Power positioning control system for rake suction dredger

Abstract Rake suction dredgers are specialized vessels used for dredging projects in water bodies. Power positioning refers to the precise control of the vessel’s position in the water using power systems to complete operational tasks or maintain specific positions. This paper proposes a PID control algorithm based on reinforcement learning to address the accuracy and stability issues of power positioning in rake suction dredgers. The algorithm utilizes a deep Q-network as the core of reinforcement learning, combined with PID control principles, to enable the intelligent agent to control the vessel in real time. In algorithm design, the problem of overestimation is effectively addressed through a dual neural network structure, and an experience replay mechanism is introduced to enhance training efficiency and stability. In the reinforcement learning process, a reward function suitable for the characteristics of rake suction dredgers is designed, considering the balance between path-tracking accuracy and vessel stability. Experimental results demonstrate that the proposed algorithm can achieve power positioning control of rake suction dredgers under different sea conditions, with high path tracking accuracy and stability, providing an effective solution for the autonomous navigation and operation of rake suction dredgers.