Torpedo Hydrodynamic Parameter Estimation: Application to Shallow Water Mobile Platform (SWAMP) Sea Run Data

摘要

The Shallow Water Mobile Platform (SWAMP) is a specially instrumented torpedolike vehicle whose purpose is to gather acoustic research data. A model of SWAMP, based on torpedo motion equations with six degrees of freedom, is presented here. Multiplexed dynamic data employed include clock time, body angular rate components, control surface deflections, propeller rotation speed, vehicle course, and depth. First-order maximum likelihood identification algorithms are used as a guide in parameter estimation. These algorithms yield an estimation of the state of the system by means of an extended 'Kalman' type filter, as well as a weighted quadratic function of the residues (i.e., differences between measurements and their computed values) as the maximum likelihood cost function. The filter computes an estimate of the full SWAMP trajectory. A cost function is provided in the form of a weighted sum of squares of the residues of the rate gyros and accumulators. Of gradient techniques investigated for the minimization of the cost function, the Fletcher-Powell technique was found to be superior (as modified because of the extreme sensitivity of the roll-damping coefficient), and only results for this technique are presented. (Author)