Improving Reliability in Multistage Processes with Autocorrelated Observations

Abstract

Multistage process surveillance is considered to effectively improve the product reliability in manufacturing or service operations. To this end, the process output is commonly inspected under specific conditions and the values of the reliability-related quality characteristic are measured. However, in some cases, the observations from the process output are autocorrelated. This brings about the situation where the use of existing monitoring schemes is futile. Therefore, a class of survival analysis regression models called the proportional hazards (PH) model has been modified to justify the effect of cascade property in line with the autocorrelation issue. Subsequently, three monitoring procedures have been devised in both the presence and absence of a censoring mechanism. The problem of unobserved heterogeneity is also addressed and remedial action has been discussed using frailty models. The performance analysis reveals that the cumulative sum (CUSUM) control chart outweighs the other two competing monitoring schemes. An example is given to illustrate the application and performance of the proposed control charts in real practice. Finally, the impact of ignoring autocorrelation has been studied which confirms the significant effect of autocorrelation on the performance of the process control.