A New Self-Organizing Double Function-Link Brain Emotional Learning Controller for MIMO Nonlinear Systems Using Sliding Surface

Chih-Min Lin,Tuan-Tu Huynh,Hiep-Binh Nguyen

doi:10.1109/access.2021.3079446

Abstract

This paper aims to propose a new type of neural network which is the self-organizing double function-link brain emotional learning controller (SO-DFL-BELC) for multiple input multiple output (MIMO) nonlinear systems. The proposed controller is a newly designed neural network containing the key mechanism of a typical brain emotional learning controller (BELC), which is a mathematical model that approximates the judgmental and emotional activity of a brain, in which it is combined with some additional functions and methods. Firstly, a double function-link (DFL) network is applied to expand the function for a BELC to improve the accuracy of the system weights. Secondly, the self-organizing mechanism is utilized to increase or decrease the number of neurons that possibly supports the main controller to adapt to the sharp change of the input and to reduce the computation time significantly. Thirdly, the learning rules of the SO-DFL-BELC are developed based on the gradient descent algorithm and sliding surface. Finally, all parameters of the system can be optimized. The proposed SO-DFL-BELC is applied to control two different MIMO nonlinear systems that are a 4D chaotic system and a four-tank system. The simulation results show the favorable control performance of the proposed control algorithm.

Highlights

For dealing with system uncertainty, many disturbance observers-based control approaches have been widely researched such as extended-state-observer-based output feedback backstepping control of hydraulic actuators with valve dynamics compensation [1], extended-state-observerbased adaptive control of electrohydraulic servomechanisms without velocity measurement [2], and an output feedback approach for time-varying input delay compensation for nonlinear systems with additive disturbance [3]
This paper proposes the design to apply function-link networks to a Brain emotional learning controller (BELC)
The amygdala channel space: The AC is built based on the amygdala, which is the emotional part of the brain

Summary

Introduction

For dealing with system uncertainty, many disturbance observers-based control approaches have been widely researched such as extended-state-observer-based output feedback backstepping control of hydraulic actuators with valve dynamics compensation [1], extended-state-observerbased adaptive control of electrohydraulic servomechanisms without velocity measurement [2], and an output feedback approach for time-varying input delay compensation for nonlinear systems with additive disturbance [3]. The associate editor coordinating the review of this manuscript and approving it for publication was Juntao Fei. based on the research of Moren, Balkenius, and LeDoux that shows some parts of the brain produce emotion signals in the amygdala cortex and the orbitofrontal cortex [4]–[8]. Based on the research of Moren, Balkenius, and LeDoux that shows some parts of the brain produce emotion signals in the amygdala cortex and the orbitofrontal cortex [4]–[8] From all of those studies, the structure of a BELC was built and utilized to control various systems such as doubly fed induction generator systems [9], omni-directional three-wheel robots [10], interior permanent-magnet synchronous motor drives [11], encoderless synchronous reluctance motor drives [12], etc. The structure of BELC is modified using some efficient methods outlined below to improve its efficiency

Objectives

Methods

Conclusion