Reinforcement learning approach for enhancing the control quality of the jacking system

The Jacking System (JS) for Jack-up Rig (JuR) in Vietnam mostly operates semi-automatically. In this paper, the authors focus on developing a reinforcement learning (RL) algorithm to enhance control quality for JS by adaptively adjusting the basic Proportional – Integral - Derivative (PID) control parameters through reinforcement learning network layers. Two testing scenarios for the proposed solution were performed in offshore conditions with the impact of wave, wind, and current components, on the Matlab 2023b computing environment. The results show that the JS lift the hull follows the desired value. Additionally, the solution was compared to a Recurrent Neural Network (RNN), which demonstrated faster response times, reduced overshoot, and lower steady-state error. This comparison confirms the effectiveness of the proposed solution.