Extreme learning machine–radial basis function (ELM-RBF) networks for diagnosing faults in a steam turbine

Research output: Contribution to journalArticlepeer-review

Abstract

A fast and reliable fault diagnosis system for a steam turbine in thermal power plant is crucial. The system will detect and classify a potential or occurring fault, hence suitable precautions steps will be correctly determined, and unplanned breakdown will be prevented. This study proposes a new application of extreme learning machine-radial basis function networks (ELM-RBF) for steam turbine fault diagnosis system. ELM-RBF recently has been known for its extremely fast computation. The proposed system was tested with real fault historical data from a steam power plant in Jakarta. To evaluate the system performance, a comparison with backpropagation neural networks (BPNN) was conducted. Four scenarios using ELM-RBF and BPNN, with and without ReliefF for feature selection were designed. The results show high accuracy in almost all the scenarios tested. The BPNN shows better accuracy than ELM-RBF, however, ELM-RBF performs considerably faster computation than BPNN without significant decrease in accuracy.

Original languageEnglish
JournalJournal of Industrial and Production Engineering
DOIs
Publication statusAccepted/In press - 2021

Keywords

  • Data-driven method
  • extreme learning machine
  • fault diagnosis
  • neural networks
  • steam turbine

Fingerprint Dive into the research topics of 'Extreme learning machine–radial basis function (ELM-RBF) networks for diagnosing faults in a steam turbine'. Together they form a unique fingerprint.

Cite this