The performance comparison of classical and fast backpropagation neural network algorithms for rudder roll stabilization of fishing vessels

Abstract

This paper presents the performance comparative results of the classic backpropagation algorithms (CBAs) and fast backpropagation algorithms (FBAs) of neural networks applied for the rudder roll stabilizer (RRS) systems. The rudder is used to keep stability of the ship motions by minimizing errors. In this paper, neural networks are designed to reduce the errors considerably by using rudders for simultaneous course keeping and roll damping. Simulation results show the effectiveness of FBA approach for rudder roil stabilization of fishing vessels when compared with CBA results.