UPFC control employing gradient descent search

Abstract

Abstract-Increasing demand coupled with limitations on new construction indicate that existing power transmission must be better controlled in order to continue reliable operation. Recent advances in FACTS devices provide a mechanism to better control power flow on the transmission network. One particular device, the Unified Power Flow Controller (UPFC), holds the most promise for maintaining operation even when the system has suffered partial failure (either naturally occurring, due to human error, or a malicious attack). In addition to the capital cost, the primary obstacles to widespread UPFC use are the combined problems of selecting the most cost effective locations for installation and maintaining proper control of them once installed.In this paper we list evidence that Gradient Descent search based on load-flow computation is more realistic and accurate than many of the optimization techniques currently in use. We then demonstrate that Gradient Descent search can be used to select control points that improve system fault tolerance more than those found by the Max-Flow technique. In addition, we demonstrate that the size of the system being computed and the number of computations is bounded and is practical for real time control.

Dynamic equivalent model development to improve the operation efficiency of wind farm

“…The gradient method is an optimization method based on the first order derivative of the optimization objective function [20][21]. It could be combined with PSO to accelerate the convergence rate to prevent being trapped locally due to its stochastic nature [22].…”