A Free-Running CFD-Data-Driven Component-Wise Hydrodynamic Operator Model for 3DoF AUV Maneuvering

This study presents a reduced-order dynamic model for three degree-of-freedom AUV maneuvering. The proposed model identifies linear operators that map physically selected, polynomially expanded kinematic subspaces to hydrodynamic forces and moments using free-running zigzag-test data obtained from computational fluid dynamics simulations. To improve prediction stability and physical interpretability, the CFD-resolved force and moment contributions from individual components, including the hull, rudders, and propeller, are extracted and modeled separately. This component-wise formulation allows each hydrodynamic contribution to be reconstructed from a corresponding physically informed kinematic subspace. The identified operators are first evaluated through forecasting validation under the training maneuver and are then applied to an untrained turning-circle maneuver. The results show that component-level hydrodynamic forces and moments can be approximated by linear operator mappings constructed from free-running CFD data. The identified relationships retain predictive capability in the untrained maneuvering scenario, indicating that the proposed framework can serve as a practical reduced-order model for CFD-based maneuvering prediction.