Revista Tecnica de la Facultad de Ingenieria Universidad del Zulia   2016 年 39 卷 3 期

2016

This paper presents a multigrid algorithm for the solution of coefficient inverse problems constrained by variable-coefficient linear reaction-diffusion partial differential equations. The aim of this work is to assess thermal conductivity of fiberboards at two different kinds of the raw materials (miscellaneous wood composite materials and log composite materials). Thermal measurements is done conducted with Neumann boundary conditions. Inversion variable is considered to be a function of space and temporal and the relationship between space and time is found to a second order polynomial function. The inverse problem is formulated as a PDEconstrained optimization. The discrete control equation by using central difference approach in this study, thentake into account Newton iterative equation for inversion parameter and full observations is used toanalyze the performance of the scheme for the non-constant coefficient case. Numerical experiments demonstrate that the effectiveness of the method for different materials. Finally the results shows that calculated results agreed comparatively well with the experiment measured value in two materials of hot-pressing, which proved that the analytical result is useful to calculate the thermal conductivity of fiberboard during hot-pressing.