An Approach Based on an Interactive Procedure for Multiple Objective Optimisation Problems

Abstract

In this paper the Interactive Procedure for Multiple Objective Optimisation Problems (IPMOOP) is presented as a tool used for the solution of multiple objective optimisation problems. The first step of this procedure is the definition of a decision-making process group (DMPG) unit, where the decision maker (DM) and the analytic programmer actively interact throughout the solution of the problem. The DMPG is responsible for providing the information about the problem’s nature and features. The main characteristic of IPMOOP is the definition of a surrogate objective function to represent the different objectives. The objectives are transformed into goals and the DM is asked to define aspiration levels which include his preferences. In each iteration, the solutions are presented to the DM who decides whether to modify the aspiration levels or not. This allows the DM to find a satisfactory solution in a progressive manner. In order to demonstrate the applicability of this approach an activity allocation problem was solved using a genetic algorithm since the surrogate function can act as the fitness function. The solutions found were satisfactory from the DM’s point of view since they achieved all the goals, aspiration levels and met all the constraints.