Robust low cost meta-modeling optimization algorithm based on meta-heuristic and knowledge databases approach: Application to polymer extrusion die design

The new method presented in this paper falls into the category of sampling methods and model management in the optimization process of surrogate related methods. This method was introduced in order to reach the global optimum with a limited number of computer experiments. During these developments, the Particle Swarm Optimization (PSO) was used as a smart sampling tool to construct the metamodel. These methods with their stochastic nature can also overcome the problems of local minima. In order to improve the efficiency and accuracy of the metamodel (Kriging), a knowledge database with smart sampling methods has been integrated into the optimization model management, to avoid unnecessary finite elements calculations and enrich the collection (sampling) in each optimization iteration. This method makes it possible to reduce the sampling size and at the same time increases the accuracy of the metamodel. For validation of the developed method, different benchmark functions were chosen in terms of features and has successfully then minimized. Finally, a practical engineering optimization problem for polymer extrusion was implemented with suggested Kriging Swarm Optimization algorithm (KSO). In this procedure, the Finite Element Analysis (FEA) was combined for simulation procedures to resolve non-isothermal non-Newtonian flow. Polymer extrusion results were applied for gathering information from design space samples and Kriging.