Performance Comparison of Fuzzy Logic and Neural Networks Controllers for Ship

Abstract

This paper explores the potential of fuzzy logic and neural networks controllers for heading motions of ship and analyses the performance of both controllers. For this purpose two separate; neural network and fuzzy logic controllers are developed. For neural network Multi-Layer Perceptron (MLP) network is used. The training of network is carried out by using back-propagation algorithm. For fuzzy logic controller Mamdani type Fuzzy Inference System (FIS) is used. The fuzzification of variables is based on triangular functions and the defuzzification is carried out by using centroid method. The processing of Inference system is carried out by developing 49 rules. For comparative analysis of both controllers during developing of controllers the parameters of controllers are kept same.