Fuzzy estimation for unknown boundary shape of fluid-solid conjugate heat transfer problem

• The DFIM is proposed for estimating the boundary shape of fluid-solid CHT system.
• The BEM coupled with FVM is used to solve the fluid-solid CHT direct problem.
• Build a weighted approach based on normal distribution and with geometry factor.
• The DFIM has a better anti-ill-posed character compared with CGM and GA.

This study proposed a decentralized fuzzy inference method (DFIM) for estimating the unknown boundary shape of a fluid-solid conjugate heat transfer (CHT) system. The method coupled boundary element method (BEM) with finite volume method (FVM) is applied to solve the direct problem. A set of decentralized fuzzy inference units is established. The deviations between the calculated and measured temperatures are taken as the input parameters of fuzzy inference units, and obtain the corresponding inference components by fuzzy inference. Then a synthesizing weighted approach based on normal distribution and with geometry factor is built to weight and synthesize the inference components, and gain the compensations of the boundary shape to revise the guess values of unknown boundary configurations. Some numerical experiments are performed to discuss the validity of the present approach by using different initial guesses of unknown boundary shape, the number of measuring points, measurement errors, etc. Comparisons with the conjugate gradient method (CGM) and genetic algorithm (GA) are also conducted.