Decentralized fuzzy inference method for estimating thermal boundary condition of a heated cylinder normal to a laminar air stream

A decentralized fuzzy inference (DFI) method based on the fuzzy theory is proposed in this study for estimating the heat flux distribution of a heated cylinder experiencing conjugate heat transfer. A group of fuzzy inference units (FIUs) are designed. The deviations between the calculated and measured temperature at each measurement point are taken as the input parameters of FIUs, and the corresponding fuzzy inference components corresponding to the measured temperature are obtained by the FIUs. According to the importance of the various measured temperatures, the fuzzy inference components are then weighted and synthesized to gain the compensations of the guessed heat flux distribution. Ultimately, the prediction of the heat flux is accomplished. Numerical tests are performed to study the effect of initial guessed heat flux, measurement point numbers, measurement errors and the coupling of measurement point numbers and measurement errors on the estimated results. The results show that the DFI method can estimate the heat flux availably and possesses a better anti-ill-posed character and higher accuracy than the conjugate gradient method (CGM). The DFI method shows superiority.