Multi-objective simulation optimization through search heuristics and relational database analysis

The SimMOp framework for generating solutions to simulation optimization problems containing multiple objectives is presented. The complexity within this subject is the conflicting multiple stochastic outputs whose estimates are only available through simulation. The framework combines a simulation model, a non-exhaustive heuristic search algorithm with an embedded multi-objective optimization technique, and database technologies to generate a set of good quality solutions. The goodness of solutions is measured from a multi-objective and stochastic perspective through analysis after the search phase of the methodology. The methodology has been tested using a discrete-event simulation model based on inventory theory.