Comparative analysis of process control charts in order to define most efficient sampling frequency

Abstract

This thesis conducts various comparative analysis of control charts in order to define most efficient sampling frequency. Control charts analyzed through this work are tools used in process of quality monitoring invented by Walter A. Shewart and have been used ever since in every branch, field or process that can or needs to be monitored. The paper contains theoretical as well as analytical part of creation of I, R and CuSum charts to define most efficient sampling frequency. Thesis focuses and explains process of making different control chart types as well as indicates problems and proposes solutions to problems faced during the making those charts, to create efficient monitoring system of water levels. According to conducted analysis for multiple charts to monitor water level for i charts real standard deviation must be used instead of approximated one and most trustful charts are created when smaller time period is monitored, one or two months. The dataset analyzed originally contains 52608 samples of water levels measured every 30 minutes through the period of 3 years. Original dataset is further categorized to make 3 different datasets, one for every year, which is even further modified to simulate sampling frequencies of 60-, and 90-minutes sampling.