Robust neural network‐based control of static var compensator

Abstract

This study addresses the problem of designing robust stabilisation control for a large class of uncertain single-machine infinite-bus electrical power systems with static var compensator (SVC). This class of systems may be perturbed by plant uncertainties, unmodelled perturbations and external disturbances. An adaptive neural network-based dynamic feedback controller is developed such that all the states and signals of the closed-loop system are bounded and the stabilisation error can be made as small as possible. As the small perturbations in the input weighting gains are neglected, an H ∞ control performance can be guaranteed. The adaptive neural network approximation systems are designed to learn the behaviours of the unknown functions, and in turn a modified procedure is proposed such that the number of the neural network basis functions can be significantly reduced. Consequently, the intelligent robust control scheme developed here possesses the properties of computational simplicity and easy implementation from the viewpoint of practical applications. The developed robust control scheme not only can handle a large class of uncertain SVC-driven power systems, but also achieve the aim of enhancing the stability performance. Finally, simulations are provided to demonstrate the effectiveness and performance of the proposed control algorithm.

Intelligent wide‐area damping controller for static var compensator considering communication delays

“…[31][32][33][34] These controllers have also been applied with the SVC for oscillation damping application and have been found to overperform the conventional controllers. [20][21][22][23][24] In this study, a linear-neuro-adaptive controller (LNAC) is presented to provide enhanced oscillation damping under varied operating scenarios. As shown in Figure 7, the LNAC consists of the system identification and controller.…”

“…Since dynamic inputs and outputs can be easily mapped using the neural network, which has led to the development of various neural network‐based adaptive control schemes 31–34 . These controllers have also been applied with the SVC for oscillation damping application and have been found to overperform the conventional controllers 20–24 …”

“…Various neural network-based control schemes have been reported to enhance oscillation damping of the power system through SVC. [20][21][22] Other neuro-based control scheme used for such application include neurofuzzy-based control. 23,24 However, the above-discussed controllers lack some essential features required for their practical realizability, like online adjustments and simple structure.…”

“…In Lu et al, 19 controller parameters were optimized using ant‐colony optimization at nominal operating conditions, and fuzzy‐based control was utilized during off‐nominal conditions. Various neural network‐based control schemes have been reported to enhance oscillation damping of the power system through SVC 20–22 . Other neuro‐based control scheme used for such application include neuro‐fuzzy‐based control 23,24 .…”