The evaluation of steady turning performance for dual-tail propulsion underwater gliders typically relies on high-fidelity unsteady CFD simulations, which remain computationally prohibitive for control optimization and multi-scenario analysis. To overcome this limitation, this paper proposes a rapid prediction framework integrating Kriging surrogate modeling with dynamic equilibrium constraints. The proposed method employs a physics-informed decoupling strategy that isolates the hydrodynamic behavior of the hull from the thrust generation of the propellers. Since these are governed by distinct physical mechanisms and operate at different spatial scales, the decoupling strategy enables efficient and targeted steady-state CFD analysis for each component subsystem. Latin Hypercube Sampling (LHS) is used to generate training data for highly accurate Kriging models, which are subsequently coupled with the glider’s balance equations to form a bidirectional solution system. The forward mode predicts turning performance from control inputs, whereas the inverse mode identifies propeller speeds required for desired trajectories. Validation via fully-coupled 6-DOF unsteady CFD simulations confirms that the framework achieves prediction errors below 10% for key turning parameters while improving computational efficiency by over an order of magnitude. The method provides an effective tool for rapid maneuverability evaluation, control system design, and real-time path planning in dual-tail propulsion underwater gliders.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号