Fishing Vessel Trawl Winch Tension Control: A BP Neural Network PID Feedforward Control Method Based on NARX Neural Network Prediction

In order to solve the problems of the poor adaptability to nonlinear systems, cumbersome parameter adjustment, and sensing-execution delay facing PID control for trawl winch tension control on fishing vessels, a prediction model for trawl winch cable tension was developed using a NARX neural network. The network was trained using historical data to achieve the accurate prediction of the trawl winch cable tension value in the future moment. The predicted value of the NARX neural network was introduced into the BP-PID controller as a feedforward quantity, and a BP-PID feedforward control strategy based on the prediction of the NARX neural network was designed. The simulation results in MATLAB software version: 9.13.0 (R2022b) show that, in comparison with the conventional PID control method, the BP-PID feedforward control strategy based on NARX neural network prediction substantially minimizes the fluctuation in trawl winch tension, enhances the control accuracy and robustness, and demonstrates excellent control performance under various sea states and load conditions.