Selecting an Optimal Alternative by Constructing a Regression Model for Each Option

Abstract

Optimization for identifying the best solution is a pivotal technique that has garnered significant attention from the scientific community. It plays a decisive role in enhancing system efficiency, particularly across multidisciplinary engineering sectors such as mechanical, medical, environmental, food, and chemical engineering... This study introduces a novel approach to selecting the optimal alternative from a set of available options. For each alternative, a unique regression model is constructed to represent the relationship between its value and the evaluation criteria. The Design of Experiments (DOE) methodology is employed to build an experimental matrix, where each evaluation criterion for the alternatives is treated as an input parameter for the experimental process. The score for each alternative is calculated using the PSI method, a multi-objective optimization technique that does not require weighting the criteria. The use of DOE allows the proposed method to quickly rank alternatives when the number of options changes (increases or decreases). A notable advantage is that the proposed method can swiftly identify the optimal alternative even when a new option is added whose criteria values fall outside the range of existing alternatives’ criteria values. This is a significant superiority over all current methods. The proposed method is applied to three numerical examples and subsequently to a typical example in the field of chemical extraction. All tests were rigorously evaluated by comparing the proposed method with other existing methods, thereby confirming the validity of the proposed approach.