Adaptive Kalman filter for marine navigation and gravity measurement

A simple adaptive Kalman filter has been developed to give optimal estimation of a ship´s course, speed and position giving sufficient accuracy to be useful for the processing of marine gravity measurements. The inputs to the filter are from the ship´s electromagnetic log, gyrocompass and radio navaid. The radio-navaid noise is assumed to be zero mean and white, but it is shown that the filter still works satisfactorily when the noise is coloured, with a spectrum similar to that obtained from long-range marine radio navaids beyond the edges of their present useful operating areas. The log and gyrocompass noises are assumed to be white and nonzero mean, and the effects of these nonzero means (due to wind and current in practice) examined. An algorithm is developed to offset the effects of the nonzero means. The sensitivity of the filter to incorrect data statistics is examined and a limited-memory radio-navaid variance estimator incorporated. Finally, the developed filter is used with real ship data obtained from the ocean survey ship, HMS Hecla.