Application of PID controller based on adaptive artificial neural network for ship control system

Traditional control methods for designing advanced control systems such as Proportional Integral Derivative (PID) controllers are for typical ships still popular because of their simple structure and with a sustainable calculation. This paper tries to develop application of PID controller based on adaptive neural network for ship navigation control system, thereby improving the quality of PID controller of this control system. At the same time, experimental design of the adaptive neural PID controller according to simulation and experiment are performed. Design of a ship model identifier using the input-output signal method is introduced and applied. The recognizer uses a multi-layer feedforward neural network, but the author trains the network online, enhancing it with good adaptation speed, capable of identifying nonlinear ship models that change over time, not just a static linear model like previous studies. By combining this neural recognition model, the control method is conducted in a real-time predictive control style, improving adaptation and control quality. PID controller with the proportional, integral, differential parameters Kp, Ki and Kd adjusted using a back-propagation neural network that is explicitly calculated and simulated. The online synthesis and modeling ability of the neural network helps the parameters of the PID control map to be fine-tuned and selected directly over time, and the adaptability of the neural network in control is utilized and promoted.