The effect of measurement errors on the performance of the Exponentially Weighted Moving Average control charts for the Ratio of Two Normally Distributed Variables
Abstract:Investigating the effect of measurement errors on the control chart monitoring the ratio of two normal random variables is an important task to facilitate the use of this kind of control chart in practice. Moreover, a deep insight into the problem can help practitioners to find a way to reduce unexpected impacts of measurement errors on the chart performance. This paper provides a study on the performance of the exponentially weighted moving average control chart monitoring the ratio in the presence of measure… Show more
“…Over the years, and especially in recent times, numerous articles have been published on control charts to optimize process monitoring through various improvements, amendments, and adaptations to current practical needs, see e.g. Teoh et al [40], Mitra et al [31], Filho & Valk [18], Song et al [37,38], Nguyen et al [35], just to name a few relevant works.…”
“…Over the years, and especially in recent times, numerous articles have been published on control charts to optimize process monitoring through various improvements, amendments, and adaptations to current practical needs, see e.g. Teoh et al [40], Mitra et al [31], Filho & Valk [18], Song et al [37,38], Nguyen et al [35], just to name a few relevant works.…”
“…The variable sampling intervals (VSI) EWMA and CUSUM schemes to monitor the ratio of two normal variables were suggested by Nguyen et al 11,12 Abubakar et al 13 propounded the run sum ratio scheme for two normal variables. Nguyen and Tran 14 considered two one-sided Shewhart charts to monitor the ratio of two normal variables when measurement errors are present, while Nguyen et al 15 evaluated the effects of measurement errors on the EWMA scheme to monitor the ratio of two normal variables. In another new development, the one-sided Shewhart charts for monitoring the ratio of two normal variables in short production runs were investigated by Tran et al 16 Numerous investigations concerning the effect of measurement errors on the run-length characteristics for univariate and multivariate schemes were suggested in the literature.…”
As far as literature of quality control is concerned, this is the first article that advocates the run sum ratio scheme with measurement errors, called the RS‐RZ ME chart. The linear covariate error model is employed in designing the RS‐RZ ME chart in detecting increases and decreases in the ratio of two variables from the normal distribution. The average run length and expected average run length values of the RS‐RZ ME chart are obtained using the Markov chain model. A comparison of the RS‐RZ ME scheme with two measurement errors based charts in the literature, namely, the Shewhart ratio and standard run sum ratio charts is conducted. The results indicate the superiority of the RS‐RZ ME chart over the aforesaid existing charts for most of the shift sizes and shift intervals considered. The findings reveal that as the values of the parameters controlling the accuracy error of the measurement system, false(θX,θYfalse)$({{\theta _X},{\theta _Y}} )$ increase, the RS‐RZ ME scheme's efficiency increases. In the same vein, as the values of the parameters controlling the precision of the measurement system, false(ηX,ηYfalse)$( {{\eta _X},{\eta _Y}} )$ decrease, the RS‐RZ ME scheme’ efficiency increases. Furthermore, as the value of the correlation coefficient between variables X and Y increases, the RS‐RZ ME chart's efficiency increases. The application of the RS‐RZ ME scheme is illustrated using data from a food industry.
“…Note that the effect of measurement errors on the performance of HWMA scheme to monitor the process mean is discussed in Thanwane et al 18,19 for Cases K and U, respectively. For some early discussions on measurement errors in SPM context, see Linna and Woodall, 27 Maravelakis et al, 28 and Maravelakis 29 ; however, for some recent contributions, see Riaz et al, 30 Sabanho et al, 31,32 Zaidi et al, 33,34 Nguyen et al, 35 and Noor-ul-Amin et al 36 The objective of this paper is to incorporate FIR features in the time-varying control limits of the HWMA scheme to monitor the mean of processes with and without measurement errors so that its sensitivity in detecting start-up problems can be improved. In the review papers by Jensen et al, 37 Psarakis et al, 38 and Does et al, 4 it is stated that the estimation of the process parameters significantly degrades the performance of a monitoring scheme; thus, the investigation of the effect of parameter estimation on the performance of the HWMĀscheme for both a constant and linearly increasing measurement system error variances is conducted.…”
Section: Introductionmentioning
confidence: 99%
“…Note that the effect of measurement errors on the performance of HWMA scheme to monitor the process mean is discussed in Thanwane et al 18,19 for Cases K and U, respectively. For some early discussions on measurement errors in SPM context, see Linna and Woodall, 27 Maravelakis et al, 28 and Maravelakis 29 ; however, for some recent contributions, see Riaz et al, 30 Sabanho et al, 31,32 Zaidi et al, 33,34 Nguyen et al, 35 and Noor‐ul‐Amin et al 36 …”
Fast initial response (FIR) features are generally used to improve the sensitivity of memory-type control charts by shrinking time-varying control limits in the earlier stage of the monitoring regime. This paper incorporates FIR features to increase the sensitivity of the homogeneously weighted moving average (HWMA) monitoring schemes with and without measurement errors under constant as well as linearly increasing variance scenarios. The robustness and the performance of the HWMA monitoring schemes are investigated in terms of numerous run-length properties assuming that the underlying process parameters are known and unknown. It is found that the FIR features improves the performance of the HWMA monitoring scheme as compared to the standard no FIR feature HWMA scheme, and at the same time, it is observed that the simultaneous use of a recently proposed FIR feature and multiple measurements significantly reduces the negative effect of measurement errors. An illustrative example on the volume of milk in bottles is used to demonstrate a real-life application.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.