A Hannan
Intelligent Control for Ship Manoeuvering
Automation in ship operation is getting everybody’s attention as the specialised knowledge of workers continues to decline. Apart from that, the increasing modern technologies often demand a promising solution of highly demanding control problems under increased uncertainty.
In such situation, conventional approaches have been proposed for several control problems. However, successful applications can only be found within well-constrained environment and none is flexible enough to perform beyond its restricted zone. Consequently, numerous advancements have been made in developing intelligent systems. One of these is inspired by human’s central nervous system called artificial neural network (ANN). This ANN consists of several interconnected simple non-linear system typically modelled by the transfer function therefore, has the capability to replicate human brains and perform the action that a human brain does in any particular situation.
In ship manoeuvering, the only two actuators are rudder and propeller revolution for the whole system. Therefore, by controlling these two for a designated purpose, automation in ship manoeuvering could be established. Classical control system like PID for track keeping or course changing has widely been used by on-board autopilot system.
However, there are number of sophisticated ship manoeuveres, where the classical control systems fail and thus, the intelligent controllers are often encouraged to take over the classical systems. This work point out the application of Artificial Neural Network (ANN) controller for automatic ship berthing and Fuzzy Logic controller for waypoint controller in particular. Nonlinear programming language (NPL) method for minimum time course changing manoeuvre is utilised to create the consistent teaching data for course changing manoeuvre and the repeated optimisation technique. Strategies as well as the formation of the two controllers are discussed briefly. In addition, simulations and experiments results are presented and compared to ensure the practical applicability of these intelligent controllers.