Ship maneuvering in restricted waterways often posses challenges to even well experienced ship operator under wind disturbances. Therefore, automation would be a great relief in such circumstances. As an intelligent controller, ANN has been found best for such MIMO (multi input – multi output) system. But to train it properly, challenges still remain to ensure the consistency in teaching data. This research therefore, focuses on the optimization process and the relevant constraints to make up the consistent teaching data that will be used for training purposes later on.
It is found that, compared to no wind where rudder angle time history followed a certain pattern for all the cases, the rudder angle time history varies significantly among the cases under wind disturbances to take into account of various wind directions. Although the steering becomes complicated under the influence of wind, reasonable optimization results are achieved (with some exception) using this proposed method for various wind speed and directions. It seems that it is difficult to converge if the steering angle restriction is relaxed and the steering control is easy. In such cases, it is necessary to reduce the limits to reduce the maximum steering angle at the first step of this repetitive optimization process.
The proposed optimization method can be used as a tool/guide while performing difficult maneuvering in restricted waterways in real life. However, this method is not fully automated and a more robust optimization method can be developed by incorporating neural networking with the offline outputs of the present method to train the neural network for online control.