Comparison of SISO and SIMO neural control strategies for ship track keeping

The paper presents an extension of an earlier paper by the authors (1995), concerned with the development of a single-input single-output (SISO) neural controller for the ship track-keeping problem. Here the single-input parameter remains as the rudder, but now a multiple output strategy based on the tracking error and heading error (SIMO) is developed and then applied to the same problems investigated by the authors. The differences in the backpropagation-based SISO and SIMO controllers are explained and their relative performances with respect to one another is discussed. In common with earlier reported research, the influence of random wind on the manoeuvring ship and the effect of random noise on the response information supplied to the controller are investigated. In summary, the authors initially review the development of track-keeping strategies since the 1980s