Neural Network Steering Controller For A Ship

Abstract

This paper presents the implementation of Multi-layer Preceptron (MLP) Feed-forward Neural Networks for direction control of an oil tanker, whose parameters vary with depth of water and randomly varying waves. The development of Artificial Neural Networks (ANN) controller is based on learning of networks, for this purpose back-propagation algorithm is used. For training, a nonlinear sliding mode controller is used as a supervisor. A network having one input, one hidden and one output layer has provided the satisfactory performance. The input layer has four neurons, the hidden layer has seven neurons and output layer has one neurons. The controller is developed for ideal conditions, to assure the robustness of controller the performance of controller is tested in different operating conditions; varying depth of water and in presence of wind generated waves. It has been demonstrated that the performance of controller remained satisfactory in all operating conditions.