Design Of Emotional Learning Based Control Schemes For Mimo Systems Under Uncertainties And Disturbances

Abstract

Biologically inspired control strategies have come to be widely used in recent times owing to newlinetheir flexibility and adaptability in the face of diverse and complicated problems. The popularity newlinehas been facilitated partly by improved computational power and a greater understanding newlineof the mammalian brain s physiology. The emotional learning based controller (ELBC) is newlinea biologically inspired strategy that mimics the emotional system of the mammalian brain. newlineThe technique involves matching two stimuli to generate the control signal. Compared newlineto other biologically inspired approaches such as neural networks and fuzzy systems, the newlineELBC has a simple structure and requires less computational effort. It is characterized newlineby a non-parametric, signal-matching, and output feedback-based design approach which newlinefacilitates its rapid and easy deployment across most embedded platforms. However, many newlineissues are associated with the existing strategies that act as an impediment towards achieving newlinesatisfactory performance under varied operating conditions. Firstly, the design methodology newlineused in most cases overemphasizes on feedback component of the stimulus. Such a strategy is newlinesuitable only for regulation around a particular operating point and concedes large transients newlineand steady-state tracking error when faced with significant set-point changes induced during newlinethe start-up and shutdown phases of a plant. Secondly, there is a lack of a systematic approach newlinetoward the stimulus design for strongly coupled multi-input multi-output (MIMO) systems. newlineThe trade-offs pertaining to tracking, regulation, and interloop coupling make the simultaneous newlinecontrol of multiple loops a challenging affair. Additionally, the existing strategies do not newlineaccount for the adverse effects of mechanical resonances on flexible structures. Thirdly, the newlinecurrent adaptive-ELBC approaches suffer from a jittery transient response, significant latency, newlineand parameter drift, making them unsuitable for real-time systems.