On improvement of a conditional mornitoring technique for condition-based maintenance

• The condition-based maintenance (CMB) is a hot research topic to overcome the drawbacks belonging to the periodic maintenance used in nuclear power plants nowadays. Auto-Associative Kernel Regression (AAKR) is a widely applied condition monitoring technique which is the basis of a CBM. In this paper, the traditional AAKR is improved by using the ensemble learning technique. The modified AAKR is tested by steady-state operational data of a Tennessee-Eastman chemical process and the results show that it can significantly improve the auto- and cross-sensitivity without reducing the accuracy. This indicates a significant improvement in performance of this condition monitoring technique.

Subject headings

NATURVETENSKAP  -- Fysik (hsv//swe)

NATURAL SCIENCES  -- Physical Sciences (hsv//eng)

Keyword

Auto-associative kernel regression

Condition-based maintenance

Ensemble learning

Publication and Content Type

ref (subject category)

kon (subject category)