Robust Diving Motion Control of an Autonomous Underwater Vehicle Using Adaptive Neuro-Fuzzy Sliding Mode Technique

Abstract

This paper presents an adaptive neuro-fuzzy sliding mode control (ANFSMC) scheme for diving motion control of an autonomous underwater vehicle (AUV) in the presence of parameter perturbations and wave disturbances. In the derivation of diving motion equations of an AUV, the pitch angle of the vehicle is often assumed to be small in the vertical plane. This is a quite strong restricting condition in underwater operations and may cause serious modeling inaccuracies in AUV's dynamics. The problem of nonlinear uncertain diving behavior with restricting assumption on the pitch angle directly is resolved by a neural network (NN) based equivalent control. The online NN estimator is designed to approximate a part of the equivalent control term containing nonlinear unknown dynamics and external disturbances. Subsequently, corrective control based on an adaptive fuzzy proportional-integral control is applied to eliminate the chattering phenomenon by smoothing the switching signal and also compensate structured uncertainties. The weights of NN are updated such that the corrective control signal of the ANFSMC converges towards zero. The adaptive laws are developed to compute coefficients of PID sliding manifold and adjust the gain of fuzzy switching control. The simulation results are presented to shows the efficacy of the control performance. INDEX TERMS Autonomous underwater vehicle, adaptive neuro-fuzzy sliding mode control, diving motion, neural network, parameter perturbations and chattering phenomenon.

Recurrent Neural Network-Based Robust Nonsingular Sliding Mode Control With Input Saturation for a Non-Holonomic Spherical Robot

“…where 1 = [ 1 , 2 , … , ] is the vector of weights of the RNN, ̂1 denotes bias, is the estimation error, and ( ) is an activation function [56][57][58].…”

“…Moreover, none of these works have considered control input saturation in the system. However, as it is evident, because of the current limitations in real actuators, the bounds of control input should be considered in real-world systems [56][57][58]. Similarly, although the singularity problem can be induced to a large control input [59][60][61][62], this has not been taken to account in most previous studies on spherical robots.…”

Control of a MIMO Coupled Plant Using a Neuro-Fuzzy Adaptive System Based on Boolean Relations

“…Another related work is [34] that presents an Adaptive Neuro-Fuzzy Sliding Mode Control (ANFSMC) architecture for diving motion control of an Autonomous Underwater Vehicle (AUV) in the presence of parameter perturbations and wave disturbances. The problem of non-linear uncertain diving behavior is addressed using a neural network designed to approximate a part that presents non-linear unknown dynamics and external disturbances.…”

Diving safety alarm based on the techniques of machine learning

“…[16] suggests a borne trans‐media aerial underwater vehicle by using a non‐linear–coupled input system, as well as implementing elevation angle detections. Finally, [17] suggests adaptive neuro‐fuzzy sliding mode control that can enrich the movement control of underwater robots, and using elevation angles to detect sea creatures.…”