LONG-TERM PREDICTION OF NAVIGATIONAL RISK FOR DESIGN OF COASTAL APPROACH CHANNELS AND HARBOR WATERS

摘要

This paper presents a new approach for long - term prediction of navigational risk regarding the possibility of ships exceeding the channel limits or navigable areas, assumed as grounding or collision with a fixed structure. The method, which is based on real time simulation results, consists of two developed models: (1) an ARMAX (Auto Regressive and Moving Average eXogenous) model is adopted to identify the ship steering dynamic system. With the help of this model, the outputs of the system (course, position, etc.) can be estimated on the basis of the system input conditions (rudder, engine, etc.); (2) the stochastic sequences of the inputs for the first model used are generated using a semi-Markov model. One implementation of the semi-Markov model for rudder actions has been described. The study used input/output measurements from a ship-handling simulator to estimate the model parameters, so the human factor has been included in the models. The method allows us to extend the results obtained from the simulator to predict future conditions of the system outputs. On the basis of the predicted results, the long- term assessment of the navigational risk can be realized, which amounts a straightforward use for optimal design of maritime engineering (e.g. coastal channels, fairways and harbor water areas).