Magnus减摇装置对豪华游艇等中小型船舶减摇具有重要意义。复杂海况下，Magnus装置旋转圆柱自身旋转、摆动与船体运动是非线性耦合动力系统，同时旋转圆柱在来流中存在流致振动。这种非线性、强耦合特点导致Magnus装置在水动力预报、多自由度减摇控制方面存在挑战。本项目针对Magnus装置复杂运动水动力特性及其控制方法开展研究。通过粘性流体力学方法对纯转动旋转柱体、耦合轴体摆动旋转柱体、耦合船舶摇荡运动多旋转柱体等开展数值模拟和水池试验研究，探索Magnus效应装置的水动力学机理。据此，设计抑制涡激振动旋转圆柱构型、构建Magnus装置多自由度减摇控制力学模型。结合深度强化学习方法，开发船舶多自由度运动减摇深度强化控制算法，并通过水池试验进行验证和优化。通过上述工作，形成Magnus装置水动力特性及其预报方法、旋转圆柱涡激振动抑制构型、多自由度减摇强化控制算法、数值程序等理论成果。

Magnus stabilizer device has the advantages of small size and full speed stabilizer, which is of great significance to the motion control of small and medium-sized high value-added ships. However, ship motions stabilization using Magnus stabilizer is a complex nonlinear dynamic process. Where the rotation, oscillation of the lift cylinder and the motions of the hull under high sea conditions are highly coupled. In addition, the rotation of the cylinder will induce the flow induced vibration. The nonlinear, strong coupling and multi parameter characteristics lead to the challenges in hydrodynamic prediction and multi-DOFs joint control of the Magnus stabilizer under multi operational modes. In this regard, the project intends to carry out research on the hydrodynamic characteristics and control methods of Magnus stabilizer device with complex operational modes. First of all, the hydrodynamics of the rotating cylinder with pure rotation, the rotating cylinder with coupled shaft swing, and the multi rotating cylinder with coupled ship motions are carried out by using viscous fluid dynamics. According to the above research, the hydrodynamic mechanism of Magnus effect device is studied, and the hydrodynamic models of Magnus stabilizer and pitch-roll joint stabilization are constructed. Combining the above dynamic models with the deep reinforcement learning method, the deep reinforcement control models for multi-DOFs joint stabilization are to be developed. Numerical simulation and experimental test will be carried out to explore the hydrodynamic characteristics and deep reinforcement learning control algorithms for Magnus stabilizer. Through the implementation of this project, the theoretical achievements such as the hydrodynamic characteristics and prediction method of Magnus device, vortex induced vibration suppression configuration of rotating cylinder, multi DOFs stabilization deep reinforcement learning control algorithm and numerical codes are formed.