Faulty sensor signal estimation in ship's propulsion using information fusion
Radovan Antonić, Zoran Vukić, Ante Munitić · International Carpathian Control Conference · 2003
In the paper, a structural analysis approach to identify most relevant variables its interrelations, components or subsystems of ship's diesel engine propulsion process with inherent redundant information has been proposed. Existing redundant relations could be used in cases of sensor faults to increase fail-safe operational performance. Sensor information fusion method was chosen to be based on using artificial neural networks. The paper presents effective artificial neural network structure for on-line estimation of faulty sensor signal by means of information fusion. Simulation have been carried out in Matlab/Simulink environment with marine diesel engine speed sensor fault example. Good practical estimation results were obtained using related engine working data.