Data-driven path-following control of underactuated ships based on antenna mutation beetle swarm predictive reinforcement learning

According to whether the mechanism model of the controlled ship system is introduced into the algorithm, the ship motion control mode can be divided into model-based mode and model-free mode. For the model-based mode, the effectiveness of ship dynamic model is the key to the realization of motion control. It is difficult to obtain the hydrodynamic parameters in the ship mechanism model, especially when the experimental conditions are limited. This paper proposes an intelligent optimization predictive reinforcement learning algorithm for motion control of an underactuated cargo ship. Firstly, the ship dynamic model training data set and reward data set are obtained through PID control. By comparing lasso, ridge and elastic network algorithms, the appropriate regression algorithms are selected to train these two types of data sets. Using line-of-sight (LOS) navigation and intelligent optimization predictive reinforcement learning algorithm, a ship motion controller is designed based on antenna mutation beetle swarm predictive reinforcement learning (AMBS-P-RL) algorithm. Simulations and experimental results of 64 TEU 4m twin-propeller twin-rudder container model ship show that the designed ship motion controller can complete the path-following task well when the hydrodynamic coefficient and ship parameters are uncertain. It has the ability to resist wind and wave disturbance, and the introduction of reinforcement learning also improves the accuracy and adaptability of ship motion control.

Automated summary

This paper introduces an intelligent optimization predictive reinforcement learning algorithm for motion control of an underactuated cargo ship. Training data sets and reward data sets are obtained through PID control, and appropriate regression algorithms are selected for training. The experimental results show that the designed ship motion controller has good path-following capability in uncertain conditions.