Generalized multiple dependent state optimal control charts under ranked set sampling for monitoring process mean

Abstract

This paper proposes two optimal control charts by using the generalized multiple dependent state (GMDS) plan to monitor the process mean. These charts make advantageous use of the ranked set sampling schemes and integrate them with the GMDS plan to improve the shift detection ability of the Shewhart‐type control charts. The proposed control charts have two pairs of control limits, inner and outer, which enables them to consider the current and previous sample information while making a decision about the process state. The run‐length profiles of the proposed charts are obtained through extensive Monte Carlo simulations. The individual and overall mean shift detection abilities of the proposed charts are compared with various competing control charts by using out‐of‐control average run length ( ) and the average extra quadratic loss ( ) as performance metrics. The comparison establishes the complete dominance of the proposed charting strategy. In the end, the practical implementation of the proposed charts is demonstrated by using a real data set related to the cement manufacturing industry.

Introducing randomness into univariate control chart sampling strategies for enhancing security against cyber‐physical attacks in production systems

“…For instance, approaches such as ranked set sampling or median ranked set sampling allow operators to judge, based upon a variable of interest (a variable other than the considered KQC), which products will be chosen for inspection. [31][32][33] Unfortunately, performance comparisons with these approaches and this paper's proposed approaches would be difficult/impossible. Specifically, assessing the performances of these alternative sampling approaches requires a mechanism for identifying which product would be inspected (e.g., ranking the products).…”

“…It should also be noted that alternative strategies to the random sampling plans proposed in this paper exist that may also increase the robustness against attacks when 100% inspection is not being performed. For instance, approaches such as ranked set sampling or median ranked set sampling allow operators to judge, based upon a variable of interest (a variable other than the considered KQC), which products will be chosen for inspection 31–33 . Unfortunately, performance comparisons with these approaches and this paper's proposed approaches would be difficult/impossible.…”