An analysis of a hybrid-driven underwater glider motion control system based on neuroendocrine controller algorithm

Abstract

This paper presents a neuroendocrine controller algorithm, which controls the motion of a hybrid-driven underwater glider. The controller is designed by combining an artificial neural network (ANN) and endocrine system (AES). The neural network predictive control based on the feedforward architecture is designed as the backbone of the controller. On the other hand, a gland cell of the AES is designed as the weight tuning factor of the ANN. The design objective is to obtain better control performance over the glider motion with the presence of disturbance as well as having adaptive behaviour. We have simulated the algorithm by using Matlab, and the results demonstrated that the neuroendocrine controller produced better control performance than the neural network controller. The cost function or performance index is reduced by 26.8%.