MODEL PREDICTIVE CONTROL DESIGN FOR SHIP FIN STABILIZER SYSTEM BASED ON RECURRENT NEURAL NETWORK

Abstract

When ships are sailing on the sea, roll motion will greatly reduce the safety of ships and cargo, as well as the health of the crew. Due to advantages of fin stabilizer, nowadays, active fin became a popular device, which usually installed on ships to reduce roll motion, and the fins roll reduction efficiency depends primarily on the controller. In our work, model predictive control (MPC) method is proposed for ship linear fin stabilizer system based on recurrent neural network. MPC is an effective method in process control, which can be used to improve efficiency of the control system. However, one of the constraints of MPC is the heavy computational burden when solving the optimization problem. To tackle this problem, the recurrent neural network (RNN) is introduced to solve quadratic programming (QP) problem so that a higher convergence can be achieved. In our work, MPC based on RNN is applied to the ship linear fin stabilizer model to derive the control strategy for this system. Finally, a numerical simulation is given to validate effectiveness of the designed algorithm.