Bootstrap Rata-Rata Bergerak Ganda Tertimbang Eksponensial dalam Proses Kontrol Multivariat

Abstract

The purpose of this study is to build alternative methods of product quality control using Bootstrap Double Exponential Weighted Moving Average in Multiva- riate Process Control (MDEWMA). The data used in this study are secondary data from the PDAM water production process. The observed variables consisted of the results of measurement of quality characteristics, namely the turbidity, pH and the remaining chlorine. The results of the study show that in controlling the mean pro- cess using the DEWMA multivariate (MDEWMA) bootstrap control diagram it is known that the weighting of 0.6 and ARL 370 are the most sensitive control di- agrams for twice doubling resampling. Four times doubling resampling shows that the weighting 0.5 and ARL 200 are the most sensitive control diagrams where all observations are statistically controlled. Based on the calculation of the capability produced, it is known that in the previous phase or stage, using the T 2 Hotelling it is known that the monitoring process is not capable (not good) and not in accor- dance with the target value, meaning that it still needs improvement for the level of accuracy in the implementation process. In the next stage using bootstrap DEWMA multivariate (MDEWMA) the process provides a reliable control chart result because of its high sensitivity in detecting out of control errors.