Piyapatr Busababodhin. Analytical derivation of the average run length for cumulative sum chart in the cases of exponential distribution. Doctoral Degree(Applied Statistics). King Mongkut's University of Technology North Bangkok. Central Library. : King Mongkut's University of Technology North Bangkok, 2012.
Analytical derivation of the average run length for cumulative sum chart in the cases of exponential distribution
Abstract:
The objective of this thesis is related to the use of Cumulative Sum chart(CUSUM) for detection of the change in three types of the process observations; (i) the exponential distribution, (ii) the autoregressive (AR(p)) with exponential distribution white noise and (iii) the trend autoregressive (trend AR(p)) with exponential distribution white noise. The explicit formulas are derived and the numericalintegrations algorithms are developed for evaluating the commonly popular characteristics of CUSUM chart. Throughout this thesis, the interesting characteristics are the Average Run Length (ARL), and the Average Delay Time (ADT). They are two conflicting criteria that had to be balanced in the control charts. In this thesis, the explicit formulas for the ARL and ADT of CUSUM chart have been derived by using the Integral Equations (IE) approach which based on the Fredholm integral equation of the second type technique. These explicit formulas are first guaranteed the existence and uniqueness of the solutions by using the Banachs fixed point theorem and then guaranteed the accuracy by numerical solutions of the Numerical Integral (NI) approach which based on Gauss-Legendre quadrature rule. The results show that the proposed formulas are easy to calculate and program. The numerical results and the values obtain from the explicit formulas are in excellent agreement. In addition, the explicit formulas obviously take the computational time much less than the numerical approximations.