Machine Learning Algorithm Based Static VAR Compensator to Enhance Voltage Stability of Multi-machine Power System

Abstract

A novel Extreme learning Machine (ELM) algorithm based tuning of the parameters of the SVC FACTS controller was implemented in order to control voltage at various buses over a wide range. The ELM algorithm is a non-iterative method which forecasts the parameters of SVC FACTS controller quickly and effectively while Back Propagation Neural Network (BPNN) algorithm is an iterative method which takes a long time for training as well prediction of parameters. The load perturbation is one of the nonlinear disturbances which is considered to investigate the operational capability of the control methodology. Standard IEEE 5 and 30 bus systems are considered as test systems and operation of two models of SVC observed with the BPNN and ELM controllers. The weakest bus is identified using L-Index method which is the optimal location of the SVC FACTS. Results show that the novel ELM method expeditiously and efficiently tunes the parameters of the SVC FACTS controller online such that the voltage regulated to desired value when there is a perturbation in load.

Improving Voltage Stability of Electrical Power System by Optimal Location of FACTS Devices Using an Evolutionary Method

“…The SVC could be used to regulate voltage in EN buses with the reactive power injected or absorbed from the network. Many research papers, such as Tran Thi Ngoat, Le Ngoc Giang and other [8,14,15], have been published in the area of FACTS devices incorporation for enhancing the operation of electrical power systems.…”

Machine Learning MOSA Monitoring System

“…A different approach of using machine learning algorithm in the enhancing voltage stability of multi-machine power system have been shown in [16] which was used for alteration of the parameters of the static var compensator controller. The authors in [17] used the back propagation neural network algorithm for time sequence prediction.…”

“…The short come of the BPFM is that update of the weights is in the upward direction instead of downward. Therefore, the momentum should be varied rather than keeping it fixed as shown in [16,17]. The energy coefficient updates consider all the weights in the Multi-layer Perceptron (MLP) to assure its main advantage.…”

Impact of Distribution Network Reconfiguration and Optimal Capacitor Placement under Wheeling Transactions

“…There are many literatures [8][9][10][11][12][13][14] extensive research is carried out on the application of FR for and Optimal Capacitor Placement (OCP) in distribution system for minimizing power loss, maximize voltage stability and improve voltage profile. To solve multi-objective optimization problems the algorithms like MOPSO algorithm for power dispatch, machine learning algorithm based static VAR compensator to enhance voltage stability, partitioning differential transform method for solving nonlinear integrodifferential equations and integration of DG system with PV module to grid with modelling are reported [15][16][17][18]. Effective methodologies are proposed to incorporate FR and OCP in distribution network under deregulated environment.…”