Exponential Smoothing of Multiple Reference Frame Components With GPUs for Real-Time Detection of Time-Varying Harmonics and Interharmonics of EAF Currents


Balouji E., Salor O., ERMİŞ M.

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, vol.54, no.6, pp.6566-6575, 2018 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 54 Issue: 6
  • Publication Date: 2018
  • Doi Number: 10.1109/tia.2018.2849059
  • Journal Name: IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.6566-6575
  • Keywords: Active power filter (APF), exponential smoothing (ES), electric arc furnace (EAF), graphical processing unit (GPU), harmonics, interharmonics, multiple synchronous reference frame (MSRF), parallel processing, power quality (PQ), ELECTRIC-ARC, FLICKER EVALUATION, POWER QUALITY, FURNACES, DECOMPOSITION, SYSTEM, FILTER, STATIONARY, ALGORITHM, OPERATION
  • Middle East Technical University Affiliated: Yes

Abstract

In this research work, a multiple synchronous reference frame (MSRF)-based analysis method, used together with exponential smoothing (ES) to accurately obtain the time-varying harmonics and interharmonics of electric arc furnace (EAF) currents, is proposed. The proposed method has been implemented on NVIDIA Geforce GTX 960 graphics card for the parallel processing of all harmonics and interharmonics so that real-time processing of the EAF currents obtained from a measurement point of the electricity transmission system is possible. To obtain the best possible accuracy for the EAF current harmonics and interharmonics estimation, the ES order has been optimized specific to each harmonic and interharmonic frequency component. It has been shown on field data that the implemented system is capable of successfully estimating all harmonics up to the 50th and all interharmonics at 5-Hz resolution in real time. Moreover, active filtering of certain harmonics and interharmonics has been achieved in the simulation environment and it has been shown that successful real-time filtering of any harmonic and interharmonic component is possible with the proposed MSRF+ES method. Hence, it can be concluded that the proposed system can be used to generate reference signals for the controllers of the new technology compensation systems such as active power filters.