Consolidated Technique of Response Surface Methodology and Data Envelopment Analysis for setting the parameters of meta-heuristic algorithms - Case study: Production Scheduling Problem

    In this study, given the sequence dependent setup times, we attempt using the technique of Response Surface Methodology (RSM) to set the parameters of the genetic algorithm (GA), which is used to optimize the scheduling problem of n job on 1 machine (n/1). It aims at finding the most suitable parameters for increasing the efficiency of the proposed algorithm. At first, a central composite design was created and then using the data relating to the plan, the complete second-degree model was fitted. Then, by solving the developed non-linear programming model the optimal values of the parameters determined. The performance of algorithm, considering the obtained parameters as inputs of the common Data Envelopment Analysis (DEA), was measured. This way, we can decide on the most effective kinds of problems that can be solved by GA in a similar volume. This study can be used as a model of setting parameters of evolutionary and meta-heuristic algorithms using scientific techniques to prevent disadvantages relating to trial and error methods.