Evaluation of multi-objective inverse heat conduction problem based on particle swarm optimization algorithm, normal distribution and finite element method

Abstract

The boundary heat transfer coefficient (BHTC) and thermal conductivity are the most essential thermal physical parameters, which have a significant effect on the calculation accuracy of physical fields in the numerical simulation. To simultaneously estimate the temperature-dependent BHTC and thermal conductivity by solving the inverse heat conduction problem (IHCP), a hybrid method (ZPSO) based on particle swarm optimization algorithm (PSO), normal distribution method and finite element method (FEM) is presented, where normal distribution method representing the transient solution space is used to improve the convergence speed. A two-dimensional direct heat conduction problem (DHCP) with the temperature-dependent thermal-physical parameters is simulated based on FEM. Some temperature curves at certain position of FEM model are attained in the simulation. The temperature-dependent BHTC and thermal conductivity are simultaneously evaluated by PSO–FEM and ZPSO–FEM according to the temperature curves attained in DHCP. Comparing the temperature-dependent thermal-physical parameters attained in the IHCP to that used in the DHCP, the results show that the BHTC and thermal conductivity evaluated by the ZPSO are well consistent with those used in the DHCP. The evaluation results show that the convergence of the hybrid method is well, and the convergence speed is accelerated by the ZPSO.