Optimal Fault Detection and Resolution During Maneuvering for AUVS

Abstract

As AUV missions become longer and more complex, increased reliability is sought using automatic fault detection and accommodation. In spite of all attempts at redundancy, failures still occur, and if detected quickly, premature mission aborts could be avoided and the mission reconfigured. This paper deals with the issue of fault detection using Kalman filter methodology for detecting a fin fault with an AUV during steering maneuvers. The problem lies in the fact that Kalman Filter fault detection residuals are not only sensitive to faults, but also to errors in the model used. Model errors for the vehicle in a maneuvering condition cause residual increases that are not easily distinguishable from fin faults. This work uses non linear optimization of both model and filter parameters to arrive at a satisfactory solution in which fin faults are detected over a range of maneuvers. The result is evaluated using in water experimental data from test runs using the 21 UUV vehicle.