Mathematical and scale model platforms for ship guidance trials

Research at the Institute of Marine Studies into mathematical ship manoeuvring models has been ongoing for the last fifteen years. Nonlinear techniques have been developed for the purposes of filtering navigational data and control of ships. The latest research into ship control employs artificial intelligence in the form of neural networks. These are designed to handle multivariate nonlinear systems and the investigation proceeded to deal with the track-keeping problem. The mathematical model has provided a test bed on which control strategies can be evaluated, the main forces acting on the vessel which disturb the system from its stable state are environmental. This paper provides a mathematical model and describes the inclusion of environmental effects using proprietary software. The model is however unsuitable for testing control architectures which are designed to handle nonlinearities greater than those provided by the computer simulation. The construction of a physical scale model provided a better platform. Handling characteristics were similar to those of the full size vessel but a faster response time was provided. Furthermore, by undertaking trials on an open lake the vessel was subject to wind and wave disturbances. In order to undertake trials on a scale model, the navigational instrumentation, errors present on such systems and computer processing tools needed to be suitably scaled. This was achieved with careful selection of components and processing algorithms. The paper describes these and concludes by providing results of controller trials which indicate the application of filtering to navigation data in real time to provide feedback