Fuzzy decision-making design of chemical plant using mixed-integer hybrid differential evolution

Abstract

In this study, we have introduced a fuzzy decision-making approach to design a multi-objective optimal design problem of a multiproduct batch chemical plant. In the crisp decision-making approach, the designer is first to solve the multi-objective optimization problem, and then to check whether each optimal objective function value satisfies the rigid preference goal. In real world application, the preference goal for each objective function is an interval bound not a rigid value so that the problem becomes a fuzzy goal optimization problem. A monotonic increasing or decreasing membership function is used to define the degree of satisfaction for each objective function so that the fuzzy goal optimization problem is then represented as an augmented minimax problem formulated as mixed-integer nonlinear programming (MINLP) models. To obtain a unique solution, we have introduced the mixed-integer hybrid differential evolution (MIHDE) to solve the MINLP problems. The MIHDE can straightforwardly solve the problem without any mathematical model transformation. Two examples were used to illustrate the applicability of the proposed interactive algorithm.