Early detection and location of a boiler leak help reduce possible equipment damage and productivity loss. In the present study, a four-element acoustic array and a set of hyperbolic equations were used to locate a power plant boiler leak. Maximum likelihood (ML) and phase transformation (PHAT) estimators were used to localize the leak source. Error rate and root mean square error (RMSE) evaluation revealed the superiority of ML over PHAT in the noisy and lowly reverberant boiler environment. To avoid distant source assumption, a genetic algorithm (GA) modified by an adaptive Gaussian mutation operator was used to search for the global hyperbolic optimum by probability calculations. The GA slightly outperformed the quasi-Newton method and required more time to converge. However, selecting a starting point near the true position is not simple in practice, and iterative process conver- gence is not assured in the quasi-Newton method. Time delay estimator errors greatly influence localization accuracy. The quaternary plane array localization error was within the permitted range of 0.01 ms, whereas that of the stereo array was 0.1 ms. Compared with the quaternary plane, the stereo array was more robust and accurate, but required more time to converge.

Hyperbolic boiler tube leak location based on quaternary acoustic array

SARTI, AUGUSTO;ANTONACCI, FABIO;
2011-01-01

Abstract

Early detection and location of a boiler leak help reduce possible equipment damage and productivity loss. In the present study, a four-element acoustic array and a set of hyperbolic equations were used to locate a power plant boiler leak. Maximum likelihood (ML) and phase transformation (PHAT) estimators were used to localize the leak source. Error rate and root mean square error (RMSE) evaluation revealed the superiority of ML over PHAT in the noisy and lowly reverberant boiler environment. To avoid distant source assumption, a genetic algorithm (GA) modified by an adaptive Gaussian mutation operator was used to search for the global hyperbolic optimum by probability calculations. The GA slightly outperformed the quasi-Newton method and required more time to converge. However, selecting a starting point near the true position is not simple in practice, and iterative process conver- gence is not assured in the quasi-Newton method. Time delay estimator errors greatly influence localization accuracy. The quaternary plane array localization error was within the permitted range of 0.01 ms, whereas that of the stereo array was 0.1 ms. Compared with the quaternary plane, the stereo array was more robust and accurate, but required more time to converge.
2011
File in questo prodotto:
File Dimensione Formato  
An-Wang-Sarti-Antonacci-Shi - Applied thermal engineering.pdf

Accesso riservato

: Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione 2.64 MB
Formato Adobe PDF
2.64 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/637559
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 19
  • ???jsp.display-item.citation.isi??? ND
social impact