Optimal Estimation of Process Capability Indices with Non-Normal Data and Inaccurate Parameters using Metaheuristics

Abstract

The current methods for estimating Process Capability Indices are based on the assumptions of normality and accuracy of process data. Under actual production conditions the data of quality characteristics of the products may be non-normal and/or have imprecise parameters. Therefore, in this paper we propose a new methodology for estimating Process Capability Indices when the data are non-normal and the specification limits are not crisp numbers. The methodology was validated using experimental data. The proposed methodology uses the Clements's method assuming a Burr type XII distribution, whose parameters are estimated through metaheuristic techniques, and considers the obtaining of fuzzy numbers using the statistical inference theory.