engleski

Dead reckoning in ship dynamic positioning systems based on EKF and dynamic neural networks

Dynamic positioning systems are used for maintaining vessel position, heading and speed, but also a predefined motion path, position mooring etc. To ensure efficient performing of these operations, it is necessary to determine an accurate estimation of low-frequency position, heading and speed. In commercial DP systems, the vessel observer is usually extended Kalman filter (EKF) which is traditionally used in marine control systems. On the other hand, one of the most critical situations that can affect the control system is sudden loss of position and heading measurements caused by sensor failures. In this case, observer must operate in a dead reckoning mode providing predicted values of lost signals. In order to increase dead reckoning capabilities of dynamically positioned marine vessels, intelligent identifier and estimator based on the nonlinear autoregressive neural network with exogenous inputs (NARX) is proposed for the purpose of intelligent identification and estimation. NARX is a recurrent dynamic neural network with feedback connections enclosing several layers of the network. The NARX model is based on the linear ARX model, which is commonly used in time-series modelling. Proposed hybrid system for sensor fusion and reconstruction of lost GPS signals that combines NARX and EKF was trained, adjusted, tested and verified with real measurements from the DP Log archive of dynamically positioned heavy-lift and J-lay pipe vessel. Obtained results clearly indicate strong synergic effect between NARX and EKF, thus presenting a significant contribution in the design of so-called virtual sensors for fault-tolerant control in low-speed manoeuvring.

dynamic positioning systems; navigation; extended Kalman filter; dynamic neural network; dead reckoning

nije evidentirano