Open AccessEnhancement of electric arc furnace reactive power compensation using Grey–Markov prediction method
Author(s) -
Samet Haidar,
Mojallal Aslan
Publication year - 2014
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2013.0698
Subject(s) - electric arc furnace , compensation (psychology) , electric arc , ac power , power (physics) , markov chain , arc (geometry) , computer science , control theory (sociology) , automotive engineering , mathematics , metallurgy , engineering , statistics , artificial intelligence , voltage , materials science , electrical engineering , chemistry , electrode , mechanical engineering , thermodynamics , psychology , physics , control (management) , psychoanalysis
The time varying nature of electric arc furnace (EAF) gives rise to voltage fluctuations, which produces the effect known as flicker. Employing reactive power compensation devices such as static VAr compensator (SVC) is one of the main approaches to mitigate this phenomenon. By utilising prediction methods to forecast EAFs reactive power consumption for a half‐cycle ahead, performance of SVC can be enhanced substantially. This study proposes a rolling Grey model and a Grey–Markov method to predict the actual reactive power of Mobarakeh Steel Company, Isfahan/Iran. To investigate the efficiency of the proposed methods the results are compared with the results of EAFs reactive power compensation when no prediction method is employed. Furthermore, autoregressive moving average (ARMA) models with updating coefficients, which are studied in the literature are used to predict EAF reactive power. Various methods for updating ARMA coefficients including normalised least mean square, recursive least square method and an online genetic algorithm are used. By comparing the indices which are defined using the concept of flicker frequency and power spectral density, the superiority of Grey–Markov and rolling Grey model over the aforementioned prediction methods is investigated.