Real-Time Residual Buoyancy Identification for Deep-Sea Autonomous Vehicle

Abstract

The autonomous underwater vehicle glides in vertical plane by controlling its buoyancy. The buoyancy of the deep-sea underwater vehicles will change with depth due to pressure hull deformation and seawater density variation. The buoyancy variation will affect the dynamic behaviors and control of the underwater vehicle. In this study, a real-time residual buoyancy identification method is proposed and applied for deep-sea autonomous vehicle to precisely identify the residual buoyancy to ensure the stable control of deep-sea autonomous vehicles and reduce the energy consumption of system regulation. The residual buoyancy identification model is designed according to the recursive least square regression model. In this on-line identification system, the residual buoyancy is regarded as the target of identification, and the vertical motion state variables of vehicles are taken as input variables. The results of residual buoyancy are obtained by iteratively calculating the vertical motion identification formulas. The simulated results demonstrate the effectiveness and inherent robustness of the proposed multi-innovation recursive least square algorithm for real-time buoyancy identification. The new on-line residual buoyancy identification method developed in this investigation is proved to be effective.