Support vector regression degradation modeling for constant fraction discriminator

• In the nuclear industries, the electronic signal processing unit plays a key role in the data processing, data analysis, control mechanism and more importantly safety of the nuclear reactor. The processing unit comprises of different modules that process pulse and current signals from detector and constant fraction discriminator which has higher criticality is one of them. Earlier the reliability was calculated using MilHdbk 217 standard and found discrepancies to the field failure. This paper studies the failure phenomenon using physics of failure approach by studying degradation and failure analysis and conducting the experiments using modified physics of failure methodology. Support vector machine (SVM) is a machine learning phenomenon using statistical learning theory. In this paper, failure data is fed to SVM for regression models intended for life prediction. From the parametric analysis, it was found that Sequential minimal optimization with RBF kernel represent the best model for degradation of the CFD. This method provides higher accuracy compared to response surface methodology.

Subject headings

TEKNIK OCH TEKNOLOGIER  -- Samhällsbyggnadsteknik -- Annan samhällsbyggnadsteknik (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Civil Engineering -- Other Civil Engineering (hsv//eng)

Keyword

Drift och underhållsteknik

Operation and Maintenance

Publication and Content Type

ref (subject category)

art (subject category)