Consistently Trained Artificial Neural Network for Automatic Ship Berthing Control

Abstract

In this paper, consistently trained Artificial Neural Network controller for automatic ship berthing is discussed. Minimum time course changing manoeuvre is utilised to ensure such consistency and a new concept named 'virtual window' is introduced. Such consistent teaching data are then used to train two separate multi-layered feed forward neural networks for command rudder and propeller revolution output. After proper training, several known and unknown conditions are tested to judge the effectiveness of the proposed controller using Monte Carlo simulations. After getting acceptable percentages of success, the trained networks are implemented for the free running experiment system to judge the network's real time response for Esso Osaka 3-m model ship. The network's behaviour during such experiments is also investigated for possible effect of initial conditions as well as wind disturbances. Moreover, since the final goal point of the proposed controller is set at some distance from the actual pier to ensure safety, therefore a study on automatic tug assistance is also discussed for the final alignment of the ship with actual pier. http://www.transnav.eu the International Journal on Marine Navigation and Safety of Sea Transportation Volume 9 Number 3

Artificial Neural Network controller for automatic ship berthing: challenges and opportunities

“…25 shows other experiment results for LHS approach. A proper justification of such behaviours is given in Ahmed and Hasegawa [15] with a scope of further improvement in this research.…”

“…To maintain the consistency in course changing manoeuvre, a concept based on virtual window is proposed in Ahmed and Hasegawa [15]. Here, virtual window denotes a safety window which ensures a ship with any particular heading passes through it will reach the imaginary line well ahead for step deceleration.…”

“…In case of transfer function, log-sigmoid is found suitable which is given as Performance of the trained network is judged depending on calculated mean squared error value (MSE). If the [15] normalized teaching data are considered in the following form:…”

“…Therefore, researchers are very keen to do such navigational tests in open spaces like a pond or river that allows the model ship to face the real environmental disturbances. Ahmed and Hasegawa [15] also planned to validate his controller for Esso Osaka 3-m model ship that Osaka University has and used the system as shown in Fig. 24, to facilitate the berthing experiments.…”

“…By following this strategy, he successfully managed to do ship berthing simulations under wind disturbances, and later on, conducted ship berthing experiments with the trained controller to justify its effectiveness in real situation operation. So far, a very limited number of experiments have been carried out for automatic ship berthing, and this research is one of those which demonstrates the effectiveness of the trained controller to berth a ship under wind up to 15 m/s from any direction [15]. In addition, to make the trained controller more versatile and universal to all port operations, a proper strategy to convert the original GPS data to a compatible set data for the trained controller was explained and verified for berthing experiments.…”