Neural network-based adaptive control for a supercavitating vehicle in transition phase

Abstract

A supercavitation is a modern technology that can reduce the frictional resistance of an underwater vehicle. To reach supercavity, which is when the cavity envelops the entire vehicle body, a vehicle passes through the transition phase from fully wetted to supercavitating operation where unsteady hydrodynamic forces and moments are created by partial cavity. This paper presents analytical and numerical investigations into the dynamics of a supercavitating vehicle in the transition phase as well as a neural network-based adaptive control method that guarantees a stable control performance with modeling uncertainty. The ventilated cavity model is used for the rapid supercavity condition, which is when the cavitation number is relatively high. The cavity profile decides the immersion depths of the fins and body to calculate the hydrodynamical effects on the body. The supercavitating system in the transition phase is replaced by linearized pitch motion dynamics to verify the performance of the design controller. In addition, a numerical simulation shows how the proposed control method can compensate for modeling uncertainty.