Title: Generalized design of control chart for weighted-count data under measurement error
Authors: Wichai Chattinnawat - Chiang Mai University (Thailand) [presenting]
Abstract: Consider an online automated high speed quality monitoring system where the process quality is determined by the number and types of nonconformity. In general, Demerit control chart can be adopted to monitor the system where the score calculated from the quantity, type, and severity of the nonconformities is used as control chart statistic. The process is considered acceptable as long as the calculated demerit score remains within present limits. However, there is no systematic investigation of optimum strategy presented in the literature on how to design the demerit weights of general types by taking into account the distribution of the demerit score which is not normally distributed as imposed by the traditional demerit control chart assumption. Moreover, the online high speed quality measurement such as based pattern recognition technology can be associated with some degree of measurement errors. Putting altogether the traditional demerit control chart needs to be studied and generalized. A generalized design methodology is presented for weighted-count type control chart with arbitrary weights structure to monitor several types of nonconformity when the nonconformity types are independent but possibly masked with measurement error. The aim is to propose a design methodology to derive the optimum weights and a new design methodology of self adapting weights.