ADRC For Ship Steering Based On RBF Neural Network

Active Disturbance Rejection Control (ADRC) is a new technique which developed by analyzing and improving the PID. The development of artificial intelligence technique provides new ideas to improve the algorithm of ADRC.Firstly, this thesis has systematically introduced the components of ADRC and its algorithm. Then establish MMG ship maneuvering mathematical model and verify the effectivity with zig-zag tests. According to the characteristics of ship motion, build the ADRC for ship steering control. Achieve the optimal results by manually adjusting and verify the effectivity and robust of the ADRC for ship steering control.It is difficult for ADRC to achieve an ideal result because of the quantity of its parameters and influences between the parameters. According to the principles of ADRC algorithm, combine the ADRC with RBF neural network to design an ADRC for ship steering based on RBF neural network. Through identifying the input and output of ship mathematic model, the output of RBF could approximate the output signal of plant. Then utilize RBF neural network identifier to adjust the two parameters of NLSEF in real time to achieve optimum ADRC control. At last, by comparing two ADRC controllers’performances to demonstrate that the RBF-ADRC can achieve more precise for ship steering control.