Journal of Quality Engineering and Management

Abstract Multi-response optimization performed by the response procedure method is very common. Before optimization, we need to select and fit the appropriate model for each response. A major problem that may occur due to incorrect fitting of models and failure to reach optimal solutions is model misidentification. The solid regression model method, which is a semi-parametric method for estimating d, can have a better performance than both parametric and non-parametric estimation methods against model misclassification. In this research, the use of a robust regression model method is proposed to improve the model estimation and the appropriate fit of each of the answers will be investigated by one of the multivariate optimization methods, namely the utility function. In the following, an applied study is presented to compare parametric, nonparametric and semi-parametric methods. The results of this study show that the performance of the stable regression model is more appropriate in many situations as well as in the modeling stage than the other two methods. Therefore, the optimization results with a stable regression model are much more reliable.

Abstract The subject of time-dependent quality indices—whose values vary over time—has attracted significant attention from quality researchers in recent years. Various methods have been developed for this purpose, among which one of the most recent approaches involves analyzing performance profiles over time. Moreover, as competition among manufactured products intensifies, product design has become a primary focus within the product life cycle for manufacturing organizations. One of the most widely used approaches in this context is robust design. The application of response surface methodology and its related optimization tools has also gained great importance in design studies. In this article, a utility function–based method is presented for the robust design optimization of time-dependent quality indices. Based on the results obtained from a pharmaceutical case study, the proposed method performs significantly better than the two existing methods.

Abstract Decision variables aimed at improving system performance are considered one of the key issues in most industries. To this end, numerous multi-objective optimization methods have been developed in recent years to address such problems. However, most of these approaches overlook the potential correlations among objectives. In this study, an optimization approach based on a utility function framework is proposed to solve the perindopril separation problem (perindopril is a widely used and effective pharmaceutical agent for treating cardiovascular diseases and hypertension). The proposed approach not only places all objectives at a minimally acceptable level of utility from the decision maker’s (DM) perspective but also accounts for the potential correlations among the objectives and their relative importance. A comparison of the results obtained from a numerical example using the proposed method with those of existing approaches in the literature demonstrates its favorable performance.