engleski

Neural Network based ship steering

The automatic ship steering is very important in shipping and shipbuilding industry, either on the clasical sea-going liner ships or on the super-high speed liners - the ships of new generation. The main goals of automatic ship steering, by computer based autopilots, are reduction in cruising times and propulsion power. But, disadvantages of this solution are nonlinear nature of the problem cause of changing the ship's characteristics and/or the sea conditions. In this paper the neural network (NN) models and controllers are proposed and presented, which give the new more precise ship model and open way for successful control once and the model of the real ship is available after learning process. If the model is an accurate representation of the real ship and the controller has been designed correctly, then the controller will work on the real ship. Both of these tasks can be completed using neural network (NN) trained with backpropagation, a common network training technique. The network will be trained to predict the next state of the system given its present state and inputs. After learning time the ship will be steered like &laquo ; ; sitting by the Nellie&raquo ; ; method. The application of this neural network (NN) based ship steering, particurarly on the liners ships would be very usefull and profitable.

Neural network; automatic ship steering; reduction in cruising time and propulsion power; sea conditions

nije evidentirano