Abstract: [Objectives] This paper proposes a vertical motion control algorithm based on deep reinforcement learning, focusing on the dependency of traditional control algorithms on precise mathematical models and system parameters. [Methods] The method achieves its goal by designing reward functions, neural network structures and adjusting relevant hyperparameters. It combines these techniques with a catamaran model. Finally, through experiments, it compares the control performance of the deep reinforcement learning DDPG algorithm and the GA-LQR algorithm under three different control modes and the robustness under different operating conditions and initial states. [Results] Under the same operating conditions, when comparing different control modes, the DDPG algorithm has a slight advantage in control performance over the GA-LQR algorithm, but its fin angle output during the control process is more aggressive. In simulated experiments under different operating conditions and initial states, when the system and environmental models undergo significant changes, the control performance of the DDPG algorithm is significantly affected. However, when the system and environment changes are small, the DDPG algorithm exhibits better adaptability and superiority over the GA-LQR algorithm. [Conclusions] Overall, this study concludes that the DDPG algorithm performs similarly to the GA-LQR algorithm in terms of performance.