Likelihood Function and Threshold Estimation for CUSUM Test Application

M. Šimandl, J. Široký, and I. Punčochář (Czech Republic)

Keywords

fault detection, CUSUM test, likelihood function estimation, threshold estimation

Abstract

The cumulative sum (CUSUM) test for fault detection is treated in the paper. The stress is laid on the estimation of likelihood functions and a threshold for the CUSUM test application. The likelihood functions are estimated as the Gaussian mixtures and the threshold of the CUSUM test is designed on the basis of measured data. The novel fault detection algorithm is illustrated using real data and the results are compared with the CUSUM test working with likelihood functions as normal distributions.

Important Links:



Go Back