SURROGATE MODEL-BASED OPTIMIZATION FOR THE DEVELOPMENT OF PRODUCTION SYSTEMS
Published in GÉP 2026/2
https://doi.org/10.70750/GEP.2026.2.9
Ruzicska Gábor
tanársegéd, Debreceni Egyetem Műszaki Kar, Gépészmérnöki Tanszék
Dr. Czégé Levente
egyetemi docens, Debreceni Egyetem Műszaki Kar, Gépészmérnöki Tanszék
Dr. Mankovits Tamás
egyetemi docens, Debreceni Egyetem Műszaki Kar, Gépészmérnöki Tanszék
ABSTRACT
Optimizing production systems using simulation models makes it possible to analyze system performance under various operating conditions; however, running simulation experiments multiple times can require significant computational resources. The goal of this research is to develop and evaluate an approach for optimizing a manufacturing system based on a surrogate model, and then to validate the resulting solution using the original simulation model. The subject of this study is a stochastic production system in which the speeds of the two conveyor belts and the processing time spent at the production station serve as control parameters. The system’s efficiency is evaluated based on three indicators: productivity, average congestion time, and equipment downtime. Based on the data obtained from the simulation experiments, we construct surrogate models that approximately describe the relationship between the control parameters and the system’s performance metrics. To find the optimal combination of parameters, a genetic algorithm is used that relies on the predictions of the surrogate model at each step of the optimization process, rather than running the simulation model directly. The resulting optimal configuration is subjected to further validation using the original simulation model. Given the stochastic nature of the system under study, validation is performed using multiple independent iterations, which allows for the evaluation of the mean values of the performance metrics and their standard deviations. The results of the research indicate that the use of surrogate models is a promising solution for reducing computation time in the optimization of manufacturing systems, while repeated simulation validation makes it possible to assess the reliability of the resulting optimal solution.
